|
Travel Model Development Project Phase 2 Final Report - Volume 1: Summary Report
Click HERE for graphic. TRAVEL MODEL DEVELOPMENT PROJECT PHASE 2 FINAL REPORT VOLUME 1: SUMMARY REPORT PREPARED BY CAMBRIDGE SYSTEMATICS, INC. 238 MAIN STREET CAMBRIDGE, MASSACHUSETTS 02142 PREPARED FOR METROPOLITAN TRANSPORTATION COMMISSION HOTEL CLAREMONT, BERKELEY, CALIFORNIA 94705 JUNE, 1980 THE PREPARATION OF THIS DOCUMENT WAS FINANCED IN PART THROUGH GRANTS FROM THE U.S. DEPARTMENT OF TRANSPORTATION, URBAN MASS TRANSPORTATION ADMINISTRATION UNDER THE URBAN MASS TRANSPORTATION ACT OF 1964. AS AMENDED; AND THROUGH GRANTS FROM THE FEDERAL HIGHWAY ADMINISTRATION AND THE CALIFORNIA STATE DEPARTMENT OF TRANSPORTATION. Click HERE for graphic. PREFACE 1 PREFACE THIS REPORT IS IN THREE VOLUMES THAT DOCUMENT THE TRAVEL DEPEND MODELS DEVELOPED IN THE TRAVEL MODEL DEVELOPMENT PROJECT (TMCP). THAT PROJECT WAS CONDUCTED FOR THE METROPOLITAN TRANSPORTATION COMMISSION (MTC) BY THE CONSULTANT TEAM OF COMSIS CORPORATION, CAMBRIDGE SYSTEMATICS, INCORPORATED, AND BARTON-ASCHMAN ASSOCIATES, INCORPORATED. ALTHOUGH THE REPORT DOCUMENTS THE WORK OF THE TMCP, RESPONSIBILITY FOR ITS CONTENT RESTS WITH CAMBRIDGE SYSTEMATICS, INCORPORATED. ADDITIONALLY, THE REPORT REPRESENTS THE STATUS OF THE MODEL SYSTEM DELIVERED TO MTC IN JUNE, 1977. REVISIONS TO AND REFINEMENTS OF THE MODELS BY MTC THEREAFTER ARE NOT INCLUDED. THIS VOLUME, SUMMARY REPORT, IS AN OVERVIEW OF THE ENTIRE SYSTEM OF DESEGREGATE TRAVEL DEMAND MODELS. IT INCLUDED A PRESENTATION OF THE CONCEPTUAL BASIS FOR ITS STRUCTURE, THE INTERRELATIONSHIPS OF THE COMPONENT MODELS, THE THREE GENERAL MATHEMATICAL FORMS OF THESE MODELS, AND A SUMMARY OF THE TWO COMPUTER PROGRAM PACKAGES DEVELOPED TO IMPLEMENT THE MODELS. FINALLY, THE VOLUME PROVIDES A BRIEF DESCRIPTION OF THE FUNCTION, FORM AND COEFFICIENTS OF EACH MODEL. VOLUME 2: DETAIL MODEL DESCRIPTIONS. PRESENTS BOTH A DETAILED DESCRIPTION OF THE TRAVEL MODEL DEVELOPMENT PROCESS, AND THE FULL DETAILS OF THIS PROCESS AS IT APPLIES SEPARATELY TO EACH OF THE 21 COMPONENT MODELS. VOLUME 3: MTCFCAST USERS' GUIDE PROVIDES INFORMATION NEEDED TO EXERCISE THE COMPLETE SET OF MODELS AS THEY HAVE BEEN IMPLEMENTED IN MTCFCAST, A COMPUTER PACKAGE PROVIDING A ZONAL-LEVEL VERSION OF THE MODELS, WITH AGGREGATION ACCOMPLISHED BY MARKET SEGMENTATION TECHNIQUES. THE VOLUME PRESENTS AN OVERVIEW OF MTCFCAST, USER DOCUMENTATION FOR EACH OF ITS PROGRAMS, AND INFORMATION ON THE DATA SETS THAT ARE CREATED BY THE PACKAGE AND USED TO TIE TOGETHER THE COMPONENT PROGRAMS. CAMBRIDGE SYSTEMATICS' ROLE IN THE TRAVEL MODEL DEVELOPMENT PROJECT WAS LED BY EARL R. RUITER (PROJECT MANAGER) AND MARVIN L. MANHEIM (PRINCIPAL-IN-CHARGE). A PRIMARY ROLE IN THE DEVELOPMENT OF THE CONCEPTUAL FRAMEWORK OF THE MODEL SYSTEM, AND IN THE SPECIFICATIONS OF THE COMPONENT MODELS, WAS PLAYED BY MOSHE E. BEN-AKIVA. THE ESTIMATION AND VALIDATION OF COMPONENT MODELS WAS CARRIED OUT BY EARL R. RUITER, RICHARD L. ALBRIGHT, ANTTI P. TALVITIE, JEFFREY M. MACMANN, AND LEONARD SHERMAN. RICHARD ALBRIGHT AND JEFFREY MACMANN CARRIED OUT THE IMPLEMENTATION OF THE MODELS IN COMPUTER PROGRAM PACKAGES. RICHARD ALBRIGHT, PATRICK O'KEEFE, AND EARL RUITER PLAYED MAJOR ROLES IN AGGREGATE VALIDATION. THE SERIES OF THREE VOLUMES OF THIS FINAL REPORT WERE INITIALLY DEVELOPED BY MOSHE BEN-AKIVA, EARL RUITER, RICHARD ALBRIGHT, AND DAVID R. WELLAND. HANNA KOLLO OF MTC AND EARL RUITER WERE RESPONSIBLE FOR THE REVIEW AND REVISION OF THE FINAL REPORT. PREFACE 2 SUPPORT STAFFS OF BOTH ORGANIZATIONS ASSISTED IN THE LARGE AMOUNT OF WORK NECESSARY TO PRODUCE THE REPORT IN FINAL FORM. THROUGHOUT THE TRAVEL MODEL DEVELOPMENT PROJECT, MONITORING OF OUR WORK AND ASSISTANCE IN CARRYING IT CUT WAS PROVIDED BY MTC STAFF MEMBERS GORDON SHUNK HANNA KOLLO, WILLIAM DAVIDSON, AND JOSE DA CUNHA. THEIR CONTRIBUTIONS ARE GRATEFULLY ACKNOWLEDGED. TABLE OF CONTENTS 1 TABLE OF CONTENTS PREFACE 1.0 INTRODUCTION 2.0 BACKGROUND 3.0 MODEL SYSTEM STRUCTURE 4.0 MODEL FORMS 5.0 LINKAGES BETWEEN THE MODELS 6.0 OVERVIEW OF THE COMPONENT MODEL DEVELOPMENT PROCESS 7.0 TRIPS REPRESENTED IN THE MODELS 8.0 THE COMPONENT MODELS - OVERVIEW 9.1 SROCC: HOME BASED WORK TRIP SHARED-RICE OCCUPANCY MODEL 9.2 NWHH: NON-WORKING HOUSEHOLD MODEL 9.3 NWHHAO: NON-WORKING HOUSEHOLD AUTO OWNERSHIP MODEL 9.4 WHHAO: WORKING HOUSEHOLD AUTO OWNERSHIP MODEL 9.5 PHBWG: PRIMARY WORKER HOPE BASED WORK TRIP GENERATION MODEL 9.6 PHBWA: PRIMARY WORKER HOME BASED WORK TRIP ATTRACTION MODEL 9.7 PHBWD: PRIMARY WORKER HOPE BASED WORK TRIP DISTRIBUTION MODEL 9.8 PHBWM: PRIMARY WORKER HOPE BASED WORK TRIP MODE CHOICE MODEL 9.9 SHBWG: SECONDARY WORKER HOME BASED WORK TRIP GENERATION MODEL 9.10 SHBWA: SECONDARY WORKER HOME BASED WORK TRIP ATTRACTION MODEL 9.11 SHBWD: SECONDARY WORKER HOME BASED WORK TRIP DISTRIBUTION MODEL 9.12 SHBWM: SECONDARY WORKER HOME BASED WORK TRIP MODEL CHOICE MODEL 9.13 HBSHG: HOME BASED SHOP TRIP GENERATION MODEL 9.14 HBSHD: HOME BASED SHOP TRIP DESTINATICN CHOICE MODEL 9.15 HBSHDM: HOME BASED SHOP TRIP DESTINATION/MODE CHOICE MODEL 9.16 HBSRG: HOME BASED SOCIAL-RECREATIONAL TRIP GENERATION MODEL 9.17 HBSRDM: HOME BASED SOCIAL-RECREATIONAL DESTINATION/MODE CHOICE MODEL 9.18 NHBAD: NON-HOME BASED AUTO TRIP DESTINATION MODEL 9.19 NHBAO: NON-HOME BASED AUTO TRIP ORIGIN MODEL 9.20 NHBTD: NON-HOME BASED TRANSIT TRIP DESTINATION MODEL 9.21 NHBTO: NON-HOME BASED TRANSIT TRIP ORIGIN MODEL 10.0 CONCLUSION 11.0 REFERENCES APPENDIX A: VARIABLE DIRECTORY INTRODUCTION 1-1 1. INTRODUCTION SIGNIFICANT ADVANCES HAVE BEEN MADE IN RECENT YEARS IN DEVELOPING AND APPLYING DESEGREGATE BEHAVIORAL TRAVEL DEMAND MODELS TO MANY ASPECTS OF URBAN TRAVEL DECISIONS (BEN-AKIVA, 1973; BEN-AKIVA AND LERMAN, 1974; CRA, 1972; MCFADDEN, 1974; THESE AND ALL OTHER REFERENCES ARE LISTED IN SECTION 11). WHAT HAS NOT PREVIOUSLY BEEN ACCOMPLISHED IS THE DEVELOPMENT OF A FULL SET OF URBAN MODELS AND THEIR INTEGRATION INTO A COMPLETE FORECASTING SYSTEM FOR USE BY A METROPOLITAN PLANNING ORGANIZATION. THE PURPOSE OF THIS SUMMARY REPORT IS TO PROVIDE AN OVERVIEW DESCRIPTION OF THE FIRST SUCH SYSTEM, WHICH HAS BEEN DEVELOPED AS THE MOCK TRAVEL DEMAND MODEL SYSTEM. TO ACCOMPLISH THIS PURPOSE, FIRST THE BACKGROUND OF THE MTC TRAVEL DEMAND MODEL DEVELOPMENT PROJECT WILL BE BRIEFLY SET OUT, FOLLOWED BY A DESCRIPTION OF THE STRUCTURE OF THE MODEL SYSTEM, THE SPECIFICATIONS OF EACH MODEL CONTAINED IN THE SYSTEM, AND THE APPLICATION PROCEDURES THAT HAVE BEEN DEVELOPED. BACKGROUND 2-1 2. BACKGROUND THE HISTORY OF MODELLING IN THE BAY AREA DATES BACK TO THE MIDDLE '60's WHEN THE BAY AREA TRANSPORTATION STUDY COMMISSION (BATSC) WAS CREATED TO CONDUCT COMPREHENSIVE TRANSPORTATION PLANNING IN THE REGION. BATSC CONCLUDED ITS TRAVEL SURVEYS IN 1965, REDUCED THE COLLECTED DATA INTO A USEABLE FORM AND PROCEEDED WITH MODEL DEVELOPMENT IN SUPPORT OF ITS PLANNING EFFORT. THE BATSC MODELS WERE OF THE TRADITIONAL AGGREGATE TYPE USED IN MOST METROPOLITAN AREAS IN THE '50's AND '60's. THEY INCLUDED TRIP GENERATION MODELS BASED ON LINEAR REGRESSION, DISTRIBUTION MODELS OF THE GRAVITY TYPE, DIVERSION CURVE MODE CHOICE MODELS, AND BOTH TRANSIT AND HIGHWAY NETWORK ASSIGNMENTS. BATSC CONCLUDED ITS WORK AND REPORTED ITS PLAN TO THE STATE LEGISLATURE IN 1969. TO CONTINUE THE TRANSPORTATION PLANNING FUNCTION IN THE BAY AREA AFTER 1969, THE STATE OF CALIFORNIA ENTERED INTO A JOINT AGREEMENT WITH THE ASSOCIATION OF BAY AREA GOVERNMENTS (ABAG) AND CREATED THE REGIONAL TRANSPORTATION PLANNING COMMITTEE (RTPC). THIS WAS A COMMITTEE OF ABAG WITH A SMALL CORE STAFF SUPPORTED BY THE STAFF OF DISTRICT 4, STATE DIVISION OF HIGHWAYS. NO SUBSTANTIVE MODELLING WORK WAS UNDERTAKEN BY RTPC. HOWEVER, MUCH OF THE BATSC TECHNICAL DOCUMENTATION WAS COMPLETED DURING THIS PERIOD. MOCK WAS CREATED BY AN ACT OF THE 1970 STATE LEGISLATURE TO ASSUME THE PERMANENT COMPREHENSIVE TRANSPORTATION PLANNING FUNCTION IN THE BAY AREA. THIS CONCLUDED THE RTPC FORMALLY. HOWEVER, A TRANSITION PERIOD WAS NEEDED FOR ORGANIZING THE COMMISSION AND STAFFING TO CARRY OUT ITS MISSION. AS THIS WAS GOING ON, THE BAY AREA RAPID TRANSIT DISTRICT (BARTD) WAS SECURING UMTA FUNDS TO UNDERTAKE BART EXTENSION STUDIES. BY 1972, FIVE MAJOR TRANSIT PLANNING STUDIES WERE UNDERWAY IN THE BAY AREA, CARRIED OUT JOINTLY BY BARTD, MOCK AND THE LOCAL JURISDICTIONS INVOLVED IN EACH STUDY. IT WAS THOUGHT, THEN, THAT A COMMON DATA BASE AND FORECASTING SYSTEM SHOULD BE USED FOR ALL THOSE STUDIES. THE REGIONAL TRANSIT TRAVEL PROJECTIONS PROJECT (RTTPP) WAS ESTABLISHED TO PRODUCE THE FORECASTING SYSTEM AND PROVIDE THE NECESSARY PROJECTIONS FOR THE STUDIES. BARTD MANAGED IT UNDER THE DIRECTION OF A TECHNICAL ADVISORY COMMITTEE FROM THE STAFFS OF BARTD, MOCK, AND THE LOCAL PARTICIPANTS. THE FIRM OF WILBUR SMITH AND ASSOCIATES WAS SELECTED AS A PRIMARY CONTRACTOR, AND DELEUW CATHER AND COMPANY AS A SUB- CONTRACTOR. MTC PROVIDED STAFF SUPPORT FOR THE PROJECT AS WELL AS PARTICIPATION AT THE TECHNICAL ADVISORY COMMITTEE. THE MODEL SYSTEMS DEVELOPED BY RTTPP WAS AN AGGREGATE ONE WITH THREE TRIP PURPOSES AND MANY DENSITY, EMPLOYMENT AND HOUSEHOLD TYPE STRATIFICATIONS. THE 1965 HOME INTERVIEW SURVEY AND THE BATSC UNADJUSTED DATA BASE WERE USED FOR THE DEVELOPMENT OF THE MODELS. THE TRIP PRODUCTION MODELS WERE TRIP RATES PER HOUSEHOLD OBTAINED USING CROSS-STRATIFICATION TECHNIQUES. THE TRIP ATTRACTION MODELS WERE TRIP RATE PER JOB MODELS, STRATIFIED BY EMPLOYMENT TYPE AND EMPLOYMENT DENSITY. TRIP DISTRIBUTION AND MODE CHOICE WERE MODELLED AS A JOINT DECISION USING A THEORETICAL PROBABILITY MODEL. MANY SIMPLIFYING ASSUMPTIONS WERE MADE FOR BACKGROUND 2-2 CALIBRATION CONVENIENCE. TRIAL AND ERROR WAS USED IN ESTIMATING THE GRAVITY DISTRIBUTION AND THE MODIFIED LOGIT FORM MODE CHOICE PARAMETERS FROM AGGREGATE DATA. THE MODEL DEVELOPMENT WAS COMPROMISED AND CURTAILED IN FAVOR OF PROVIDING PRELIMINARY PROJECTIONS AND DATA FOR THE STUDIES IT WAS INTENDED TO SERVE. AS WORK ON THE RTTPP MODEL NEARED COMPLETION, THE MTC STAFF SOUGHT NEW APPROACHES FOR IMPROVING THE ACCURACY OF BAY AREA TRAVEL MODELS, AND FOR INCREASING THE RESPONSIVENESS TO AGENCY INFORMATION NEEDS. DISCUSSIONS AND DELIBERATIONS LED TO A DECISION TO PURSUE A SUBSTANTIAL EFFORT ON MODEL DEVELOPMENT AS OPPOSED TO PATCHWORK IMPROVEMENTS OF THE RTTPP MODELS, OR A REINSTATEMENT OF THE BATSC MODELS. A PROCESS WAS OUTLINED FOR IDENTIFYING THE TYPES OF MODELS NEEDED IN THE BAY REGION. THE STATUS OF THE DATA BASE AND MODEL SYSTEMS WAS SUMMARIZED FOR QUALIFIED CONSULTANTS WHO WERE INVITED TO SUBMIT PROPOSALS IN RESPONSE TO AN RFP. THE TEAM OF COMSIS CORPORATION, BARTON-ASCHMAN ASSOCIATES, AND CAMBRIDGE SYSTEMATICS INC. WAS SELECTED TO DO THE WORK. THUS THE TRAVEL MODEL DEVELOPMENT PROJECT (TMDP) CAME INTO BEING. THE WORK STARTED IN JULY, 1975, WITH A TWO PHASE SCOPE OF BEST GEAR THE PROJECT TO MTC NEEDS. IN PHASE 1, THE CONSULTANTS REVIEWED MTC MODELING NEEDS AND DATA BASES. THEY DEFINED A GENERAL FRAMEWORK FOR A LONG TERM COMMITMENT TO MODELLING, FORECASTING AND DATA BASES FROM WHICH PLANNING STUDIES WOULD DRAW THEIR TECHNICAL DATA. THIS WOULD BE DEVELOPED INCREMENTALLY, AND WOULD BE BASED ON A SET OF DESEGREGATE MODEL COMPONENTS THAT COULD BE STRUCTURED IN DIFFERENT WAYS TO RESPOND TO AGENCY NEEDS. IN DETERMINING THE PURPOSES, NEEDS AND USE OF MODELS AT MTC, A STRUCTURED PROCESS WAS FOLLOWED. MTC STAFF WAS SURVEYED FOR THEIR NEEDS, AND EXTERNAL REPRESENTATIVES WERE INVITED TO SMALL GROUP WORKSHOPS TO AIR THEIR VIEWS AND HELP IDENTIFY RELEVANT ISSUES IN ORDER TO NARROW THE SCOPE OF WORK TO MATCH THE RESOURCES AVAILABLE. THE RESULTS WERE PRESENTED TO THE CONSULTANT TEAM TO USE IN THEIR FINAL RECOMMENDATIONS. PHASE 1 WAS CONCLUDED WITH A DECISION TO DEVELOP A SET OF DESEGREGATE MODEL COMPONENTS AND IMPLEMENT THEM IN A SOFTWARE PACKAGE ALONG THE LINES OF TRADITIONAL TRANSPORTATION ANALYSIS SYSTEMS. ADDITIONALLY, A SUBSET OF THESE MODEL COMPONENTS WAS TO BE IMPLEMENTED IN A SHORT RANGE GENERALIZED POLICY (SRGP) ANALYSIS PACKAGE TO FORM THE BASIS FOR SKETCH PLANNING AND FUTURE DEVELOPMENT AND APPLICATION TECHNIQUES. THE DATA BASE FOR MODEL DEVELOPMENT WAS THE 1965 BATSC HOME INTERVIEW SURVEY REFACTORED TO NEW ESTIMATES OF DWELLING UNITS (THE 1965 BACKCAST DATA), AND ADJUSTED TO SCREENLINE COUNTS. ALTHOUGH MANY SPECIAL PURPOSE SURVEYS HAVE BEEN CONDUCTED SINCE 1965, THE BATSC SURVEYS REMAINED THE MOST RECENT AND COMPLETE TRAVEL DATA SET. IN ADDITION ALL NECESSARY RELATED DATA WERE AVAILABLE FOR 1965: HIGHWAY AND TRANSIT NETWORKS, ZONAL LEVEL-OF-SERVICE DATA (ACCESS TIMES AND DISTANCES, PARKING COSTS, TOLLS, FARES, ETC.), AND LAND USE DATA (OBTAINED LARGELY FROM BACKGROUND 2-3 BACKCASTING 1970 CENSUS DATA TO 1965, USING THE PROJECTIVE LAND USE MODEL, PLUM). PHASE 2 WAS STARTED IN DECEMBER 1975, AND A FINAL SET OF MODELS WAS DELIVERED) TO MTC IN JUNE, 1977. MODEL SYSTEM STRUCTURE 3-1 3. MODEL SYSTEM STRUCTURE IN GENERAL, A TRAVEL DEMAND MODEL IS CONCERNED WITH THOSE HOUSEHOLD AND INDIVIDUAL DECISIONS THAT RESULT IN TRIPS BEING MADE. HOWEVER, SOME OTHER CHOICES ARE SO INTERRELATED WITH ACTUAL TRIPMAKING CHOICES THAT IT IS IMPOSSIBLE TO SEPARATE THEM FROM SUCH DECISIONS. FOR EXAMPLE, THE CHOICE OF RESIDENTIAL LOCATION IS NOT IN ITSELF A TRIPMAKING DECISION; HOWEVER, THE COMBINATION OF A WORKER'S EMPLOYMENT LOCATION CHOICE AND HIS OR HER HOUSEHOLD'S RESIDENTIAL LOCATION DECISION HAS AS ITS CONSEQUENCE A TRIP CHOICE, I.E., DAILY WORK TRIPS. FOR THIS REASON, THE GENERAL FRAMEWORK FROM WHICH THE MODELS DISCUSSED IN LATER SECTIONS OF THIS REPORT ARE DERIVED BEGINS WITH A PARTITION OF ALL POSSIBLE HOUSEHOLD AND HOUSEHOLD MEMBER DECISIONS INTO TWO SETS: THOSE RELEVANT TO TRANSPORTATION ANALYSIS AND THOSE THAT CAN, FOR PRACTICAL PURPOSES, BE IGNORED. THIS PARTITION PRODUCES THE FOLLOWING VECTOR OF TRAVEL-RELATED HOUSEHOLD DECISIONS: (1) EMPLOYMENT LOCATION (FOR ALL WORKERS) (2) RESIDENTIAL LOCATION (3) HOUSING TYPE (4) AUTOMOBILE OWNERSHIP (5) MODE OF TRAVEL TO WORK (FOR ALL WORKERS) (6) FREQUENCY (FOR NON-WORK TRIPS FOR EACH PURPOSE) (7) DESTINATION (FOR NON-WORK TRIPS OF EACH PURPOSE) (8) TIME OF DAY (FOR NON-WORK TRIPS OF EACH PURPOSE) (9) MODE OF TRAVEL (FOR NON-WORK TRIPS OF EACH PURPOSE) (10) ROUTE (FOR ALL TRIPS) FOR MOST SITUATIONS, THIS VECTOR DESCRIBES THE DECISIONS THAT A COMPLETE MODEL SYSTEM MUST CONSIDER. (THERE ARE POSSIBLE EXCEPTIONS. FREQUENCY AND TIME OF DAY FOR WORK TRIPS MAY BE DISTINCT CHOICES FOR SOME HOUSEHOLDS RATHER THAN CONSTRAINED BY THE WORK SITUATION. SOME DECISIONS MAY BE CONSTRAINED BY LACk OF INCOME OR LACK OF AN AVAILABLE AUTO, THOUGH THE LATTER MAY WELL BE PART OF SOME DECISION VECTORS.) IN THEORY, EACH DECISION MAY BE DEPENDENT ON THE REST. FOR EXAMPLE, WHERE ONE CHOOSES TO LIVE IS OBVIOUSLY LINKED TO THE HOUSING TYPE AND THE LEVEL OF AUTOMOBILE OWNERSHIP ONE SELECTS. SIMILARLY, SHOPPING TRIP DESTINATION AND MODE ARE LIKELY TO BE CLOSELY LINKED. IF CARRIED THROUGH COMPLETELY, THIS PERSPECTIVE WOULD PRODUCE A MODEL OF UNMANAGEABLE DIMENSIONS, SINCE THE ENTIRE VECTOR OF POSSIBLE CHOICES WOULD, FOR PRACTICAL PURPOSES, BE LIMITLESS, AND USEFUL MODELS WOULD BE IMPOSSIBLE TO DEVELOP. FORTUNATELY, THERE ARE SOME INTERRELATIONSHIPS AMONG COMPONENTS OF THIS VECTOR THAT ARE OF A FUNDAMENTALLY DIFFERENT CHARACTER THAN OTHERS. SOME OF THE DECISIONS, SUCH AS RESIDENTIAL LOCATION CHOICE, HAVE HIGH TRANSACTION COSTS AND ARE CONSEQUENTLY STABLE OVER FAIRLY LONG INTERVALS; OTHER CHOICES SUCH AS FREQUENCY OF SOCIAL AND RECREATIONAL TRIPS ARE ALTERED ON A DAILY BASIS. SOME DECISIONS ARE MORE LOGICALLY REPRESENTED AS BEING MADE COLLECTIVELY BY THE HOUSEHOLD, WHILE OTHERS CAN BE APPROXIMATED AS INDIVIDUAL CHOICES. MODEL SYSTEM STRUCTURE 3-2 THUS, IT IS POSSIBLE TO FORMULATE EXPLICIT BEHAVIORAL HYPOTHESES, AND TO ESTABLISH A STRUCTURE OF THE TOTAL VECTOR OF CHOICES AS A LOGICAL WORKING HYPOTHESIS. SUCH AN EXPLICIT STRUCTURE GREATLY SIMPLIFIES MODEL DEVELOPMENT. THIS STRUCTURE IS TERMED A HIERARCHY OF CHOICE (BEN-AKIVA ET AL, 1979). FIGURE 3-1 ILLUSTRATES THE THREE-STAGE CHOICE HIERARCHY REPRESENTED BY THE MTC MODEL SYSTEM. AT THE HIGHEST LEVEL COME URBAN DEVELOPMENT DECISIONS THAT ARE LONG-RUN IN NATURE: EMPLOYERS DECIDE WHERE TO PROVIDE JOBS AND DEVELOPERS DECIDE WHERE TO PROVIDE HOUSING OF VARIOUS TYPES. NEXT COME HOUSEHOLD MOBILITY DECISIONS MADE MORE FREQUENTLY: WHERE TO LIVE AND WORK, HOW MANY HOUSEHOLD MEMBERS WILL HAVE JOBS, HOW OFTEN EACH WILL GO TO WORK, HOW MANY AUTOS WILL BE OWNED, AND WHAT MODES WILL BE USED TO MAKE WORK TRIPS. FINALLY COME SHORT-RUN TRAVEL DECISIONS MADE ALMOST DAILY: FREQUENCY, DESTINATION, AND MODE FOR NON-WORK TRIPS; TIME-OF-DAY AND ROUTE FOR ALL TRIPS. THE URBAN DEVELOPMENT MODEL SYSTEM IMPLEMENTED BY THE ASSOCIATION OF BAY AREA GOVERNMENTS (ABAG) PREDICTS EACH OF THE DEVELOPMENT DECISIONS SHOWN IN FIGURE 3-1 PLUS THE RESIDENTIAL LOCATION DECISION. THE TRAVEL DEMAND MODELS DEVELOPED IN THE THIS PROJECT PREDICT EACH OF THE REMAINING HOUSEHOLD DECISIONS. FIGURE 3-2 SHOWS, IN FURTHER DETAIL, HOW THE MTC TRAVEL DEMAND MODELS ARE INTERRELATED WITHIN THE GENERAL CHOICE STRUCTURE SHOWN IN FIGURE 3- 1. EACH BLOCK IN FIGURE 3-2 REPRESENTS ONE OR MORE DESEGREGATE (ESTIMATED USING OBSERVED HOUSEHOLD OR TRIPMAKER BEHAVIOR) TRAVEL DEMAND MODELS. THE STRUCTURE IS STRONGLY RELATED TO THE CONCEPTUAL SYSTEM OF DESEGREGATE TRAVEL DEMAND MODELS SHOWN IN FIGURE 3-1, BUT INCORPORATES A NUMBER OF FEATURES AND APPROXIMATIONS NECESSARY FOR PRODUCING A PRACTICAL REGIONAL FORECASTING SYSTEM. THESE ARE: (1) THERE ARE TWO MOBILITY BLOCKS, ONE FOR HOUSEHOLDS WITH WORKERS AND ONE FOR HOUSEHOLDS WITHOUT WORKERS. THE RESIDENTIAL LOCATION AND HOUSING TYPE CHOICES ARE EXTERNAL TO THE PRESENT MODEL DEVELOPMENT EFFORT, AS THEY ARE PREDICTED BY ABAG'S URBAN DEVELOPMENT MODEL SYSTEM (WHICH ALSO PREDICTS INCOME DISTRIBUTIONS AND WORK FORCE BY ZONE). (2) THE FIRST MOBILITY CHOICE BLOCK DISTINGUISHES BETWEEN PRIMARY AND SECONDARY WORKERS IN A HOUSEHOLD. EACH HOUSEHOLD WITH WORKERS HAS ONLY ONE PRIMARY WORKER, OR BREADWINNER. ALL ADDITIONAL WORKERS ARE TERMED SECONDARY. (3) THE MODELLING SYSTEM DEALS SEPARATELY WITH HOME BASED AND NON- HOME BASED TRIPS. THIS SIMPLIFIES THE REPRESENTATION OF TRIP CHAINS (I.E., A TRIP FROM HOME, FOLLOWED BY ONE OR MORE NON- HOME BASED TRIPS, FOLLOWED FINALLY BY A TRIP HOME), AN AREA IN WHICH BASIC CONCEPTUAL DEVELOPMENT IS CONTINUING (ADLER, 1976). ALSO, IT ALLOWS THE MODEL SYSTEM TO DEAL WITH ONE-WAY TRIPS IN ACCORDANCE WITH TRADITIONAL PRACTICE, RATHER THAN WITH THE ROUND TRIPS MORE COMMONLY CONSIDERED IN MODEL SYSTEM STRUCTURE 3-3 DESEGREGATE MODELLING. (4) A SERIES OF SEQUENTIAL MODELS HAS BEEN DEVELOPED FOR A NUMBER OF CLOSELY RELATED TRAVEL DECISIONS FOR WHICH JOINT MODELS HAVE BEEN PREVIOUSLY USED. EXAMPLES ARE AUTO OWNERSHIP AND MODE CHOICE FOR PRIMARY WORKERS, AND NON-WORK TRIP FREQUENCY, DESTINATION AND MODE CHOICE. HOWEVER, AND NON-WORK TRIP FREQUENCY, DESTINATION AND MODE CHOICE. HOWEVER, DUE TO THE STRUCTURES OF THESE SEQUENTIAL MODELS (SEE NEXT ITEM), JOINT EFFECTS ARE NOT IGNORED. (5) JOINT EFFECTS ARE INCLUDED IN THE SEQUENTIAL STRUCTURE BY MEANS OF "FEEDBACK" LOOPS, SHOWN IN FIGURE 3-2 BY THE ***> ARROWS. EACH OF THESE REPRESENTS AN ACCESSIBILITY-LIKE VARIABLE IN THE HIGHER-LEVEL MODEL (E.G., AUTO OWNERSHIP FOR HOUSEHOLDS WITHOUT WORKERS) WHICH IS OBTAINED FROM A LOWER- LEVEL MODEL (E.G., HOME BASED OTHER DESTINATION AND MODE CHOICE). EACH OF THESE VARIABLES IS A COMPOSITE OF ALL THE VARIABLES OF THE LOWER-LEVEL MODEL. AN EXAMPLE OF SUCH A VARIABLE IS THE NATURAL LOGARITHM OF THE DENOMINATOR OF THE WORK MODE CHOICE MODEL, WHICH IS USED AS A VARIABLE IN THE WORK DESTINATION CHOICE MODELS. THESE VARIABLES ALLOW CONSISTENT REPRESENTATION OF LEVEL-OF-SERVICE EFFECTS IN SPITE OF THE SEQUENTIAL STRUCTURE OF THE SEPARATE MODELS. (6) THE TIME-OF-DAY DECISION IS MODELLED USING HISTORICAL PEAKING CHARACTERISTICS RATHER THAN A CHOICE MODEL BASED ON THE RELATIONSHIP BETWEEN PEAK AND OFFPEAK TRANSPORTATION SYSTEM CHARACTERISTICS. (7) THE VEHICLE OCCUPANCY CHOICE DECISION FOR NON-WORK TRAVEL IS MADE USING HISTORICALLY OBSERVED RATES RATHER THAN DESEGREGATE CHOICE MODELS. (8) THE ROUTE CHOICE DECISION IS MODELLED USING CONVENTIONAL ALL- OR-NOTHING AND CAPACITY RESTRAINT ASSIGNMENT TECHNIQUES. MODEL SYSTEM STRUCTURE 3-4 Click HERE for graphic. MODEL SYSTEM STRUCTURE 3-5 Click HERE for graphic. MODEL SYSTEM STRUCTURE 3-6 Click HERE for graphic. MODEL FORMS 4-1 4. MODELS FORMS EACH OF THE MODELS SHOWN IN FIGURE 3-2 TAKES ONE OF THREE GENERAL MATHEMATICAL FORMS, WHICH ARE DESCRIBE BRIEFLY IN THIS SECTION. 4.1 THE LOGIT FORM MOST OF THE COMPONENT MODELS TAKE THE LOGIT FORM WHICH CAN BE DERIVED AS A REPRESENTATION OF UTILITY MAXIMIZATION AMONG A DISCRETE SET OF ALTERNATIVES UNDER UNCERTAINTY (MCFADDEN, 1974; RICHARDS AND BEN-AKIVA, 1975). THE RESULTING PROBABILITY MODELS TAKE THE FOLLOWING FORM: P(I:A(Q)) :: = EXP(U(I,Q))/(EXP(U(K,Q), SUMMED OVER K IN A(Q)) WHERE: P(I:A(Q)) IS THE PROBABILITY OF DECISION-MAKER Q CHOOSING ALTERNATIVE I FROM THE SET OF AVAILABLE ALTERNATIVES, A(Q). U(K,Q) IS THE UTILITY OF ALTERNATIVE K TO Q. IN A LOGIT FORM MODEL, U(K, Q IS A LINEAR FUNCTION OF ANY NUMBER OF TERMS, EACH TERM BEING THE PRODUCT OF A COEFFICIENT AND AN INDEPENDENT VARIABLE: U(K,Q) :: = (B(J)*X(J,K,Q), SUMMED OVER J) WHERE: B(J) IS THE COEFFICIENT OF THE J-TH TERM. X(J,K,Q) IS THE INDEPENDENT VARIABLE OF THE J-TH TERM WHICH DESCRIBES ALTERNATIVE K AND/OR DECISION-MAKER Q. B, THE VECTOR OF COEFFICIENTS, CAN BE ESTIMATED USING A SAMPLE OF OBSERVED CHOICES AND MAXIMUM LIKELIHOOD TECHNIQUES. FOR EACH DECISION-MAKER, THE SAMPLE MUST INCLUDE INDEPENDENT VARIABLES DESCRIBING HIM AND HIS ACTUAL CHOICE OF ALTERNATIVE, AND, FOR EACH ALTERNATIVE AVAILABLE TO HIM, THE INDEPENDENT VARIABLES DESCRIBING IT. 4.2 THE LINEAR FORM SOME OF THE COMPONENT MODELS TAKE A LINEAR FORM IN WHICH AN EXPECTED VALUE (RATHER THEN A PROBABILITY, AS IN THE CASE OF THE LOGIT FORM OF THE DEPENDENT VARIABLE IS PREDICTED: E(S) :: = (B(J)*X(J,S), SUMMED OVER J) WHERE: E(S) IS THE PREDICTED VALUE OF THE DEPENDENT VARIABLE IN SITUATION S. B(J) IS THE COEFFICIENT OF THE J-TH TERM. X(J,S) IS THE INDEPENDENT VARIABLE OF THE J-TH TERM WHICH DESCRIBES SITUATION S. B, THE VECTOR OF COEFFICIENTS, CAN BE ESTIMATED USING LINEAR MODEL FORMS 4-2 REGRESSION BY SUPPLYING, FOR EACH SITUATION, THE INDEPENDENT VARIABLES DESCRIBING IT AND THE OBSERVED VALUE OF THE DEPENDENT VARIABLE. 4.3 THE NON-LINEAR FORM ANOTHER FORM EMPLOYED FOR A FEW OF THE COMPONENT MODELS IS ONE IN WHICH EXPECTED VALUES ARE PREDICTED USING A GENERAL NON-LINEAR FUNCTION OF THE INDEPENDENT VARIABLES. THIS FORM IS A GENERALIZATION OF THE PREVIOUS ONE: E(S) :: = F(B,X(S)) WHERE: E(S) IS THE PREDICTED VALUE OF THE DEPENDENT VARIABLE IN SITUATION S. B IS A VECTOR OF COEFFICIENTS. X(S) IS A VECTOR OF INDEPENDENT VARIABLES DESCRIBING SITUATION S. F IS A NON-LINEAR FUNCTIONAL FORM. B, THE VECTOR OF COEFFICIENTS, CAN BE ESTIMATED FOR A NUMBER OF FUNCTIONAL FORMS USING NON-LINEAR REGRESSION BY SUPPLYING, FOR EACH SITUATION, THE INDEPENDENT VARIABLES DESCRIBING IT AND THE OBSERVED VALUE OF THE DEPENDENT VARIABLE. LINKAGES BETWEEN THE MODELS 5-1 5. LINKAGES BETWEEN THE MODELS THE COMPONENT MODELS SHOWN IN FIGURE 3-2 ARE LINKED IN TWO WAYS: (1) "LOWER LEVEL" MODELS ARE CONDITIONAL ON THE PREDICTED CHOICES BY "HIGHER LEVEL" MODELS, AS INDICATED BY THE SOLID ARROWS. (2) FEEDBACK IN THE FORM OF COMPOSITE, OR ACCESSIBILITY, VARIABLES CALCULATED USING "LOWER LEVEL" MODELS AND INCLUDED IN "HIGHER LEVEL" MODELS, AS INDICATED BY THE ***> ARROWS. THE FIRST TYPE OF LINKAGE IS DETERMINED BY THE ASSUME CHOICE HIERARCHY AND THE RESULTING SEQUENCE OF MODELS. VARIABLES RESULTING FROM "HIGHER LEVEL" CHOICES ARE PREDETERMINED FOR "LOWER LEVEL" CHOICES, AND ARE ATTRIBUTES OF THE HOUSEHOLD OR THE INDIVIDUAL THAT DO NOT VARY AMONG ALTERNATIVE "LOWER LEVEL" CHOICES. FOR EXAMPLE, AUTO OWNERSHIP IS TREATED AS A HOUSEHOLD CHARACTERISTIC IN THE NON-WORK MODELS, AFTER IT IS PREDICTED BY THE "HIGHER LEVEL" AUTO OWNERSHIP MODELS. THE COMPOSITE VARIABLES REPRESENT EXPECTATIONS OF THE OUTCOMES OF "LOWER LEVEL" CHOICES WHICH COULD BE DIFFERENT AMONG ALTERNATIVES OF "HIGHER LEVEL" CHOICES. FOR EXAMPLE, LEVEL OF SERVICE BY TRANSIT FOR SHOPPING TRIPS AFFECTS AUTO OWNERSHIP. HOWEVER, THIS VARIABLE DEPENDS ON THE HOUSEHOLD CHOICE OF THE SHOPPING TRIPS, A DECISION WHICH IS MADE ONLY CONDITIONAL ON THE HOUSEHOLD AUTO OWNERSHIP. THUS, THE SPECIFIC SHOPPING LEVEL OF SERVICE IS INDETERMINATE IN THE CHOICE OF AUTO OWNERSHIP. HOWEVER, COMPOSITE VARIABLES REPRESENTING OVERALL SHOPPING LEVEL OF SERVICE FOR ALTERNATIVE AUTO OWNERSHIP LEVELS CAN BE DETERMINED. THE ATTRIBUTES WHICH VARY AMONG "LOWER LEVEL" CHOICES ARE AGGREGATED AND INCLUDED AS COMPOSITE VARIABLES IN THE MODELS OF "HIGHER LEVEL" CHOICES. ALL SYSTEMS OF TRAVEL DEMAND MODELS INCLUDE, TO SOME EXTENT, SUCH COMPOSITE VARIABLES. EXAMPLES ARE WEIGHTED "INCLUSIVE PRICES" (CRA, 1972), AND THE TRANSIT ACCESSIBILITY VARIABLE USED IN A TRIP GENERATION MODEL BY THE METROPOLITAN WASHINGTON COUNCIL OF GOVERNMENTS (1974). HOWEVER, THE FORMULATION OF THESE COMPOSITE VARIABLES IS OFTEN ARBITRARY AND RESULTS IN COUNTER-INTUITIVE PREDICTIONS. THE COMPOSITE VARIABLES USED IN THE MTC MODEL SYSTEM ARE DERIVED IN A WAY WHICH IS CONSISTENT WITH THE UNDERLYING ASSUMPTION OF THE MODELS. IF A "LOWER LEVEL" CHOICE IS MODELLED USING THE LOGIT MODEL, THE COMPOSITE VARIABLE DEFINED OVER THESE CHOICE ALTERNATIVES IS CONSTRUCTED AS THE EXPECTED MAXIMUM UTILITY FROM THIS CHOICE PROCESS. IF THE OUTCOME OF THIS CHOICE WERE KNOWN, THEN THE COMPOSITE VARIABLE WOULD BE TAKEN AS THE EXPECTED UTILITY OF THE CHOSEN ALTERNATIVE. HOWEVER, SINCE IT IS ASSUMED, IN DEVELOPING A CHOICE MODEL SUCH AS LOGIT, THAT THE ALTERNATIVE WITH THE HIGHEST UTILITY IS SELECTED, WE CAN CALCULATE THE EXPECTED VALUE OF THE MAXIMUM UTILITY. FOR THE LOGIT MODEL SHOWN IN 4.1, THIS IS EQUAL TO THE FOLLOWING EXPRESSION (SEE, FOR EXAMPLE, BEN-AKIVA AND LERMAN, 1977): LINKAGES BETWEEN THE MODELS 5-2 LN(EXP(U(K,Q)), SUMMED OVER K IN A(Q)) THIS IS THE NATURAL LOGARITHM OF THE DENOMINATOR OF THE LOGIT FUNCTION WHICH MUST ALWAYS BE COMPUTED IN FORECASTING, AND THEREFORE REQUIRES LITTLE OR NO ADDITIONAL COMPUTATION. IN THE MTC MODEL SYSTEM SUCH VARIABLES WERE USED IN SEVERAL MODELS AS THE EXPECTED MAXIMUM UTILITY FROM AN ENTIRE SET OF ALTERNATIVES OR AS THE EXPECTED MAXIMUM UTILITY FROM A SUBSET OF THE ALTERNATIVES. FOR EXAMPLE, THE SHOPPING TRIP FREQUENCY MODEL INCLUDES AS AN INDEPENDENT VARIABLE THE NATURAL LOGARITHM OF THE DENOMINATOR OF THE SHOPPING DESTINATION AND MODE CHOICE MODEL. THE AUTO OWNERSHIP MODEL INCLUDES THE RATIOS OF EXPECTED MAXIMUM UTILITIES FOR SHOPPING TRAVEL BY AUTO AND TRANSIT FOR DIFFERENT AUTO OWNERSHIP LEVELS, WHICH ARE CALCULATED USING APPROPRIATE PARTIAL SUMS OF THE DENOMINATOR. LARGER VALUES OF THIS RATIO INDICATE A GREATER NEED FOR A CAR FOR SHOPPING TRAVEL WHICH WILL THEREFORE RESULT IN INCREASED AUTO OWNERSHIP. OVERVIEW OF THE COMPONENT MODEL DEVELOPMENT PROCESS 6-1 6. OVERVIEW OF THE COMPONENT MODEL DEVELOPMENT PROCESS FOLLOWING THE SPECIFICATION OF THE OVERALL MODEL SYSTEM STRUCTURE DESCRIBED IN SECTION 3, EACH COMPONENT MODEL HAS DEVELOPED BY CARRYING OUT THE FIVE-STEP PROCESS DESCRIBED. AS IN ANY EMPIRICAL MODEL DEVELOPMENT EFFORT, THE ESSENTIAL STEPS IN DEVELOPING COMPONENT TRAVEL DEMAND MODELS ARE THE FOLLOWING: (1) DEVELOPMENT OF ONE OR MORE CANDIDATE MODEL SPECIFICATIONS, INCLUDING FUNCTIONAL FORM, DEPENDENT VARIABLE, AND INDEPENDENT VARIABLES. THESE MODEL SPECIFICATIONS CAN BE BASED ON A PRIOR 1 KNOWLEDGE OF THE PHENOMENA BEING MODELLED AND/OR ON CONCEPTUAL AND THEORETICAL STRUCTURES OF THE PHENOMENA. (2) ESTIMATION OF THE COEFFICIENTS OF ALTERNATIVE MODEL SPECIFICATIONS, LEADING POSSIBLY TO THE DEVELOPMENT OF ADDITIONAL MODEL SPECIFICATIONS BASED ON THE RESULTS OBTAINED IN EARLY ESTIMATIONS. (3) VALIDATION OF A SUBSET OF THE ALTERNATIVE MODEL SPECIFICATIONS ACCOMPLISHED BY COMPARING MODEL PREDICTIONS WITH OBSERVED DATA. IDEALLY, THE OBSERVED DATA SHOULD DIFFER FROM THAT USED IN THE MODEL ESTIMATION PROCESS. (4) IMPLEMENTATION OF THE MODELS AS PART OF A PROCESS THAT CAN BE USED TO FORECAST THE VALUE OF THE DEPENDENT VARIABLE FOR VARIOUS COMBINATIONS OF INDEPENDENT VARIABLE VALUES. IN THE PARTICULAR CASE OF DESEGREGATE TRAVEL DEMAND MODELS, THE MODEL ESTIMATION STEP IS DONE USING DESEGREGATE TRAVEL DATA, REPRESENTING INDIVIDUAL HOUSEHOLDS, INDIVIDUALS, OR TRIPS. MODEL IMPLEMENTATION, HOWEVER, MUST INCLUDE A PROCESS TO AGGREGATE MODEL FORECASTS TO REPRESENT ENTIRE POPULATIONS OF TRIPMAKERS RATHER THAN INDIVIDUALS. FOR THIS REASON, THE VALIDATION STEP WAS DONE TWO WAYS: (1) DESEGREGATE VALIDATION, IN WHICH THE PREDICTED AND OBSERVED BEHAVIOR OF INDIVIDUAL HOUSEHOLDS, INDIVIDUALS, OR TRIPS ARE COMPARED. (2) AGGREGATE VALIDATION, IN WHICH OBSERVED TOTALS BY ZONE (OR OTHER UNIT TO WHICH INDIVIDUAL PREDICTIONS ARE AGGREGATED) ARE COMPARED WITH THE RESULTS PREDICTED BY AN AGGREGATE VERSION OF THE IMPLEMENTED MODEL. THE TRAVEL DEMAND MODELLING PROJECT INCLUDED EACH OF THE STEPS DESCRIBED ABOVE. A GENERAL OVERVIEW THAT SHOWS THE INTER- RELATIONSHIPS OF THESE STEPS, AS WELL AS THE MAJOR COMPUTER PROGRAMS USED IN EACH STEP, IS SHOWN IN FIGURE 6-1. THE SUBSECTIONS THAT FOLLOW DESCRIBE, FIRST, THE KINDS OF INPUT DATA USED AND THEIR SOURCES AND, SECOND, THE PROGRAMS AND PROCEDURES EMPLOYED IN EACH OF THE STEPS IN MODEL DEVELOPMENT. OVERVIEW OF THE COMPONENT MODEL DEVELOPMENT PROCESS 6-2 6.1 INPUT DATA FOR MODEL DEVELOPMENT THE BASIC INPUTS FALL INTO THREE CLASSES: (1) HOME INTERVIEW DATA: INCLUDED ARE SOCIO-ECONOMIC AND TRIPMAKING DATA FOR HOUSEHOLDS, INDIVIDUALS, AND TRIPS. THEY ARE OBTAINED FROM TWO SUBSAMPLES OF THE 1965 BATSC HOME SURVEY; ONE USED FOR MODEL ESTIMATION, AND ONE FOR DESEGREGATE MODEL VALIDATION. EACH OF THESE SUBSAMPLES WAS DESEGREGATE MODEL VALIDATION. EACH OF THESE SUBSAMPLES WAS OF ALL HOUSEHOLDS IN A DIFFERENT SET OF ZONES, SELECTED RANDOMLY WITHIN EACH OF THE NINE COUNTIES OF THE STUDY AREA. BE SELECTED PER COUNTY TO PREVENT UNDERREPRESENTATION AT THE COUNTY LEVEL. THE RESULTING CALIBRATION SUBSAMPLE CONTAINS 1641 HOUSEHOLDS IN 18 ZONES; THE VALIDATION SUBSAMPLE CONTAINS 1715 HOUSEHOLDS IN 16 ZONES. (2) SOCIO-ECONOMIC DATA: THESE DATA ITEMS EXIST FOR EACH OF THE 290 MTC TRAFFIC ANALYSIS ZONES IN THE 1965 ZONAL SYSTEM. THEY INCLUDE POPULATION, HOUSING, EMPLOYMENT, AND LAND USE ACREAGE DATA. (3) LEVEL-OF-SERVICE DATA: INCLUDED ARE ZONE-TO-ZONE TIMES, COSTS, AND DISTANCES BY MODE, WITH VARIOUS COMPONENTS OF TRAVEL TIME (IN-VEHICLE, WAITING, WALKING, TRANSFERRING) SEPARATELY SPECIFIED. ALSO INCLUDED ARE ZONAL VALUES REPRESENTING HIGHWAY NETWORK ACCESS/EGRESS TIMES, COSTS, AND DISTANCES. 6.2 DESEGREGATE ESTIMATION BECAUSE THE PROGRAMS USED IN THE DESEGREGATE ESTIMATION STEP SHOWN IN FIGURE 6-1 DEPEND ON THE MODEL FORM BEING ESTIMATED, THE ESTIMATION PROCESS AS IT APPLIES TO EACH MODEL FORM IS DISCUSSED SEPARATELY. 6.2.1 LOGIT FORM MODELS FOR LOGIT FORM MODELS, THE MODEL ESTIMATION STEP INVOLVES THE USE OF LOGIN, AN INPUT DATA PREPARATION PROGRAM THAT COMBINES THE THREE TYPES OF INPUT DATA INTO A FILE CONTAINING ONE RECORD FOR EACH OBSERVED HOUSEHOLD, INDIVIDUAL, OR TRIP CHOSEN FOR INCLUSION IN THE ESTIMATION SAMPLE. EACH OF THESE RECORDS CONSISTS OF THE VALUES OF EACH VARIABLE CHOSEN FOR INCLUSION IN A PARTICULAR CANDIDATE MODEL SPECIFICATION OF EACH MODEL THIS PROGRAMMING TASK IS MINIMIZED, HOWEVER, THROUGH THE USE OF A STANDARD LOGIN SKELETON THAT HANDLES STANDARD TYPES OF INPUT AND OUTPUT, PROVIDES OVERALL PROGRAM CONTROL, AND PROVIDE USER ENTRY POINTS AT WHICH MODEL- AND SPECIFICATION-SPECIFIC PROGRAM SEGMENTS CAN BE INSERTED. ONCE THE MODEL-SPECIFIC PROGRAMMING HAS BEEN DONE FOR A PARTICULAR MODEL, SPECIFICATION-SPECIFIC PROGRAMMING OVERVIEW OF THE COMPONENT MODEL DEVELOPMENT PROCESS 6-3 USUALLY INVOLVES MINOR CHANGES (CAMBRIDGE SYSTEMATICS, 1976A). THE ESTIMATION PROGRAM USED IN THE DESEGREGATE ESTIMATION STEP FOR LOGIT FORM MODELS IS LOGIT. LOGIT READS THE OBSERVATION FILE GENERATED BY LOGIN, PLUS USER CONTROL, AND USES AN ITERATIVE MAXIMUM LIKELIHOOD ESTIMATION METHOD TO PRODUCE STATISTICAL ESTIMATES OF THE MODEL COEFFICIENT VECTOR B, AS WELL AS VARIOUS STATISTICAL MEASURES OF GOODNESS OF FIT (CAMBRIDGE SYSTEMATICS, 1976B). 6.2.2. LINEAR FORM MODELS FOR LINEAR FORM MODELS, THE LINEAR REGRESSION CAPABILITIES OF SPSS (NIE, 1975) WERE USED FOR MODEL ESTIMATION. IN MOST CASES, SPECIAL VERSIONS OF THE INPUT DATA PREPARATION PROGRAM LOGIN WERE USED TO CREATE THE SPSS INPUT FILE FOR THESE MODELS. THESE PROGRAMS SELECTED OBSERVED HOUSEHOLDS, INDIVIDUALS, OR TRIPS FOR INCLUSION IN THE ESTIMATION SAMPLE, AND COMBINED ALL OF THE POTENTIAL VARIABLES OF EACH OF THE INPUT DATA TYPES DESCRIBED IN 6.1 INTO THE RECORD FOR EACH SITUATION (OR "CASE", TO USE SPS TERMS). DUE TO THE ABILITY OF SPSS TO SELECT AND TRANSFORM VARIABLES IN ITS INPUT FILE PRIOR TO ESTIMATION, IT IS NORMALLY NECESSARY TO DEVELOP JUST A SINGLE LOGIN PROGRAM PER LINEAR MODEL. 6.2.3 NON-LINEAR FORM MODELS A CAMBRIDGE SYSTEMATICS PROGRAM FOR NON-LINEAR REGRESSION, JMVX85, A REVISION OF THE BIOMEDICAL COMPUTER PROGRAM BMDX85, WAS USED TO ESTIMATE ALL NON-LINEAR FORM MODELS. THE BMD PROGRAM BMD07R (DIXON, 1973) COULD ALSO BE USED TO ESTIMATE THESE MODELS. SPECIAL VERSIONS OF THE INPUT DATA PREPARATION PROGRAM LOGIN WERE USED TO CREATE THE INPUT FILES TO JMVX85, WITH A SINGLE LOGIN PROGRAM REQUIRED PER MODEL, AS FOR LINEAR FORM MODELS. EACH ALTERNATIVE SPECIFICATION OF A GIVEN MODEL REQUIRES A MINIMUM OF SPECIAL CODING (FOR DATA TRANSFORMATIONS AND THE EVALUATION OF PARTIAL DERIVATIVES) OF TWO OF THE SUBROUTINES OF THE JMVX85 ESTIMATION PROGRAM. 6.3 DESEGREGATE VALIDATION AS FOR MODEL ESTIMATION, THE PROGRAMS USED FOR DESEGREGATE VALIDATION DEPEND ON THE MODEL FORM. 6.3.1 LOGIT FORM MODELS ONCE A PROVISIONALLY ACCEPTABLE MODEL SPECIFICATION WAS OBTAINED IN THE ESTIMATION STEP, DESEGREGATE VALIDATION OF THE LOGIT FORM MODELS WAS DONE BY EXPANDING THE LOGIN INPUT PREPARATION PROGRAMS USED PRIOR TO MODEL ESTIMATION TO PERFORM THE FOLLOWING TASKS: (1) THE ESTIMATED MODEL IS APPLIED TO EACH HOUSEHOLD, INDIVIDUAL, OR TRIP SELECTED FROM THE DESEGREGATE VALIDATION FILE, PRODUCING AN ESTIMATED PROBABILITY OF CHOICE OF EACH AVAILABLE ALTERNATIVE. OVERVIEW OF THE COMPONENT MODEL DEVELOPMENT PROCESS 6-4 (2) AFTER EXPANSION USING THE APPROPRIATE SURVEY FACTOR, THE PREDICTED PROBABILITIES OF EACH ALTERNATIVE, AS WELL AS THE OBSERVED CHOICE, ARE SUMMED TO PRODUCE SUBTOTALS SPECIALLY DEFINED FOR EACH MODEL. THESE SUBTOTALS WERE DESIGNED TO DISPLAY THE MODEL'S PERFORMANCE BY SUCH RELEVANT STRATIFICATIONS AS HOUSEHOLD SIZE, HOUSEHOLD INCOME, TRIP LENGTH, ETC. SUMMARY TABLES CONTAINING THE STRATIFIED OBSERVED AND PREDICTED SUBTOTALS ARE PRINTED. (3) FOR EACH CELL IN THE SUMMARY TABLES, A 67 PERCENT ACCEPTABILITY RANGE IS COMPUTED BASED ON THE SAMPLING RATE OF THE SURVEY FROM WHICH THE OBSERVATIONS CAME AND THE OBSERVED VOLUME IN THE CELL. THE RANGE GIVES THE +/- ONE STANDARD DEVIATION BOUNDS WITHIN WHICH THE TRUE OBSERVED NUMBER WILL FALL 67 TIMES OUT OF 100 (ASSUMING A NORMAL DISTRIBUTION). FOR EACH CELL, IF THE PREDICTED VALUE FALLS WITHIN THIS RANGE, NO MODEL ERROR IS IMPLIED. FOR ALL CELLS, NO MODEL ERROR IS IMPLIED IF 67 PERCENT OF THEM HAVE PREDICTED VALUES WITHIN THE ACCEPTABILITY RANGES. THE OUTPUTS BY CELL ARE ALSO USEFUL FOR IDENTIFYING MODEL BIASES, SUCH AS THE CONSISTENT OVER- PREDICTION OF TRANSIT USE FOR TRIPS BY HIGH INCOME TRAVELERS. 6.3.2 LINEAR AND NON-LINEAR FORM MODELS DESEGREGATE VALIDATION OF BOTH THE LINEAR AND NON-LINEAR FORM MODELS WAS ACCOMPLISHED USING BOTH LOGIN AND SPSS PROGRAMS. THE FIRST STEP IN THIS PROCESS WAS THE USE OF THE SAME LOGIN PROGRAM USED FOR MODEL ESTIMATION, APPLIED TO THE VALIDATION (RATHER THAN ESTIMATION) DATA SET OF HOUSEHOLDS, INDIVIDUALS, OR TRIPS. THE DATA SET CREATED BY LOGIN WAS THEN INPUT TO SPSS, AND ITS VARIABLE TRANSFORMATION AND ARITHMETIC CAPABILITIES WERE USED TO COMPUTE THE PREDICTED EXPECTED VALUES FOR EACH SAMPLED OBSERVATION. FOLLOWING THIS COMPUTATION, THE SPSS CROSS TABULATION, BREAKDOWN, SCATTERGRAM, AND T-TEST CAPABILITIES WERE USED TO OBTAIN MEAN PREDICTED AND OBSERVED EXPECTED VALUES BY MARKET SEGMENT, TO TEST THE DIFFERENCES BETWEEN THESE MEANS FOR SIGNIFICANCE, AND TO PRINT SCATTERGRAMS OF PREDICTED VS. OBSERVED VALUES. OVERVIEW OF THE COMPONENT MODEL DEVELOPMENT PROCESS 6-5 Click HERE for graphic. OVERVIEW OF THE COMPONENT MODEL DEVELOPMENT PROCESS 6-6 6.4 MODEL IMPLEMENTATION AS PART OF THE TRAVEL MODEL DEVELOPMENT PROJECT, TWO COMPUTERIZED PROCEDURES WERE DEVELOPED TO IMPLEMENT THE DEMAND MODELS DESCRIBED IN THIS REPORT. THE FIRST IS ORIENTED TOWARD DETAILED REGIONAL NETWORK ANALYSIS IN EITHER THE SHORT OR LONG RANGE TIME FRAME. IN THE SHORT TO MEDIUM RANGE TIME FRAME. EACH OF THESE PROCEDURES IS DESCRIBED IN THIS SECTION. 6.4.1 MTCFCAST - REGIONAL TRAVEL DEMAND FORECASTING PACKAGE MTCFCAST IS A PACKAGE OF PROGRAMS, BASED ON ALL THE TRAVEL DEMAND MODELS SHOWN IN FIGURE 3-2, WHICH PREDICTS REGIONAL TRAVEL PATTERNS AND VOLUMES FROM REGIONAL SOCIOECONOMIC INFORMATION AND THE LEVEL OF SERVICE DATA FOR EXISTING AND PROPOSED MODES OF TRANSPORTATION. THIS COMPUTER SYSTEM REPRESENTS AN ALTERNATIVE TO THE DEMAND ESTIMATION PORTIONS OF THE TRADITIONAL URBAN TRANSPORTATION PLANNING PACKAGES (FHWA, 1974; UMTA, 1976), AND IS INTEGRATED WITH THEM IN TERMS OF ITS EXTERNAL DATA STRUCTURE AND ITS USE OF THEIR DATA PROCESSING, REPORT FORMATTING, AND TRAFFIC ASSIGNMENT PROGRAMS. THIS COMPATIBILITY WITH THE POWERFUL UTPS PACKAGE GREATLY ENHANCES THE EFFECTIVENESS OF MTCFCAST. THE PRIMARY PRODUCT OF MTCFCAST IS A SET OF 24-HOUR PERSON TRIP TABLES: (1) DRIVE ALONE HOME BASED WORK TRIPS (P/A) (2) SHARED-RIDE HOME BASED WORK TRIPS (P/A) (3) TRANSIT HOME BASED WORK TRIPS (P/A) (4) AUTO HOME BASED SHOP TRIPS (P/A) (5) TRANSIT HOME BASED SHOP TRIPS (P/A) (6) AUTO HOME BASED SOCIAL-RECREATIONAL TRIPS (P/A) (7) TRANSIT HOME BASED SOCIAL=RECREATIONAL TRIPS (P/A) (8) AUTO NON-HOME BASED TRIPS (O/D) (9) TRANSIT NON-HOME BASED TRIPS (O/D) (10) AUTO HOME BASED TRIPS (O/D) (11) TRANSIT HOME BASED TRIPS (O/D) THE ORIGIN/DESTINATION (O/D) TRIP TABLES MAY BE INPUT DIRECTLY TO THE UTPS NETWORK ASSIGNMENT ROUTINES (UROAO AND UNET) TO PRODUCE 24-HOUR PERSON HIGHWAY AND TRANSIT TRIP ASSIGNMENTS. THE PRODUCTION/ATTRACTION (P/A) TRIP TABLES MAY BE INPUT TO TRIP PEAKING AND AUTO OCCUPANCY FACTORING (AVAILABLE IN UTPS) ROUTINES WHICH PRODUCE PEAK PERIOD AND VEHICLE TRIP TABLES WHICH THEN MAY BE INJECTED BACK INTO THE MTCFCAST SYSTEM FOR CONVERSION FROM P/A TO O/D FORMATS. THE RESULTING PEAK PERIOD AND VEHICLE O/D TRIP TABLES CAN THEN BE INPUT TO THE UTPS ASSIGNMENT ROUTINES. THIS COMPOSITE MTCFCAST/UTPS ANALYSIS FRAMEWORK IS DIAGRAMMED IN FIGURE 6-2. THE MTCFCAST FORECASTING SYSTEM IS SIMILAR TO, BUT MORE SOPHISTICATED THAN, THE CONVENTIONAL TRIP GENERATION-TRIP DISTRIBUTION-MODAL SPLIT METHODOLOGY. IT INCORPORATES EACH OF THE TWENTY-ONE MODELS DESCRIBED IN SECTION 9. THESE MODELS ARE OVERVIEW OF THE COMPONENT MODEL DEVELOPMENT PROCESS 6-7 MTCFCAST'S OVERALL LOGICAL STRUCTURE, ALONG WITH ITS MAJOR INPUTS AND OUTPUTS, ARE ALSO SHOWN IN FIGURE 6-3, IN WHICH EACH OF THE BLOCKS IDENTIFIED IN FIGURE 3-2 IS SHOWN AS ONE OR TWO BOXES. REFLECTING THE LOGICAL STRUCTURE OF THE MODEL SYSTEM, MTCFCAST USES SOCIO-ECONOMIC AND LEVEL-OF-SERVICE INPUTS TO PREDICT, FIRST OF ALL, THE NUMBER OF WORKERS PER HOUSEHOLD USING THE NWHH MODEL (SECTION 9.2). THIS ALLOWS TOTAL HOUSEHOLDS TO BE SEGMENTED INTO TWO GROUPS: HOUSEHOLD WITH AND WITHOUT WORKERS. MOBILITY IS NEXT PREDICTED FOR EACH OF THESE GROUPS SEPARATELY. FOR HOUSEHOLDS WITH WORKERS, THE MOBILITY DECISIONS INCLUDE BOTH THE NUMBER, DESTINATION, AND MODE FOR WORK TRIPS (PREDICTED USING THE HBW MODELS PRESENTED IN SECTION 9.5-9.12; AND THE NUMBER OF AUTOS OWNED (PREDICTED USING THE HBW MODEL IN SECTION 5.4). FOR HOUSEHOLDS WITHOUT WORKERS, ONLY AUTO OWNERSHIP MUST BE PREDICTED. THIS IS DONE USING THE NWHHAO MODEL (SECTION 9.3). THESE TWO HOUSEHOLD GROUPS ARE NEXT COMBINED, AND THEIR HOME BASED NON-WORK TRAVEL (SECTIONS 9.12-9.15), AND NON-HOME BASED TRAVEL (SECTIONS 9.13- 917), AND NON-HOME BASED TRAVEL (SECTIONS 9.18-9.21) IS PREDICTED. THE RESULTS ARE AVAILABLE FOR PEAK HOUR FACTORING AND FOR BOTH HIGHWAY AND TRANSIT NETWORK ASSIGNMENTS. THROUGHOUT THE MTCFCAST SYSTEM, AGGREGATION IS PERFORMED BY MARKET SEGMENTATION, USING AVERAGE SOCIOECONOMIC VALUES FOR EACH OF THREE INCOME GROUPS INITIALLY, FOLLOWED BY SEGMENTATION BASED ON AUTO OWNERSHIP LEVEL AFTER THE PREDICTION OF AUTO OWNERSHIP IN THE WHHAO AND NWHHAO MODELS. DUE TO THE COMPLEXITIES OF THE USE OF ACCESSIBILITY-TYPE (LOGSUM) VARIABLES, MARKET SEGMENTATION, AND THE NEED TO CREATE MODEL VARIABLES AS SPECIAL TRANSFORMATIONS OF THE INPUT DATA ITEMS, THE ACTUAL SEQUENCE OF PROCESSING STEPS IN MTCFCAST DIFFERS FROM THE LOGICAL SEQUENCE DESCRIBED ABOVE. THIS SEQUENCE, NOT IMPORTANT TO AN UNDERSTANDING OF THE MODELS PRESENTED IN THIS SUMMARY REPORT, IS DESCRIBED IN DETAIL IN VOLUME 3 OF THIS REPORT. OVERVIEW OF THE COMPONENT MODEL DEVELOPMENT PROCESS 6-8 Click HERE for graphic. OVERVIEW OF THE COMPONENT MODEL DEVELOPMENT PROCESS 6-9 Click HERE for graphic. OVERVIEW OF THE COMPONENT MODEL DEVELOPMENT PROCESS 6-10 Click HERE for graphic. OVERVIEW OF THE COMPONENT MODEL DEVELOPMENT PROCESS 6-11 6.4.2 SRGP - SHORT-RANGE GENERALIZED POLICY ANALYSIS SRGP IS A COMPUTERIZED PROCEDURE THAT APPLIES A SUBSET OF THE MTC TRAVEL DEMAND MODELS FOR SHORT RANGE GENERALIZED POLICY ANALYSIS. (SHORT RANGE CAN BE DEFINED AS LESS THAN FIVE YEARS.) THE PROCEDURE IS DESIGNED TO PRODUCE RAPID TURNAROUND ESTIMATES OF THE CONSEQUENCES OF BROADLY DEFINED TRANSPORTATION POLICY OPTIONS. SRGP PROCESSING AND OUTPUTS ARE BASED ON AN INPUT SAMPLE OF HOME INTERVIEW SURVEY HOUSEHOLDS. THE PROGRAM ESTIMATES THE TRAVEL BEHAVIOR OF EACH INDIVIDUAL HOUSEHOLD, SUBJECT TO USER-CONTROLLED FACILITIES FOR EXPANDING THE RESULTS IN WHATEVER MANNER IS APPROPRIATE FOR THE PROBLEM UNIVERSE. THIS APPROACH TAKES FULL ADVANTAGE OF THE DESEGREGATE NATURE OF THE DEMAND MODELS. AGGREGATION DOES NOT TAKE PLACE UNTIL THE EXPANSION, AFTER ALL ESTIMATION IS COMPLETE, AND CAN BE STRAIGHTFORWARD AND WITHOUT BIAS. THE USE OF SAMPLE HOUSEHOLDS ALSO LENDS SRGP ITS SHORT RANGE APPLICABILITY. THE PROCEDURE HAS NO FACILITIES FOR MODELLING THE LONG RANGE DYNAMICS OF ATTRIBUTES OF HOUSEHOLDS SUCH AS LOCATION AND LIFE CYCLE PROGRESSION, OR THE NUMBER OF WORKERS AND THEIR CHOICE OF OCCUPATIONS AND JOB LOCATIONS. ACTIVITY DISTRIBUTIONS (THE ZONAL EXTENT AND INTENSITY OF EMPLOYMENT, SHOPPING FACILITIES, SOCIAL AND RECREATIONAL OPPORTUNITIES, ETC.) ARE ALSO PROVIDED EXOGENOUSLY TO THE PROCEDURE. IT WAS PREVIOUSLY APPLIED TO OTHER URBAN AREAS BY ATHERTON ET AL (1976) AND BEN-AKIVA AND ATHERTON (1977). DUE TO SRGP'S ORIENTATION, ONLY THE MODELS REPRESENTING SHORT RANGE CHOICES ARE INCLUDED (SEE FIGURE 6-4): (1) AUTO OWNERSHIP FOR WORKING HOUSEHOLDS -WHHAO (2) AUTO OWNERSHIP FOR NON-WORKING HOUSEHOLDS -NWHHAO (3) HOME BASED WORK MODE CHOICE - PHBWB AND SHBWM (4) HOME BASED SHOP TRIP PRODUCTION - HBSHG (5) HOME BASED SOCIAL-RECREATIONAL TRIP PRODUCTION - HBSRG (6) HOME BASED SHOP DESTINATION/MODE CHOICE - HBSHDM (7) HOME BASED SOCIAL/RECREATIONAL DESTINATION/MODE CHOICE - HBSHDM (8) HOME BASED WORK SHARED-RIDE VEHICLE OCCUPANCY CHOICE SROCC THE MANNER IN WHICH THE MODELS ARE APPLIED BY SRGP IS OUTLINED BRIEFLY BELOW. WHILE THE USER HAS THE OPTION OF SUPPRESSING THE EXECUTION OF THE AUTO OWNERSHIP MODELS AND ANY OF THE TRIP PURPOSE MODELS, THIS DISCUSSION ASSUMES A FULL APPLICATION OF ALL MODELS. srgp TREATS EACH HOUSEHOLD AS, IN EFFECT, THREE HOUSEHOLDS THAT ARE IDENTICAL EXCEPT FOR THEIR AUTO OWNERSHIP LEVELS, WEIGHTED BY THE PROBABILITIES PREDICTED BY THE APPROPRIATE AUTO OWNERSHIP MODEL. THE PROCEDURE REPEATS THIS MODEL SEQUENCE FOR EACH HOUSEHOLD: (1) EACH HOUSEHOLD WORKER'S HBW TRIP MODE CHOICE PROBABILITIES ARE ESTIMATED AT EACH HOUSEHOLD AUTO OWNERSHIP LEVEL, CONDITIONED ON HIS/HER OBSERVED WORK TRIP DESTINATION. OVERVIEW OF THE COMPONENT MODEL DEVELOPMENT PROCESS 6-12 (2) HBW DRIVE ALONE PROBABILITIES AND SHARED RIDE PROBABILITIES MULTIPLIED BY ESTIMATED OCCUPANCIES ARE SUMMED OVER ALL THE HOUSEHOLD'S WORK TRIPS, AND THE TOTALS ARE USED IN TURN TO ESTIMATE AUTOS REMAINING AFTER WORK TRIPS AT EACH HOUSEHOLD AUTO OWNERSHIP LEVEL. (3) MODE/DESTINATION CHOICE AND CONDITIONAL TRIP PRODUCTION ARE ESTIMATED AT EACH AUTO OWNERSHIP LEVEL AND ITS ASSOCIATED "AUTOS REMAINING" FOR HBSR TRAVEL. SIMILARLY, BUT WITHOUT CONSIDERATION OF AUTOS REMAINING, THE HBSH MODELS ARE EXECUTED FOR DESTINATION, MODE CHOICE, AND TRIP FREQUENCY AT EACH AUTO OWNERSHIP LEVEL. (4) FROM THE MODE, AUTO OWNERSHIP, AND PURPOSE-SPECIFIC UTILITY FUNCTIONS THAT ARE EVALUATED DURING THE ABOVE CALCULATIONS, PROBABILITIES OF EACH OF THE THREE HOUSEHOLD AUTO OWNERSHIP LEVEL ALTERNATIVES ARE ESTIMATED. (5) THE VARIOUS TRAVEL VOLUMES PREVIOUSLY ESTIMATED AT EACH LEVEL OF AUTO OWNERSHIP ARE THEN COMBINED IN PROPORTION TO THEIR RESPECTIVE PROBABILITIES AND FINALLY EXPANDED. SUMMARY IMPACTS ARE ACCUMULATED AND REPORTED FOR HOUSEHOLD INCOME CLASS GROUPS OR OTHER SEGMENTATION. SRGP ALSO HAS THE CAPABILITY TO RETRIEVE THE RESULTS OF A PREVIOUS RUN TO PRODUCE COMPARISON TABLES. THE SRGP PACKAGE IS DESCRIBED IN GREATER DETAIL IN A PROGRAM DOCUMENTATION REPORT (CAMBRIDGE SYSTEMATICS, 1976). OVERVIEW OF THE COMPONENT MODEL DEVELOPMENT PROCESS 6-13 Click HERE for graphic. OVERVIEW OF THE COMPONENT MODEL DEVELOPMENT PROCESS 6-14 Click HERE for graphic. OVERVIEW OF THE COMPONENT MODEL DEVELOPMENT PROCESS 6-15 6.5 AGGREGATE VALIDATION THE MTCFCAST COMPUTER PACKAGE WAS APPLIED TO BASE YEAR (1965) LAND- USE, SOCIO-ECONOMIC, AND LEVEL-OF-SERVICE DATA TO VALIDATE THE TRAVEL DEMAND MODELS AT THE AGGREGATE LEVEL. THE ENTIRE MODEL SYSTEM WAS APPLIED IN SEQUENCE, WITH PAUSES FOLLOWING EACH MODEL FOR COMPARISON WITH OBSERVED DATA. AT EACH OF THESE PAUSES, THE MODELS WERE ADJUSTED, IF NECESSARY, TO MATCH DISTRICT OR REGIONAL OBSERVED DATA BEFORE CONTINUING WITH SUBSEQUENT MODEL APPLICATION STEPS. A COMPARISON OF PREDICTED AND OBSERVED DATA WAS PERFORMED AT THE REGIONAL LEVEL FOR ALL MODELS THAT PREDICT NON-TRAVEL (BUT TRAVEL RELATED) VARIABLES. INCLUDED IN THIS GROUP ARE THE FOLLOWING MODELS: (1) NON-WORKING HOUSEHOLDS (NWHH) (2) NON-WORKING HOUSEHOLD AUTO OWNERSHIP (NWHHAO) (3) WORKING HOUSEHOLD AUTO OWNERSHIP (WHHAO) REGIONAL AVERAGE WERE ALSO USED TO VALIDATE THE SHARED RIDE OCCUPANCY MODEL (SROCC) AND THE TRIP END COMPONENTS OF THE NHB MODELS (NHBAD, NHBAO, NHBTD, NHBTO). ALL OTHER MODELS WERE VALIDATED BY COMPARING PREDICTED AND OBSERVED DATA AT THE DISTRICT AND DISTRICT PAIR LEVEL, WITH ADJUSTMENTS MADE TO THE MODELS TO ENSURE THAT THEY MATCHED THE FOLLOWING OBSERVED DATA: PERSON TRIPS PRODUCED BY PURPOSE AND DISTRICT (HBW, HBSH, AND HBSR). PERSON TRIPS ATTRACTED BY PURPOSE AND DISTRICT (HBW). AVERAGE PERSON TRIP LENGTH (IN MILES) BY PURPOSE AND DISTRICT OF PRODUCTION (ALL PURPOSES). TRIPS PRODUCED BY MODE, PURPOSE, AND DISTRICT (HBW, HBSH, AND HBSR). TRIP INTERCHANGES BY MODE, PURPOSE, AND DISTRICT PAIR (ALL PURPOSES). MODEL ADJUSTMENTS OF A NUMBER OF TYPES WERE MADE TO ENSURE THAT EACH OF THESE BASES OF COMPARISON WERE SATISFIED. THESE ADJUSTMENTS WERE REQUIRED FOR A NUMBER OF REASONS, ALL OF WHICH EXIST TO VARYING DEGREES FOR EACH MODEL INCLUDED IN THE SYSTEM: (1) DIFFERENCES IN ESTIMATION, DESEGREGATE VALIDATION, AND AGGREGATE VALIDATION DATA SETS NOT ADEQUATELY REFLECTED IN THE MODEL SPECIFICATIONS. AS SUBSAMPLES DRAWN FROM THE LARGER AGGREGATE VALIDATION DATA SET, THE ESTIMATION AND DESEGREGATE VALIDATION HOUSEHOLD DATA SETS CAN BE EXPECTED TO DIFFER RANDOMLY FROM EACH OTHER AND FROM THE AGGREGATION OF THE LARGER SAMPLE (THE ENTIRE HOME INTERVIEW SURVEY) INTO TRIP TABLES BY PURPOSE AND MODE. SIMILARLY, THE ZONAL-LEVEL SOCIO-ECONOMIC DATA USED FOR AGGREGATE VALIDATION DIFFERED FROM THE DESEGREGATE SUBSAMPLES, EVEN FOR THE ZONES INCLUDED OVERVIEW OF THE COMPONENT MODEL DEVELOPMENT PROCESS 6-16 IN THE SUBSAMPLES, BECAUSE IT WAS ABSTAINED FROM A DIFFERENT SOURCE - THE "BACKCAST" OF CENSUS DATA TO 1965 - RATHER THAN THE AGGREGATION OF SURVEY DATA. (2) INHERENT DIFFERENCES IN MODEL PREDICTIONS WHEN THEY ARE APPLIED TO AGGREGATIONS OF HOUSEHOLD (AGGREGATED BY ORIGIN ZONE AND EITHER INCOME LEVEL OR AUTO OWNERSHIP) RATHER THAN TO INDIVIDUAL HOUSEHOLDS. (3) THE CHAINING OF THE MODELS IN THE AGGREGATE VALIDATION PROCESS, RESULTING IN THE USE OF THE OUTPUTS OF A HIGHER LEVEL MODEL AS INPUTS TO LOWER LEVEL MODELS. THIS DIFFERS FROM THE DESEGREGATE VALIDATION PROCESS, IN WHICH ONLY OBSERVED INPUTS ARE USED TO OBTAIN MODEL-PREDICTED OUTPUTS THAT CAN BE COMPARED WITH OBSERVED VALUES. FOR EACH OF THESE REASONS, IT WAS NECESSARY TO DEVELOP TWO SETS OF FINAL MODEL FORMULATIONS: ONE FOR DESEGREGATE APPLICATIONS OF THE MODELS, AND ONE FOR AGGREGATE APPLICATIONS. GENERALLY, THE AGGREGATE FORMULATIONS IN THE MTC MODEL SYSTEM DIFFER FROM THE DESEGREGATE FORMULATIONS IN THAT THEY INCLUDE BOTH DIFFERENT VALUES OF SOME ADJUSTMENTS ALSO FOUND TO BE REQUIRED DURING DESEGREGATE VALIDATION, AND ADDITIONAL ADJUSTMENTS, MANY DEVELOPED ON A DISTRICT-SPECIFIC BASIS RATHER THAN ON A REGIONAL BASIS. SPECIFICALLY, THE FOLLOWING ADJUSTMENT FACTORS WERE ADDED TO THE MODELS AS A RESULT OF THE AGGREGATE VALIDATION PROCESS: (1) FACTORS TO ADJUST TO OBSERVED REGIONAL MEAN OR TOTAL VALUES: +-----+--------------+---------------------------------------+ MODEL FACTOR TO ADJUST TO REGIONAL: +-----+--------------+---------------------------------------+ SROCC $SROCCF MEAN SHARED-RIDE VEHICLE OCCUPANCY PHBWG $PHBWGFR TOTAL PRIMARY HBW TRIP PRODUCTIONS PHBWA $PHBWAFR TOTAL PRIMARY HBW TRIP ATTRACTIONS SHBWG $SHBWGFR TOTAL SECONDARY HBW TRIP PRODUCTIONS SHBWA $SHBWAFR TOTAL SECONDARY HBW TRIP ATTRACTIONS HBSHG $HBSHGFR TOTAL HBSH TRIP PRODUCTIONS HBSRG $HBSRGFR TOTAL HBSR TRIP PRODUCTIONS +-----+--------------+---------------------------------------+ OVERVIEW OF THE COMPONENT MODEL DEVELOPMENT PROCESS 6-17 (2) SCALE FACTORS TO ADJUST TO OBSERVED DISTRICT-LEVEL TOTAL VALUES: +-----+--------------+---------------------------------------+ MODEL FACTOR TO ADJUST TO DISTRICT-LEVEL TOTAL: +-----+--------------+---------------------------------------+ PHBWG $PHBWGF PRIMARY HBW TRIP PRODUCTIONS PHBWA, $HBWAF PRIMARY & SECONDARY HBW SHBWA TRIP ATTRACTIONS SHBWG $SHBWGF SECONDARY HBW TRIP PRODUCTIONS HBSHG $HBSHGF HBSH TRIP PRODUCTIONS HBSRG $HBSRGF HBSR TRIP PRODUCTIONS +-----+--------------+---------------------------------------+ (3) ALTERNATIVE-SPECIFIC CONSTANTS TO ADJUST TO OBSERVED REGIONAL SHARES IN LOGIT MODELS: +-----+--------------+---------------------------------------+ MODEL CONSTANT ALTERNATIVE +-----+--------------+---------------------------------------+ NWHHAO C HOUSEHOLDS WITH 1+ WORKERS NWHHAO C(1) HOUSEHOLDS WITH 1 AUTO C(2) HOUSEHOLDS WITH 2+ AUTOS WHHAO C(1) HOUSEHOLDS WITH 1 AUTO C(2) HOUSEHOLDS WITH 2+ AUTOS PHBWM C(1) DRIVE-ALONE MODE C(2) SHARED-RIDE MODE SHBWM C(1) DRIVE-ALONE MODE C(2) SHARED-RIDE MODE HBSHDM C AUTO MODE HBSRDM C AUTO MODE NHBAD C ZERO FREQUENCY NHBAD C ZERO FREQUENCY NHBTD C ZERO FREQUENCY NHBTD C ZERO FREQUENCY +-----+--------------+---------------------------------------+ (4) ALTERNATIVE-SPECIFIC CONSTANTS TO ADJUST TO OBSERVED PRODUCTION DISTRICT-LEVEL SHARES IN MODE CHOICE MODELS: +-----+--------------+---------------------------------------+ MODEL FACTOR TO ADJUST TO REGIONAL: +-----+--------------+---------------------------------------+ PHBWM $HBWMAMCF DRIVE-ALONE MODE SHBWM $HBWMSMCF SHARED-RICE MODE HBSHDM $HBSHMCF AUTO MODE HBSRDM $HBSRMCF AUTO MODE +-----+--------------+---------------------------------------+ OVERVIEW OF THE COMPONENT MODEL DEVELOPMENT PROCESS 6-18 (5) COEFFICIENTS TO ADJUST TO OBSERVED PRODUCTION DISTRICT-LEVEL AVERAGE TRIP LENGTHS IN DESTINATION CHOICE MODELS: +---------+-------------+ MODEL COEFFICIENT +---------+-------------+ PHBWD $PHBWDC SHBWD $SHBWDC HBSHDM $HBSHDC HBSRDM, $HBSRDC NHBTD, NHBTO +--------+-------------+ (6) COEFFICIENTS TO ADJUST TO OBSERVED OBO TRIP PRODUCTIONS OR ATTRACTIONS: +-----------+--------------+------------+--------------------+ MODEL VARIABLE MODE COEFFICIENT +-----------+--------------+------------+--------------------+ HBSHG LN(RETSERD) PERSON B(4) HBSHDM CBD AUTO B(4) CBD TRANSIT B(7) HBSRDM CBD AUTO B(6) CBD TRANSIT B(10) NHBAD CBD AUTO B(1) NHBAD CBD AUTO B(1) NHBTD CBD TRANSIT B(6) NHBTO CBD TRANSIT B(1) +-----------+--------------+------------+--------------------+ (7) FACTORS TO ADJUST TO OBSERVED DISTRICT TO DISTRICT INTERCHANGES BY PURPOSE AND MODE: +----------------+----------------------+--------------------+ TRIP PURPOSE MODE FACTOR +----------------+----------------------+--------------------+ HBW DRIVE ALONE $HBWPAF SHARED-RIDE $HBWFSF TRANSIT $HBWPTF HBSH AUTO $HBSHMAF TRANSIT $HBSHMTF HBSR AUTO $HBSRMAF TRANSIT $HBSRMTF NHB AUTO $NHBFAF TRANSIT $NHBMTF +----------------+----------------------+--------------------+ ALTHOUGH THIS LAST SET OF FACTORS IS AVAILABLE FOR APPLICATION TO THE RESULTS OF THE MODELS AS DOCUMENTED INT HIS FINAL REPORT, THEY ARE NOT EXPLICITLY INCORPORATED INTO THE MATHEMATICAL EXPRESSION OF THE MODELS THIS IS BECAUSE, AS TRIP TABLE (RATHER THAN MODEL) ADJUSTMENT FACTORS, THEY APPLY TO THE COMBINED RESULTS OF COMPUTING ENTIRE CHAINS OF MODELS. TRIPS REPRESENTED IN THE MODELS 7-1 7. TRIPS REPRESENTED IN THE MODELS ALL TRIPS USED IN MODEL DEVELOPMENT REPRESENT WEEKDAY TRAVEL, AS REPORTED BY SURVEY RESPONDENTS IN THE 1965 MTC HOME INTERVIEW SURVEY. TWO CLASSIFICATIONS OF THESE TRIPS - BY PURPOSE AND BY MODE - WERE BASIC TO THE MODEL DEVELOPMENT PROCESS. THE TRIP PURPOSES USED IN MODEL DEVELOPMENT ARE THE FOLLOWING: (1) HOME BASED WORK TRIPS (HBW) - ALL TRIPS BETWEEN HOME AND WORK ARE INCLUDED. (2) HOME BASED SHOPPING TRIPS (HBSH) - SHOPPING IS USED IN A GENERALIZED SENSE, TO INCLUDE ALSO MEDICAL-DENTAL, BUSINESS- RELATED, AND PASSENGER PURPOSES. ALL TRIPS BETWEEN HOME AND ANY OF THESE PURPOSES ARE INCLUDED. (3) HOME BASED SOCIAL-RECREATION TRIPS INCLUDE EATING OUT, VISITING AND RECREATION. ALL TRIPS BETWEEN HOME AND ANY OF THESE PURPOSES ARE INCLUDED. (4) NON-HOME BASED (NHB) - ALL TRIPS THAT DO NOT BEGIN OR END AT HOME OR AT SCHOOL ARE INCLUDED. THESE FOUR PURPOSE GROUPS INCLUDE ALL SURVEYED TRIPS EXCEPT SCHOOL TRIPS. THE MODELS CONSIDERED IN THE MODELS INCLUDE AUTO AND PICKUP DRIVERS AND PASSENGERS, AS WELL AS ALL BUS, STREETCAR, RAILROAD, AND JITNEY TRIPS. THESE MODES OF TRAVEL ARE AGGREGATED TO THREE MODAL ALTERNATIVES FOR WORK TRIPS: DRIVE ALONE, SHARED RIDE, AND TRANSIT. FOR ALL OTHER TRIPS, TWO MODAL ALTERNATIVES ARE USED: AUTO (BOTH DRIVERS AND PASSENGERS) AND TRANSIT. TRIPS BY THE FOLLOWING MODES WERE NOT USED IN MODEL DEVELOPMENT: TAXI DRIVER AND PASSENGER, TRUCK DRIVER AND PASSENGER, WALK, AND OTHER. THE COMPONENT MODELS-OVERVIEW 8-1 8. THE COMPONENT MODELS - OVERVIEW THE NEXT SECTION CONTAINS A GENERAL DESCRIPTION OF THE FUNCTION OF EACH OF THE 21 MODELS IN THE MTC TRAVEL DEMAND MODEL SYSTEM, AND MATHEMATICAL STATEMENTS OF THE FINAL MODEL, IN BOTH THEIR DESEGREGATE AND AGGREGATE FORMS. THESE TWO FORMS CORRESPOND TO THE MODELS IMPLEMENTED IN SRGP AND MTCFCAST, RESPECTIVELY (SEE 6.4). THESE MODEL DESCRIPTIONS ARE SUMMARIES OF THE COMPLETE INFORMATION CONTAINED IN PART II OF VOLUME II OF THIS REPORT, WHICH SHOULD BE REFERRED TO FOR FURTHER DETAILS ON THE MODELS AND ON THEIR DEVELOPMENT PROCESS. READERS CAN OBTAIN ONLY A VERBAL DESCRIPTION OF EACH MODEL BY READING JUST THE FIRST SUBSECTION, LABELED GENERAL DESCRIPTION, FOR THE MODELS. THE SECOND AND THIRD SUBSECTIONS PROVIDE COMPLETE MATHEMATICAL DESCRIPTIONS. AS EACH MODEL DESCRIPTION IS READ, THE READER WILL FIND IT USEFUL TO REFER BACK TO FIGURE 3-2, 6-3, AND 6-4 TO HELP IDENTIFY THE ROLE OF THE COMPONENT MODEL WITHIN THE COMPLETE MODEL SYSTEM. BEFORE BEGINNING THE MODEL DESCRIPTIONS THEMSELVES, THE NOTATIONAL AND VARIABLE NAMING CONVENTIONS USED IN THESE DESCRIPTIONS ARE EXPLAINED. 8.1 MODEL NAMING CONVENTIONS THE COMPONENT MODELS, THEIR MNEMONIC NAMES AND THEIR ORDER IN SECTION 9 (WITH MODEL NUMBER EQUAL TO SUBSECTION NUMBER) ARE LISTED IN TABLE 8-1. ALSO SHOWN IS THE MATHEMATICAL FORM AND THE METHOD OF STRATIFICATION USED WHEN THE MODEL IS APPLIED IN MTCFCAST. THE COMPONENT MODELS-OVERVIEW 8-2 -------+----------------+-------------+------------+---------+ MTCFCAST NO. MODEL NAME MNEMONIC MODEL FORM STRATIFI- CATION -------+----------------+-------------+------------+---------+ 1 HOME BASED WORK TRIP SROCC LINEAR NONE SHARED-RIDE OCCUPANCY MODEL 2 NON-WORKING HOUSEHOLD NWHH LOGIT INCOME MODEL 3 NON-WORKING HOUSEHOLD NWHHAO LOGIT INCOME AUTO OWNERSHIP MODEL 4 WORKING HOUSEHOLD WHHAO LOGIT INCOME AUTO OWNERSHIP MODEL 5 PRIMARY WORKER HOME PHBWG LINEAR INCOME BASED WORK TRIP GENERATION MODEL 6 PRIMARY WORKER HOME PHBWA LINEAR NONE BASED WORK TRIP ATTRACTION MODEL 7 PRIMARY WORKER HOME PHBWO LOGIT INCOME BASED WORK TRIP DEST. CHOICE MODEL 8 PRIMARY WORKER HOME PHBWM LOGIT AO1 BASED WORK TRIP MODE CHOICE MODEL 9 SECONDARY WORKER HOME SHBWG LINEAR AO BASED WORK TRIP GENERATION MODEL 10 SECONDARY WORKER HOME SHBWA LINEAR NONE BASED WORK TRIP ATTRACTION MODEL 11 SECONDARY WORKER HOME SHBWO LOGIT AO BASED WORK TRIP DEST. CHOICE MODEL 12 SECONDARY WORKER HOME SHBWM LINEAR AO BASED WORK TRIP MODE CHOICE MODEL 13 HOME BASED SHOP TRIP HBSHG NON-LINEAR AO GENERATION MODEL 14 HOME BASED SHOP TRIP HBSHO LOGIT AO=0 DEST. CHOICE MODEL ONLY 15 HOME BASED SHOP TRIP HBSHOM LOGIT AC DESTINATION/MODE CHOICE MODEL 16 HOME BASED SOCIAL/ HBSRG NON-LINEAR AO RECREATIONAL TRIP GENERATION MODEL -------+----------------+-------------+------------+---------+ ___________________________ (1) AO = AUTO OWNERSHIP TABLE 8-1 (CONTINUED) MODELS IN THE MTC TRAVEL DEMAND MODEL SYSTEM -------+----------------+-------------+------------+---------+ MTCFCAST NO. MODEL NAME MNEMONIC MODEL FORM STRATIFI- CATION -------+----------------+-------------+------------+------------+ 17 HOME BASED SOC/REC HBSROM LOGIT AO1 DESTINATION/MODE CHOICE MODEL 18 NON-HOME BASED AUTO NHBAD LOGIT NONE TRIP DESTINATION CHOICE MODEL 19 NON-HOME BASED AUTO NHBAD LOGIT NONE TRIP ORIGIN CHOICE MODEL 20 NON-HOME BASED TRANSIT NHBTD LOGIT NONE TRIP DESTINATION CHOICE MODEL 21 NON-HOME BASED TRANSIT NHBTO LOGIT NONE TRIP ORIGIN CHOICE MODEL +------+----------------+-------------+------------+------------+ ___________________________ (1) AO = AUTO OWNERSHIP TABLE 8-1 MODELS IN THE MTC TRAVEL DEMAND MODEL SYSTEM THE COMPONENT MODELS-OVERVIEW 8-4 8.2 NOTATIONAL CONVENTIONS (1) MATHEMATICAL OPERATIONS OF ADDITION, SUBTRACTION, MULTIPLICATION, DIVISION AND EXPONENTATION ARE INDICATED BY THE STANDARD SYMBOLS +, -, *, /, AND ** RESPECTIVELY. (2) A NUMBER OF COMMON MATHEMATICAL OPERATIONS ARE EXPRESSED USING ALPHABETIC SHORTHAND: SYMBOL SIGNIFICANCE LN(X) NATURAL LOGARITHM OF THE VARIABLE X (LOGARITHM TO THE BASE E, WHERE E = 2.71828...) EXP(X) EXPONENTATION: E TO THE POWER X (EQUIVALENT TO E**X) SQRT(X) THE SQUARE ROOT OF X (EQUIVALENT TO X**0.5) MIN(X,Y,...) THE SMALLEST NUMERICAL VALUE IN THE SET X, Y,... MAX(X,Y,...) THE LARGEST NUMERICAL VALUE IN THE SET X,Y,... XEY SCIENTIFIC NOTATION FOR NUMBERS: X MULTIPLIED BY (10 RAISED TO THE POWER Y). (EQUIVALENT TO X*(10**Y) (3) THE OPERATION "IS DEFINED BY" IS DENOTED BY THE SYMBOL ::= (4) A SYMBOL ENCLOSED IN BRACKETS, SUCH AS, DENOTES A SITUATION IN WHICH THE GENERIC SYMBOL X MUST BE REPLACED BY SPECIFIC SYMBOLS TO OBTAIN THE SPECIFIC FORMS OF A VARIABLE OR SUBSCRIPT. FOR EXAMPLE, IF THE GENERIC VARIABLE Y IS GIVEN, AND IF THE ALLOWABLE SYMBOLS REPRESENTED BY X ARE 1, 2, AND 3, THEN THE SPECIFIC FORMS OF THE VARIABLE ARE Y1, Y2, AND Y3. NOTE THAT THE BRACKETS APPEAR ONLY IN THE GENERIC VARIABLES, AND NOT IN THE ALTERNATIVE-SPECIFIC VARIABLES. 8.3 VARIABLE NAMING CONVENTIONS THE FOLLOWING VARIABLE NAMES ARE COMMONLY USED IN THE DESCRIPTION OF MANY MODELS: U - A UTILITY WHICH APPEARS IN A LOGIT MODEL; SEE 4.1 FOR A GENERAL DISCUSSION OF THIS MODEL FORM. P - A PROBABILITY; IN THIS REPORT ALWAYS THE DEPENDENT VARIABLE OF A LOGIT MODEL; SEE 4.1. B - AN ESTIMATED COEFFICIENT OF A LOGIT, LINEAR OR NON-LINEAR MODEL. IN THE CASE OF THE LINEAR AND LOGIT MODELS, THESE B'S ARE MULTIPLICATIVE COEFFICIENTS OF THE COMPONENT MODELS - OVERVIEW 8-5 THEIR RELATED VARIABLES; IN THE CASE OF NON-LINEAR MODELS A NUMBER OF RELATIONSHIPS BETWEEN COEFFICIENT AND VARIABLE ARE POSSIBLE. C - AN ESTIMATED CONSTANT IN A LINEAR OR LOGIT MODEL. THESE QUANTITIES ARE ADDED DIRECTLY INTO LOGIT UTILITIES AND LINEAR DEPENDENT VARIABLES; THEY ARE NOT MULTIPLIED BY OTHER VARIABLES. # - WHEN THIS SYMBOL PRECEDES AN ALPHABETIC VARIABLE NAME, IT DENOTES A DISAGGREGATE VARIABLE, BASED ON HOUSEHOLD OR TRIPMAKER INFORMATION. $ - WHEN THIS SYMBOL PRECEDES AN ALPHABETIC VARIABLE NAME, IT DENOTES AN UNESTIMATED MODEL PARAMETER; A BOUND, A REGIONAL MEAN OR A ADJUSTMENT FACTOR. WHEN NO SYMBOL PRECEDES AN ALPHABETIC VARIABLE NAME, THE VARIABLE IS AN AGGREGATE ONE, BASED ON ZONAL OR ZONE-TO-ZONE DATA. 8.4 SUBSCRIPT NAMING CONVENTIONS THE FOLLOWING SUBSCRIPT CODES ARE USED CONSISTENTLY THROUGHOUT SECTION 9: A - DENOTED THE DRIVE ALONE MODE IN THE HBW MODELS, AND THE AUTO (DRIVER AND PASSENGER) MODE IN ALL THE OTHER MODELS. - DENOTES AUTO OWNERSHIP LEVEL, WHICH CAN BE THE SYMBOL O, 1, OR 2 (TWO OR MORE). - DENOTES INCOME LEVEL, WHICH CAN BE THE SYMBOL 1, 2, OR 3, CORRESPONDING TO GROUPS WITH LOW, MEDIUM AND HIGH AVERAGE HOUSEHOLD INCOMES. THE ANNUAL HOUSEHOLD INCOMES DIVIDING THESE GROUPS CAN BE SET BY THE USER OF THE MODELS; NORMALLY THE TOTAL NUMBER OF HOUSEHOLDS IN EACH GROUP SHOULD BE APPROXIMATELY EQUAL. H - DENOTES A SPECIFIC HOUSEHOLD. I - DENOTES A SPECIFIC PRODUCTION OR ORIGIN ZONE. J - DENOTES A SPECIFIC ATTRACTION OR DESTINATION ZONE. THE COMPONENT MODELS - OVERVIEW 8-6 - DENOTES MODE, WHICH MAY BE ONE OF THE FOLLOWING SYMBOLS: MODEL TYPE SYMBOL HBW OTHER A DRIVE ALONE AUTO S SHARED-RIDE T TRANSIT TRANSIT G - DENOTES A TRIP OR A TRIPMAKER, DEPENDING ON THE MODEL. SROCC 9.1-1 9. MODEL DESCRIPTIONS 9.1. HOME BASED WORK TRIP SHARED-RIDE OCCUPANCY MODEL (SROCC) 9.1.1. GENERAL DESCRIPTION IN ITS DISAGGREGATE FORM THE SROCC MODEL PREDICTS THE AUTO OCCUPANCY (PERSONS/VEHICLE) FOR A SHARED-RIDE HOME BASED WORK TRIP. IN ITS AGGREGATE FORM IT PREDICTS THE MEAN SHARED-RIDE HOME BASED WORK TRIP AUTO OCCUPANCY BY PRODUCTION ZONE - ATTRACTION ZONE PAIR. IN MTCFCAST, THE INVERSE OF SROCC IS PREDICTED TO FACILITATE MATRIX MULTIPLICATION IN LATER PROCESSING. THE MODEL IS INDEPENDENT OF THE OUTPUTS OF ANY OTHER MODEL. BECAUSE THE AUTO OCCUPANCY OF DRIVE-ALONE PERSON TRIPS IS ONE BY DEFINITION, TOTAL HOME BASED WORK VEHICLE TRIPS CAN BE OBTAINED AS THE SUM OF THE DRIVE-ALONE PERSON TRIOS, AND SHARED-RIDE PERSON TRIPS DIVIDED BY THE SHARED-RIDE OCCUPANCY PROVIDED BY THIS MODEL. THE MODEL IS USED IN THIS WAY, AND ALSO TO COMPUTE PER-PERSON SHARED-RIDE TRAVEL COSTS. SHARED-RIDE AUTO OCCUPANCY IS CALCULATED AS A LINEAR FUNCTION OF TWO VARIABLES: DISPOSABLE INCOME OF THE TRIPMAKER, AND THE SHARED- RIDE IN-VEHICLE TRAVEL TIME FOR THE TRIPMAKER'S WORK TRIP. SHARED- RIDE OCCUPANCY DECREASES WITH INCREASING INCOME AND INCREASES WITH INCREASING TRAVEL TIME. 9.1.2 DISAGGREGATE MODEL FORMULATION THIS LINEAR MODEL HAS THE FOLLOWING EQUATION: #SROCC(Q) :: = MAX(2,(C=B(1) *#DINC(H(Q))+B(2)*STIMET(1(Q),J(Q)))) WHERE: #SROCC(Q) IS THE HOME BASED WORK TRIP SHARED-RIDE AUTO OCCUPANCY PREDICTED FOR TRIPMAKER Q (PERSONS/VEHICLE) #DINC(H(G)) IS THE ANNUAL DISPOSABLE INCOME OF Q'S HOUSEHOLD, H(G) (1965 DOLLARS), COMPUTED AS: #DINC(H(Q)) :: = #INCOME(H(Q)) - $750*#PHH(H(Q)) WHERE: #INCOME(H(Q)) IS THE ANNUAL GROSS INCOME OF Q'S HOUSEHOLD, H(Q) (1965 DOLLARS), COMPUTED. #PHH(H(Q)) IS THE ANNUAL GROSS INCOME OF Q'S HOUSEHOLD, H(Q) (PERSONS). STIMET(I(Q),J(Q)) IS THE 1-WAY PEAK PERIOD SHARED-RIDE IN- VEHICLE TIME FROM ZONE I(Q) TO ZONE J(Q) (MINUTES). C,B(1),B(2) ARE MODEL COEFFICIENTS: SROCC 9.1-2 VARIABLE COEFF VALUE ------- C 2.542 #DINC B(1) -4.7171E - 5 STIMET B(2) .01116 9.1.3 AGGREGATE MODEL FORMULATION SROCC(I,J) :: = MAX(2,$SROCCF*(C+B(1)*DINC(1)+B(2)*STIME(I,J)))1 WHERE: SROCC(I,J) IS THE SHARED-RIDE AUTO OCCUPANCY PREDICTED FOR HOME BASED WORK TRIPS MADE BY THE RESIDENTS OF ZONE I WORKING IN ZONE J (PERSONS/VEHICLE). DINC(I) IS THE MEAN ANNUAL GROSS HOUSEHOLD INCOME IN ZONE I (1965 DOLLARS), COMPUTED AS: DINC(I) :: = INCOME(I)-$750*PHH(I) WHERE: INCOME(I) IS THE MEAN GROSS ANNUAL HOUSEHOLD INCOME IN ZONE I (1965 DOLLARS). PHH(I) IS THE MEAN HOUSEHOLD SIZE IN ZONE I (PERSONS). STIMET(I,J) IS THE 1-WAY PEAK PERIOD SHARED-RIDE IN-VEHICLE TRAVEL TIME FROM ZONE I TO ZONE J (MINUTES). C, B(1), B(2) ARE MODEL COEFFICIENTS: VARIABLE COEFF VALUE -------- C 2.542 DINC B(1) -4.717E - 5 STIMET B(2) .01116 $SROCCF IS A REGIONAL ADJUSTMENT FACTOR: .944 RECOMMENDED FOR SAN FRANCISCO APPLICATIONS; 1.0 RECOMMENDED OTHERWISE. NWHH 9.2-1 9.2. NON-WORKING HOUSEHOLD MODEL (NWHH) 9.2.1 GENERAL DESCRIPTION IN ITS DISAGGREGATE FORM, THE NWHH MODEL PREDICTS THE PROBABILITY OF A HOUSEHOLD HAVING NO WORKERS. IN ITS AGGREGATE FORM, THE NWHH MODEL ESTIMATES THE FRACTION OF A GROUP OF HOUSEHOLDS THAT HAS ZERO WORKERS. IN MTFCAST, THESE HOUSEHOLD GROUPS ARE DEFINED BY RESIDENCE ZONE AND INCOME LEVEL. NON-WORKER AND WORKER HOUSEHOLD VOLUME PREDICTIONS MAY BE OBTAINED BY MULTIPLYING THE MODEL RESULTS BY THE APPROPRIATE BASE HOUSEHOLD NUMBERS; THESE VOLUMES MAY BE USED AS BASES FOR APPLICATION OF THE HOME BASED TRIP FREQUENCY MODELS (PHBWG, SHBEG, HBSHG, HBSRG) AND THE NWHHAD MODEL. THE MODEL HAS THE LOGIT FORM, WITH TWO ALTERNATIVES: ZERO WORKERS ONE OR MORE WORKERS THE INDEPENDENT VARIABLES ARE HOUSEHOLD INCOME, PERSONS PER HOUSEHOLD, AND A SPECIAL LOW INCOME VARIABLE INDICATING THE AMOUNT (IF ANY) BY WHICH HOUSEHOLD INCOME IS LESS THAN $4500. THE PROBABILITY OF A HOUSEHOLD HAVING NO WORKERS DECREASES WITH INCREASING INCOME AND HOUSEHOLD SIZE AND INCREASES AS THE LOW INCOME VARIABLE INCREASES. 9.2.2 DISAGGREGATE MODEL FORMULATION THIS LOGIT MODEL HAS THE FOLLOWING UTILITY EQUATIONS: U(C,H) :: = 0 U(I+,H) :: = C+B(1)*#INCOME(H)+B(2)*#PHH(H)+B(3)*#FINC(H) WHERE: U( ,H) IS THE UTILITY OF HAVING WORKERS IN HOUSEHOLD H. #INCOME(H) IS THE GROSS ANNUAL INCOME FOR HOUSEHOLD H (1965 DOLLARS). #PHH(H) IS THE NUMBER OF PERSONS IN HOUSEHOLD H (PERSONS). #FINC(H) IS COMPUTED AS MAX (0,(4500-#INCOME(H))) (1965 DOLLARS) C,B(1),B(2),B(3) ARE MODEL COEFFICIENTS: VARIABLE COEFF VALUE ------- C .3120 #INCOME B(1) 4.844E - 5 #PHH B(2) .6834 #FINC B(3) -5.814E - 4 NWHH 9.2-2 9.2.2 AGGREGATE MODEL FORMULATION U(O,I,G) :: = O U(1+,I,G) :: = C+B(1)*INCOMEG(I,G)+B(2)*PHHG(I,G)+B(3)*FINC(I,G) WHERE: U( ,I,G) IS THE UTILITY TO HOUSEHOLDS OF INCOME GROUP G RESIDING IN ZONE I OF HAVING WORKERS. INCOMEG(I,G) IS THE MEAN GROSS ANNUAL HOUSEHOLD INCOME OF HOUSEHOLDS IN INCOME GROUP G IN ZONE I (1965 DOLLARS) PHHG(I,G) IS THE AVERAGE HOUSEHOLD SIZE IF HOUSEHOLDS IN INCOME GROUP G IN ZONE I (PERSONS). FINC(I,G) IS COMPUTED AS MAX(O,(4500-INCOMEG(I,G))) (1965 DOLLARS) C,B(1),B(2),B(3) ARE MODEL COEFFICIENTS: VARIABLE COEFF VALUE -------- C -.3400 INCOMEG B(1) 4.844E - 5 PHHG B(2) .6834 FINC B(3) -9.814E - 4 NWHHAO 9.3-1 9.3 NON-WORKING HOUSEHOLD AUTO OWNERSHIP MODEL (NWHHAO) 9.3.1 GENERAL DESCRIPTION IN ITS DISAGGREGATE FORM, THE NWHHAO MODEL PREDICTS THE PROBABILITY OF A HOUSEHOLD WITHOUT WORKERS OWNING O, 1, 2 OR MORE AUTOMOBILES. IN ITS AGGREGATE FORM, IT PREDICTS THE FRACTION OF A GROUP OF HOUSEHOLDS BELONGING TO EACH OF THE THREE AUTO OWNERSHIP LEVELS. IN MTCFCAST, THESE HOUSEHOLD GROUPS ARE DEFINED BY RESIDENCE ZONE AND INCOME LEVEL. THE AGGREGATE FORM OF THE NWHAO MODEL IS DESIGNED TO BE APPLIED TO THE PREDICTED NUMBER OF NON-WORKING HOUSEHOLDS, WHICH CAN BE OBTAINED FROM THE NWHH MODEL. THE MODEL ALSO USES A COMPOSITE ACCESSIBILITY VARIABLE PROVIDED BY THE HBSHO MODEL. THE MODEL RESULTS ARE USED TO PROVIDE HOUSEHOLD TOTALS BY AUTO OWNERSHIP LEVEL AS STRATIFICATIONS FOR USE BY THE NON-WORK TRIP GENERATION MODELS (HBSHG AND HBSRG). THE INDEPENDENT VARIABLES, AND THEIR EFFECTS ON PREDICTED AUTO OWNERSHIP ARE THE FOLLOWING: (1) PER CAPITA HOUSEHOLD INCOME REMAINING AFTER PROVIDING THE BASIC PERSONAL NEEDS OF EACH HOUSEHOLD MEMBER. AUTO OWNERSHIP INCREASES AS INCOME INCREASES. (2) POPULATION DENSITY IN THE ZONE OF RESIDENCE, REPRESENTING THE DIFFICULTY OF AUTO STORAGE AT HOME. MULTIPLE AUTO OWNERSHIP DECREASES AS POPULATION DENSITY INCREASES. (3) THE RELATIVE ACESSIBILITY (TRANSIT VS. AUTO) TO SHOPPING DESTINATIONS FROM THE RESIDENCE ZONE. AUTO OWNERSHIP DECREASES AS THIS RELATIVE ACCESSIBILITY INCREASES. 9.3.2 DISAGGREGATE MODEL FORMULATION THIS LOGIT MODEL HAS THE FOLLOWING UTILITY EQUATIONS: U(0,H) :: = B(1)*#RSUMSHA(H:AO=0) U(1,H) ::p = C(1)+B(2)*#LPCDINC(H) U(2+,H) :: = C(2)+B(3)*#LPCDINC(H)+B(4)*POPCEN(I(H)) WHERE: U( ,H) IS THE UTILITY TO HOUSEHOLD H OF OWNING AO AUTOS. #RSUMSHA(H:AO=0) IS THE RATIO OF DENOMINATORS OF THE DISAGGREGATE HBSHD MODEL (TRANSIT OVER AUTO) FOR HOUSEHOLD H. #LPCDINC(H) IS THE NATURAL LOG OF PER CAPITA DISPOSABLE INCOME FOR HOUSEHOLD H (1965 DOLLARS/PERSON), COMPUTED AS: NWHHAC 9.3-2 #LPCDINC(H) :: = LN(#INCOME(H)/#PHH(H) - $750) WHERE: #INCOME(H) IS HOUSEHOLD H'S GROSS ANNUAL INCOME (1965 DOLLARS) #PHH(H) IS THE NUMBER OF PERSONS IN HOUSEHOLD H (PERSONS) POPDEN(I(H)) IS THE POPULATION DENSITY IN H'S RESIDENCE ZONE, I(H) (PERSONS PER ACRE) C(1),C(2),B(1)-B(4) ARE MODEL COEFFICIENTS: VARIABLE COEFF VALUE ------- C(1) -.8695 ------- C(2) -8.357 #RSUMSHA B(1) .6070 #LPCDINC B(2) .3188 #LPCDINC B(3) 1.227 POPDEN B(4) -.0682 9.3.3 AGGREGATE MODEL FORMULATION U(0,I,G) :: = B(1)*RSUMSH(I) U(1,I,G) :: = C(1)+B(2)*LPCDINC(I,G) U(2+,I,G) :: = C(2)+B(3)*LPCDINC(I,G)+B(4)*POPDEN(I) WHERE: U( ,I,G) IS THE UTILITY TO HOUSEHOLDS OF INCOME GROUP G RESIDING IN ZONE I AND OWNING AO AUTOMOBILES. RSUMSH(I) IS THE RATIO OF DENOMINATORS OF THE HBSHD MODEL (TRANSIT OVER AUTO) FOR HOUSEHOLDS IN ZONE I WHICH OWN ZERO AUTOS. LPCDINC(I,G) IS THE NATURAL LOG OF PER CAPITA DISPOSABLE INCOME FOR HOUSEHOLDS OF INCOME GROUP G IN ZONE I (1965 DOLLARS/PERSON), COMPLETED AS: LPCDINC(I,G) :: = LN(INCOMEG(I,G)/PHHG(I,G) - $750 WHERE: INCOME(I,G) IS THE MEAN GROSS ANNUAL INCOME IN ZONE I FOR INCOME GROUP G (1965 DOLLARS). PHH(I,G) IS THE MEAN SIZE OF HOUSEHOLDS IN ZONE I FOR INCOME GROUP G (PERSONS). POPDEN(I) IS THE POPULATION DENSITY OF ZONE I (PERSONS/ACRE). C(1),C(2),B(1)-B(4) ARE MODEL COEFFICIENTS: NWHHAC 9.3-3 VARIABLE COEFF VALUE ------- C(1) -1.8695 ------- C(2) -8.773 RSUMSHA B(1) .6070 LPCDINC B(2) .3188 LPCDINC B(3) 1.227 POPDEN B(4) -.0682 WHHAO 9.4-1 9.4 WORKING HOUSEHOLD AUTO OWNERSHIP MODEL (WHHAO) 9.4.1 GENERAL DESCRIPTION IN ITS DISAGGREGATE FORM, THE WHHAO MODEL PREDICTS THE PROBABILITIES OF A WORKING HOUSEHOLD OWNING 0, 1, AND 2 OR MORE CARS. IN ITS AGGREGATE FORM, IT PREDICTS THE FRACTIONS OF A GROUP OF WORKING HOUSEHOLDS THAT OWN 0, 1, AND 2 OR MORE CARS, GIVEN THE ZONE OF EMPLOYMENT FOR THE PRIMARY WORKER. IN MTCFCAST, THESE HOUSEHOLD GROUPS ARE DEFINED BY RESIDENCE ZONE AND INCOME LEVEL. BOTH FORMS OF THE MODEL USE COMPOSITE VARIABLES BASED ON THE PHBWM AND THE HBSHD MODELS. IN ADDITION,M THE AGGREGATE FORM USES A COMPOSITE INCOME VARIABLE (LNRINC) THAT IS COMPUTED PROBABILISTICALLY USING OUTPUTS FROM THE PHBWM MODEL. TO OBTAIN AUTO OWNERSHIP VOLUMES THE AGGREGATE MODEL MUST BE APPLIED TO A PHBW TRIP TABLE, OBTAINED FROM THE PHBWD MODEL. THE AGGREGATE FORM OF THE MODEL IS DESIGNED TO BE APPLIED TO THE PREDICTED NUMBER OF PRIMARY HOME BASED WORK TRIPS THAT ARE OBTAINED FROM THE PHBWD MODEL. THE MODEL RESULTS ARE USED TO PROVIDE WORK TRIP TOTALS BY AUTO OWNERSHIP LEVEL AS STRATIFICATIONS FOR USE BY THE PRIMARY WORKER WORK MODE CHOICE MODEL (PHBWM) AND FOR EACH OF THE FOLLOWING TRIP GENERATION MODELS: SHBWG, HBSHG, AND HBSRG. THE INDEPENDENT VARIABLES AND THEIR EFFECTS ON PREDICTED AUTO OWNERSHIP ARE THE FOLLOWING: (1) HOUSEHOLD INCOME REMAINING AFTER PROVIDING THE BASIC PERSONAL NEEDS OF EACH HOUSEHOLD MEMBER, PAYING FIXED AUTO OWNERSHIP COSTS, AND PAYING THE EXPECTED COSTS OF WORK TRIPS TO THE CHOSEN ATTRACTION ZONE. AUTO OWNERSHIP INCREASES AS INCOME INCREASES. (2) HOUSING TYPE, SINGLE FAMILY HOUSING SERVES AS A PROXY FOR THE EASE OF AUTO STORAGE AT HOME. AUTO OWNERSHIP INCREASES AS THE EASE OF AUTO STORAGE INCREASES. (3) HOUSEHOLD SIZE, REPRESENTING THE INCREASED DEMAND FOR AUTOS WHICH EXISTS IN LARGER HOUSEHOLDS. (4) RETAIL AND SERVICE EMPLOYMENT DENSITY IN THE RESIDENCE ZONE, REPRESENTING BOTH THE PROBABILITY OF BEING ABLE TO WALK TO WORK AND SHOPPING ATTRACTIONS, AND THE DIFFICULTY OF AUTO STORAGE AT HOME. AUTO OWNERSHIP DECREASES AS THIS DENSITY MEASURE INCREASES. (5) THE RELATIVE ACCESSIBILITY (TRANSIT/AUTO) TO SHOPPING DESTINATIONS FROM THE RESIDENCE ZONE. AUTO OWNERSHIP DECREASES AS THIS RELATIVE ACCESSIBILITY INCREASES. (6) THE RELATIVE ACCESSIBILITY (TRANSIT/AUTO) TO THE CHOSEN PRIMARY WORKER TRIP ATTRACTION ZONE. AUTO OWNERSHIP DECREASES AS THIS RELATIVE ACCESSIBILITY INCREASES. WHHAO 9.4-2 9.4.2 DISAGGREGATE MODEL FORMULATION THIS LOGIT MODEL HAS THE FOLLOWING UTILITY EQUATIONS: U(0,H) :: = B(1)*#LNRINC(H:AO=0)+B(2)*#REXP(H:AO=0) +B(3)*#RSUMSHA(H:AO=0) U(0,H) :: = C(1)+B(1)*#LNRINC(H:AO=1)+B(2)*#REXP(H:AO=1) +B(3)*RETSERD(I(H)) +B(5)*#INVPHH(H)+B(6)*#HOUSE(H) U(0,H) :: = C(2)+B(1)*#LNRINC(H:AO=2+)+B(2)*#REXP(H:AO=2+) +B(4)*RETSERD(I(H)) +B(5)*$MAO(2+)*#INVPHH(H)+B(7)*#HOUSE(H) WHERE: U( ,H) IS THE UTILITY TO HOUSEHOLD H OF OWNING AO AUTOMOBILES. #LNRINC(H: ) IS THE NATURAL LOGARITHM OF THE REMAINING INCOME OF HOUSEHOLD H, GIVEN AUTO OWNERSHIP LEVEL AO (REMAINING INCOME IN 1965 DOLLARS). #RINC IS COMPUTED AS: #RINC :: = (ANNUAL GROSS HOUSEHOLD INCOME)-$500*(EXPECTED COST OF PRIMARY WORKER TRIP)-$750*PERSONS IN HOUSEHOLD)- $1000*(AVERAGE AUTO OWNERSHIP AT GIVEN AO LEVEL) #REXP(H: ) IS THE RATIO OF THE EXPONENTIATED UTILITIES OF THE PHBWM MODEL (TRANSIT OVER AUTO) FOR HOUSEHOLDS, GIVEN AUTO OWNERSHIP LEVEL AO. ##RSUMSHA(H: ) IS THE RATIO OF THE DENOMINATOR OF THE HBSHD MODEL (TRANSIT OVER AUTO) FOR HOUSEHOLD H, GIVEN AUTO OWNERSHIP LEVEL AO. RETSERD(I(H)) IS THE RETAIL AND SERVICE EMPLOYMENT DENSITY IN HOUSEHOLD H'S ZONE OF RESIDENCE I(H) (EMPLOYEES PER ACRE). #INVPHH(H) IS THE INVERSE OF HOUSEHOLD H'S HOUSEHOLD SIZE (HOUSEHOLD SIZE IN PERSONS). #HOUSE(H) IS THE HOUSING TYPE CODE FOR HOUSEHOLD H (=1 FOR SINGLE FAMILY DWELLING, =0 FOR MULTIPLE FAMILY DWELLING). C(1),C(2),B(1)-B(7) ARE MODEL COEFFICIENTS WHHAC 9.4-3 VARIABLE COEFF VALUE ------- C(1) 4.989 ------- C(2) 5.689 #LNRINC B(1) .7919 #REXP B(2) .06814 #RSUMSHA B(3) .5608 RETSERD B(4) -.05419 #INVPHH B(5) -2.689 #HOUSE B(6) .3935 #HOUSE B(7) 1.342 $MAO(2+) IS THE MEAN REGIONAL HOUSEHOLD AUTO OWNERSHIP FOR THOSE HOUSEHOLDS IN THE 2+ AUTO OWNERSHIP LEVEL (AUTOS/HOUSEHOLD); 2.236 RECOMMENDED FOR 1965. 9.4.3 AGGREGATE MODEL FORMULATION U(0,I,J,G) :: = B(1)*LNRINC(I,J,G:AO=0)+B(2)*REXP(I,J:AO=0) +B(3)*RSUMSH(I) U(1,I,,J,G) :: = C(1)+B(1)*LNRINC(I,J,G:AO=1)+B(2)*REXP(I,J:AO=1) +B(4)*RETSERD(I)+B(5)*($PHH/$PHHG(O))*INVPHH(I) +B(6)*HOUSE(I) U(2+,I,J,G) :: = C(2)+B(1)*LNRINC(I,J,G:AO=2+)+B(2)*REXP(I,J:AO=2+) +B(4)*RETSERD(I) +B(5)*$MAO(2+)*($PHH/$PHHG(G))*INVPHH(I) +B(7)*HOUSE(I) WHERE: U( ),I,J,G) IS THE UTILITY OF OWNING AO AUTOMOBILES TO HOUSEHOLDS SITUATED IN ZONE I WHOSE PRIMARY WORKERS WORK IN ZONE J, AND BELONG TO INCOME GROUP G. LNRINC(I.J.G: ) IS THE NATURAL ALGORITHM OF THE REMAINING INCOME OF HOUSEHOLDS SITUATED IN ZONE I, BELONGING TO INCOME GROUP G, WHOSE PRIMARY WORKERS WORK IN ZONE J, FOR AUTO OWNERSHIP LEVEL AP (REMAINING INCOME IN 1965 DOLLARS). RINC IS COMPUTED AS: RINC(I,J,G,A) :: = INCOME G (I,G) -$500*ECOST(I,J,A) -$750*PHHG(I,G) -$1000*$MAO(A) WHERE: INCOMEG(I,G) IS THE MEAN GROSS ANNUAL HOUSEHOLD INCOME IN ZONE I FOR THE INCOME GROUP G (IN 1965 DOLLARS). ECOST (I,J,A) IS THE EXPECTED TRAVEL COST FROM ZONE I TO ZONE J FOR PRIMARY HBW TRIPS MADE BY WORKERS IN HOUSEHOLDS IN AUTO OWNERSHIP GROUP A (1965 DOLLARS). WHHAO 9.4-4 PHHG(I,G) IS THE MEAN SIZE OF HOUSEHOLDS IN ZONE I FOR INCOME GROUP G (PERSONS/HOUSEHOLD). $MAO(A) IS THE MEAN REGIONAL HOUSEHOLD AUTO OWNERSHIP FOR HOUSEHOLDS IN AUTO OWNERSHIP GROUP A (AUTOS/HOUSEHOLD). IN 1965, FOR A = 0, 1, 2+, $MAO = 0, 1, 2.236. REXP(I,J: ) IS THE RATIO OF THE EXPONENTIATED UTILITIES OF THE PHBWM MODEL (TRANSIT OVER AUTO) FOR HOUSEHOLDS SITUATED IN ZONE I, WHOSE PRIMARY WORKERS WORK IN ZONE J, AND FOR AUTO OWNERSHIP LEVEL AO. RSUMSH(I) IS THE RATIO OF THE DENOMINATORS OF THE HBSHD MODEL (TRANSIT OVER AUTO FOR HOUSEHOLDS IN ZONE I, GIVEN ZERO AUTO OWNERSHIP. RETSERD(I) IS THE RETAIL AND SERVICE EMPLOYMENT DENSITY IN ZONE I (EMPLOYEES/ACRE). INVPHH(I) IS THE INVERSE OF THE AVERAGE HOUSEHOLD SIZE IN ZONE I (HOUSEHOLD SIZE IN PERSONS). HOUSE(I) IS THE RATIO OF SINGLE FAMILY DWELLINGS TO TOTAL NUMBER OF DWELLINGS IN ZONE I. C(1), C(2),B(1)-B(7) ARE MODEL COEFFICIENTS: VARIABLE COEFF VALUE ------- C(1) 4.475 ------- C(2) 5.836 LNRINC B(1) .7919 REXP B(2) .06814 RSUMSHA B(3) .5608 RETSERD B(4) -.05419 INVPHH B(5) -2.689 HOUSE B(6) .3935 HOUSE B(7) 1.342 $MAO(2+) IS THE MEAN REGIONAL HOUSEHOLD AUTO OWNERSHIP FOR THOSE HOUSEHOLDS IN THE 2+ AUTO OWNERSHIP LEVEL (AUTOS/HOUSEHOLD); 2.236 RECOMMENDED FOR 1965. $PHH IS THE MEAN REGIONAL HOUSEHOLD SIZE (PERSONS/HOUSEHOLD); 3.81 RECOMMENDED. $PHHG(G) IS THE MEAN REGIONAL HOUSEHOLD SIZE FOR HOUSEHOLDS IN INCOME GROUP G (PERSONS PER HOUSEHOLD); 2.96, 3.53 AND 4.22 RECOMMENDED FOR INCOME GROUPS 1, 2, AND 3 RESPECTIVELY. THESE ARE USED WITH $PHH TO ESTIMATE THE AVERAGE RELATIVE HOUSEHOLD SIZE BY INCOME GROUP. PHBWG 9.5-1 9.5 PRIMARY WORKER HOME BASED WORK TRIP GENERATION MODEL (PHBWG 9.5.1 GENERAL DESCRIPTION IN ITS DISAGGREGATE FORM, THE PHBWG MODEL PREDICTS THE NUMBER OF DAILY HOME BASED WORK TRIPS MADE BY A PRIMARY WORKER. IN ITS AGGREGATE FORM, THE MODEL PREDICTS THE VOLUME OF PRIMARY WORKER HOME BASED WORK TRIPS PRODUCED BY A GROUP OF PRIMARY WORKERS. IN MTCFCAST, THESE GROUPS ARE DEFINED BY RESIDENCE ZONE AND HOUSEHOLD INCOME LEVEL. IN ITS AGGREGATE FORM, THE MODEL REQUIRES THE NUMBER OF PRIMARY WORKERS (I.E., THE NUMBER OF HOUSEHOLDS WITH WORKERS) BY ZONE, OBTAINED FROM THE NWHH MODEL. THE MODEL RESULTS -- THE PRIMARY WORKER TRIP PRODUCTION TOTALS -- ARE ALLOCATED TO DESTINATIONS BY THE PWHB MODEL. VARIATIONS IN DAILY HOME BASED WORK TRIPS FROM THE NOMINAL VALUE OF TWO TRIPS PER DAY MADE BY PRIMARY WORKERS, OCCUR FOR A NUMBER OF REASONS. INCREASES CAN BE CAUSED BY WORKERS WHO GO HOME FOR LUNCH OR HAVE MORE THAN ONE JOB. DECREASES CAN BE CAUSED VACATIONS AND DAYS OFF (EITHER ELECTED OR REQUIRED BY THE EMPLOYER), COMPLEX TRIPS TO WORK WHICH INCLUDE STOPS FOR NON-WORK PURPOSES, AND WALKING TO WORK. THE MODEL'S INDEPENDENT VARIABLES, HOUSEHOLD INCOME AND TOTAL EMPLOYMENT DENSITY IN THE RESIDENCE ZONE, ARE PROXIES FOR ONE OR MORE OF THESE SOURCES OF VARIATION. TRIPS PER WORKER INCREASE WITH INCREASING INCOME AND DECREASE WITH INCREASING EMPLOYMENT DENSITY. 9.5.2 DISAGGREGATE MODEL FORMULATION THIS LINEAR MODEL HAS THE FOLLOWING EQUATION: #PHBWGV(Q) :: = MAX($MINGP,(C+B(1)*LN(#INCOME(H(Q))/1000) +B(2)*TOTEMPD(I(Q)))) WHERE: #PHBWGV(Q) IS THE PREDICTED DAILY HOME BASED WORK TRIP PRODUCTION RATE FOR PRIMARY WORKER Q (TRIPOS/DAY). #INCOME(H(Q) IS THE ANNUAL GROSS INCOME OF Q'S HOUSEHOLD, H(Q) (1965 DOLLARS). TOTEMPD(I(Q) IS THE GROSS EMPLOYMENT DENSITY OF Q'S ZONE OF RESIDENCE, I(Q) (EMPLOYEES/ACRE). C+B(1)B(2) ARE MODEL COEFFICIENTS: VARIABLE COEFF VALUE ---------- C 1.422 LN(#INCOME/1000) B(1) .0755 TOTEMPD B(2) -.0127 $MINGP IS A LOWER BOUND ON THE TRIP GENERATION RATE PHBWG 9.5-2 (TRIPS/WORKER) : .05 RECOMMENDED. 9.5.3 AGGREGATE MODEL FORMULATION PHBWGV(I,G) :: = WHHSG(I,G)*$PHBWGFR*MAX($MINGP,(C +B(1)*LN(INCOMEG(I,G)/1000)+B(2)*TOTEMPD(I))) WHERE: PHBWGV(I,G) IS THE NUMBER OF PRIMARY WORKER HOME BASED WORK TRIPS PRODUCED IN ZONE I BY HOUSEHOLDS IN INCOME GROUP G (TRIPS). WHHSG(I,G) IS THE VOLUME OF WORKING HOUSEHOLDS IN ZONE I BELONGING TO INCOME GROUP G (HOUSEHOLDS). INCOMEG(I,G) IS THE MEAN GROSS ANNUAL HOUSEHOLD INCOME FOR HOUSEHOLDS IN ZONE I BELONGING TO INCOME GROUP G (1965 DOLLARS). TOTEMPD(I) IS THE GROSS EMPLOYMENT DENSITY IN ZONE I (EMPLOYEES/ACRE). C+B(1)B(2) ARE MODEL COEFFICIENTS: VARIABLE COEFF VALUE ---------- C 1.422 LN(#INCOME/1000) B(1) .0755 TOTEMPD B(2) -.0127 $MINGP IS A LOWER BOUND ON THE TRIP GENERATION RATE (TRIPS/WORKER) : .05 RECOMMENDED. $PHBWGFR IS A REGIONAL ADJUSTMENT FACTOR; .9335 RECOMMENDED FOR SAN FRANCISCO BAY AREA APPLICATIONS ONLY; 1.0 RECOMMENDED OTHERWISE. $PHBWGF(I) IS A ZONAL ADJUSTMENT FACTOR FOR POPULATION ZONE I; FACTORS DERIVE FROM THE DISTRICT FACTORS GIVEN IN TABLE 9.5-1 ARE RECOMMENDED FOR SAN FRANCISCO BAY AREA APPLICATIONS ONLY; OTHERWISE 1.0 RECOMMENDED FOR ALL ZONES. PHBWG 9.5-3 DIST VALUE DIST VALUE DIST VALUE 1 .951 11 1.006 21 .935 2 .967 12 1.042 22 1.209 3 1.062 13 .958 23 1.575 4 .896 14 .948 24 1.176 5 .922 15 1.007 25 .752 6 .920 16 1.079 26 .924 7 .893 17 .964 27 .976 8 .924 18 1.065 28 1.033 9 1.109 19 .879 29 1.036 10 1.046 20 .956 30 .956 TABLE 9.5-1 PHBWG AND SHBWG AGGREGATE VALIDATION ADJUSTMENTS $PHBWGF AND $SHBWGF DISTRICT VALUES PHBWA 9.6-1 9.6 PRIMARY WORKER HOME BASED WORK TRIP ATTRACTION MODEL (PHBWA) 9.6.1 GENERAL DESCRIPTION THE PHBWA MODEL PREDICTS THE NUMBER OF PRIMARY WORKER HOME BASED WORK TRIPS ATTRACTED TO A ZONE. THIS IS AN AGGREGATE MODEL ONLY; NO DISAGGREGATE FORM EXISTS. THE PHBWA MODEL IS INDEPENDENT OF THE OUTPUTS OF ALL OTHER MODELS IN THE MTC SYSTEM. ITS RESULTS ARE USED AS AN ATTRACTION-ZONE-SIZE VARIABLE IN THE PHBWD MODEL. VARIATIONS IN TRIP ATTRACTIONS FROM THE NORMAL VALUE IF TWO TRIPS PER EMPLOYEE PER DAY OCCUR FOR THE SAME REASONS THEY DO FOR TRIP PRODUCTIONS (SEE 9.5.1). IN ADDITION, THE MANUFACTURING EMPLOYMENT GROUP WAS FOUND TO CONSISTENTLY ATTRACT THE HIGHEST NUMBER OF PRIMARY WORKER TRIPS PER JOB. SINCE NO SOCIO-ECONOMIC DATA FOR WORKERS IS AVAILABLE AT THE ATTRACTION END OF THEIR TRIPS, VARIABLES SUCH AS HOUSEHOLD INCOME CANNOT BE USED IN THIS MODEL. THE INDEPENDENT VARIABLES IN THE MODEL ARE THE FRACTION OF ZONAL EMPLOYMENT THAT IS MANUFACTURING, AND WORKERS LIVING IN THE ZONE DIVIDED BY TOTAL ZONAL JOBS, BY POPULATION SERVING JOBS, AND BY MANUFACTURING JOBS. THESE LAST THREE VARIABLES ARE PROXIES FOR THE LIKELIHOOD OF LIVING CLOSE ENOUGH TO WORK TO GO HOME FOR LUNCH AND/OR TO AVOID MAKING WORK TRIPS AS PART OF COMPLEX TRIP CHAINS.L WITH THE RELATIVELY LARGE SIZE OF THE MTC ZONES (THE AVERAGE ZONE SIZE OF THE 290 ZONE SYSTEM IS GREATER THAN 24 SQUARE MILES), IT IS NOT SURPRISING THAT NO ZONAL VARIABLE COULD BE FOUND TO SERVE AS A PROXY FOR THE LIKELIHOOD OF WALKING TO WORK. 9.6.2 AGGREGATE MODEL FORMULATION THIS LINEAR MODEL HAS THE FOLLOWING EQUATION: PHBWA(I) :: = MAX($HBWAF(I)*$PHBWAFR*TOTEMP(I)*(C +B(1)*WORKERS(I)/TOTEMP(I) +B(2)*WORKERS(I)/PSREMP(I) +B(3)*WORKERS(I)/MFGEMP(I) +B(4)*MFGEMP(I)/TOTEMP(I)),0) WHERE: PHBWA(I) IS THE TOTAL NUMBER OF PRIMARY HOME BASED WORK TRIPS ATTRACTED TO ZONE I (TRIPS). TOTEMP(I) IS THE TOTAL EMPLOYMENT OF ZONE I (EMPLOYEES). WORKERS(I) IS THE TOTAL NUMBER OF WORKERS RESIDING IN ZONE I (WORKERS). PSREMP(I) IS THE POPULATION-SERVING EMPLOYMENT IN ZONE I (EMPLOYEES). MFGEMP(I) IS THE TOTAL NUMBER OF EMPLOYEES OF MANUFACTURING INDUSTRIES IN ZONE I (EMPLOYEES). C,B(1)-B(4) ARE MODEL COEFFICIENTS: PHBWA 9.6-2 VARIABLE COEFF VALUE -------- C .874 WORKERS/TOTEMP B(1) .0757 WORKERS/PSREMP B(2) -.0396 WORKERS/MFGEMP B(3) -5.00E - 5 MFGEMP/TOTEMP B(4) .502 $PHBWAFR IS A REGIONAL ADJUSTMENT FACTOR: 1.0217 RECOMMENDED FOR SAN FRANCISCO BAY AREA APPLICATIONS ONLY, 1.0 RECOMMENDED OTHERWISE. $HBWAF(I) IS A ZONAL ADJUSTMENT FACTOR FOR ATTRACTION ZONE I; FACTORS FROM THE DISTRICT ADJUSTMENT FACTORS GIVEN IN TABLE 9.6-1 ARE RECOMMENDED FOR SAN FRANCISCO BAY AREA ONLY; OTHERWISE 1.0 RECOMMENDED FOR ALL ZONES. PHWBA 9.6-3 DIST VALUE DIST VALUE DIST VALUE 1 1.006 11 1.005 21 .856 2 .844 12 1.104 22 1.263 3 .926 13 .980 23 1.399 4 .816 14 .914 24 .958 5 1.004 15 1.032 25 .940 6 .818 16 1.029 26 1.042 7 .896 17 .977 27 .918 8 1.051 18 1.031 28 1.027 9 1.161 19 .959 29 1.103 10 .899 20 1.017 30 .959 TABLE 9.6-1 PHBWA AND SHBWA AGGREGATE VALIDATION ADJUSTMENTS $HBWAF DISTRICT VALUES PHBWD 9.7-1 9.7 PRIMARY WORKER HOME BASED WORK TRIP DISTRIBUTION MODEL (PHBWD) 9.7.1 GENERAL DESCRIPTION IN ITS DISAGGREGATE FORM THE PHBWD MODEL PREDICTS THE PROBABILITY OF A PRIMARY WORKER'S HOME BASED WORK TRIP BEING ATTRACTED TO A PARTICULAR ZONE. IN ITS AGGREGATE FORM IT PREDICTS THE FRACTION OF A GROUP OF HOME BASED PRIMARY WORK TRIPS ATTRACTED TO EACH DESIGNATION ZONE. IN MTCFCAST, THESE TRIP GROUPS ARE DEFINED BY ORIGIN ZONE AND INCOME LEVEL. THE PHBWD MODEL REQUIRES INPUTS FROM THE PHBWA MODEL AND THE PHBWM AND WHHAO MODELS. ALSO REQUIRED ARE TRIP PRODUCTIONS TO BE DISTRIBUTED; THESE ARE PROVIDED BY THE PHBWG MODEL. THE MODEL RESULTS ARE USED TO PROVIDE PRIMARY WORKER PERSON TRIPS BY ZONE PAIR, WHICH ARE ALLOCATED TO AUTO OWNERSHIP GROUPS BY THE WHHAQ MODEL. THE MODEL HAS THREE INDEPENDENT VARIABLES: (1) THE EXPECTED ACCESSIBILITY (WEIGHTED BY AUTO OWNERSHIP PROBABILITIES) OF EACH ALTERNATIVE ATTRACTION ZONE. THIS VARIABLE INCORPORATES THE ACCESSIBILITY EFFECTS OF BOTH ALTERNATIVE MODES (IN MODE CHOICE VARIABLES) AND ALTERNATIVE AO LEVELS (IN AO PROBABILITIES) BY USING THE RESULTS OF SUBSEQUENT MODELS IN THE PROCESSING SEQUENCE, RATHER THAN BY REQUIRING COMPLETE KNOWLEDGE OF MODE CHOICE AND AO LEVEL IN ORDER TO PREDICT ATTRACTION ZONE CHOICE. IN ADDITION, THIS VARIABLE COLLAPSES MOST OF THE IMPEDENCE TERMA AFFECTION ATTRACTION ZONE CHOICE INTO A SINGLE VARIABLE. THE PROBABILITY OF CHOOSING A GIVEN ATTRACTION ZONE INCREASES AS ITS ACCESSIBILITY INCREASES. (2) THE TRIP LENGTH (DISTANCE) CORRECTION VARIABLE TO EACH ATTRACTION ZONE; THE AMOUNT BY WHICH THE PRODUCTION TO ATTRACTION ZONE TRIP DISTANCE EXCEEDS THE THRESHOLD VALUES OF THREE MILES. THIS VARIABLE REPRESENTS THE DECREASING KNOWLEDGE OF POTENTIAL WORK LOCATIONS AL THE DISTANCE FROM HOME INCREASES. THE PROBABILITY OF CHOOSING A GIVEN ATTRACTION ZONE DECREASES AS TRIP DISTANCE INCREASES, FOR ALL DISTANCES GREATER THAN THREE MILES. (3) THE NUMBER OF PRIMARY WORKER HBW TRIP ATTRACTIONS IN THE ATTRACTION ZONE, AS PREDICTED BY THE PHBWA MODES. THE PROBABILITY OF CHOOSING A GIVEN ATTRACTION ZONE INCREASES AS ITS TRIP ATTRACTIONS INCREASE. 9.7.2 DISAGGREGATE MODEL FORMULATION THIS LOGIT MODEL HAS THE FOLLOWING UTILITY EQUATION: U(J,Q) :: = B(1)*LNPHBWA(J)+B(2)*#WLSUM(J,Q)+B(3)*PDIST3(I(Q),J) PHBWD 9.7-2 WHERE: U(J,Q) IS THE UTILITY OF ATTRACTION ZONE J TO PRIMARY WORKER Q. LNPHWBWA(J) IS THE NATURAL LOGARITHM OF THE TOTAL NUMBER OF PRIMARY WORKER HOME BASED WORK TRIPS ATTRACTED TO ZONE J (PHBWA IN TRIPS). #WLSUM(J,Q) IS THE WEIGHTED MEAN NATURAL LOGARITHM OF THE AO SPECIFIC DENOMINATORS OF THE PHBWM MODEL FOR PRIMARY WORKER Q AND DESTINATION ZONE J, WHERE THE WEIGHTS ARE THE PROBABILITIES OF Q'S HOUSEHOLD OWNING 0, 1 AND 2+ CARS AS PREDICTED BY THE DISAGGREGATE WHHAO MODEL, GIVEN J AS THE WORK DESTINATION ZONE. PDIST 3(I(Q),J) IS A DISTANCE CORRECTION VARIABLE, COMPUTED AS: PDIST3(I,J) :: = MAX ((&ADIST(I,J)-3),0) WHERE: &ADIST(I,J) IS THE 1-WAY PEAK PERIOD AUTO DISTANCE FROM I TO J (MILES). B(1)-B(3) ARE MODEL COEFFICIENTS: VARIABLE COEFF VALUE LNPHBWA B(1) 1.00 #WLSUM B(2) 1.811 PDIST3 B(3) -.056 9.7.3 AGGREGATE MODEL FORMULATION U(I,J,G) :: = B(1)*LNPHBWA(J)+B(2)*WLSUM(I,J,G)+B(3)*PHBWDCF(I,J) WHERE: U(I,J,G) IS THE UTILITY OF ATTRACTION ZONE J TO PRIMARY WORKERS IN ZONE I BELONGING TO INCOME GROUP G. LNPHBWA(J) IS THE NATURAL LOGARITHM IF THE TOTAL NUMBER OF PRIMARY WORKER HOME BASED WORK TRIPS ATTRACTED TO ZONE J (PHBWA IN TRIPS). WLSUM(I,J,G) IS THE WEIGHTED MEAN NATURAL LOGARITHM IF THE AO-SPECIFIC DENOMINATORS OF THE FHBWM MODEL FOR ORIGIN ZONE I, DESTINATION ZONE J AND INCOME GROUP G, WHERE THE WEIGHTS ARE THE PROBABILITIES IF A HOUSEHOLD OF INCOME GROUP G IN ZONE I OWNING 0, 1 AND 2+ CARS AS PREDICTED BY THE AGGREGATE WHHAO MODEL, GIVEN J AS THE WORK ZONE. PHBWDCF(I,J) IS A DISTANCE CORRECTION FACTOR, COMPUTED AS: PHBWDCF(I,J) :: = MIN($PHBWDC(I)*(&ADIST(I,J)-3),0) WHERE: &ADIST(I,J) IS THE 1-WAY PEAK PERIOD AUTO DISTANCE FROM 9.7-3 I TO J (MILES) $PHBDWDC(I) IS A COEFFICIENT OF THE DISTANCE CORRECTION VARIABLE FOR ZONE I; DERIVED FROM DISTRICT VALUES GIVEN IN TABLE 9.7-1 FOR SAN FRANCISCO BAY AREA APPLICATIONS ONLY; OTHERWISE, -.060 RECOMMENDED FOR ALL ZONES. B(1)-B(3) ARE MODEL COEFFICIENTS: VARIABLE COEFF VALUE LNPHBWA B(1) 1.00 WLSUM B(2) 1.811 PHBWDCF B(3) 1.000 9.7-4 DIST VALUE DIST VALUE DIST VALUE 1 -.001 11 -.055 21 -.084 2 -.021 12 -.003 22 -.075 3 -.061 13 -.113 23 -.158 4 -.031 14 -.094 24 -.068 5 -.062 15 -.067 25 -.054 6 -.063 16 -.051 26 -.062 7 -.035 17 -.060 27 -.047 8 -.050 18 -.094 28 -.006 9 -.030 19 -.106 29 -.044 10 -.044 20 -.068 30 -.047 TABLE 9.7-1 PHBWD AGGREGATE VALIDATION ADJUSTMENTS COEFFICIENTS OF DISTANCE (TRIP LENGTH) CORRECTION VARIABLE $PHBWDC DISTRICT VALUES 9.8-1 9.8 PRIMARY WORKER HOME BASED WORK TRIP MODE MODEL (PHBWM) 9.8.1 GENERAL DESCRIPTION IN ITS DISAGGREGATE FORM, THE PHBWM MODEL PREDICTS THE PROBABILITIES OF A PRIMARY WORKER CHOOSING EACH OF THREE MODES (DRIVE ALONE, SHARED-RIDE, AND TRANSIT) FOR HIS WORK TRIP, GIVEN HIS WORK LOCATION. IN ITS AGGREGATE FORM, THE MODEL PREDICTS, FOR A GROUP OF PRIMARY WORKERS LIVING IN A PARTICULAR ZONE AND WORKING IN A PARTICULAR ZONE, THE FRACTIONS THAT WILL USE EACH IF THE THREE MODES ON THEIR HOME BASED WORK TRIPS. IN MTCFCAST, THE GROUPS ARE BASED ON SEGMENTATION BY AUTO OWNERSHIP LEVEL (0, 1, 2+). NONE OF THE INDEPENDENT VARIABLES IS DIRECTLY DEPENDENT IN ANY OTHER MODEL IN THE MTC SYSTEM. HOWEVER, THE SHARED-RIDE COST VARIABLES ARE USUALLY COMPUTED USING AUTO OCCUPANCIES PRODUCED BY THE SROCC MODEL. MOREOVER, TO OBTAIN MODAL TRIP TABLES, THE AGGREGATE MODEL MUST BE APPLIED TO PRIMARY WORKER TRIP TABLES STRATIFIED BY AUTO OWNERSHIP, USUALLY OBTAINED FROM THE WHHAO MODEL. THE MODEL RESULTS, IN THE FORM IF PRODUCTION/ATTRACTION WORK TRIPS BY MODE, CAN BE CONVERTED TO ORIGIN/DESTINATION WORK TRIPS USING SPLITTING FACTORS GENERATED BY THE NHB MODELS. THE PRODUCTION/ATTRACTION WORK TRIPS, SUMMED BY ATTRACTION ZONE, ALSO SERVE AS A PORTION IF THE POPULATION OF POTENTIAL NHB TRIPS. THE SHARED-RIDE TRIPS CAN BE CONVERTED FROM PERSON TRIPS TO VEHICLE TRIPS USING THE RESULTS OF THE SROCC MODEL. IN ADDITION, ACCESSIBILITY VARIABLES FROM THIS MODEL ARE USED IN THE WHHAO AND PHBWD MODELS. THE INDEPENDENT VARIABLES AND THEIR IMPACTS IN MODE CHOICE ARE: (1) HOUSEHOLD INCOME REMAINING AFTER PROVIDING THE BASIC PERSONAL NEEDS OF EACH HOUSEHOLD MEMBER. THE CHOICE OF TRANSIT DECREASES AS INCOME INCREASES. (2) NUMBER OF WORKERS IN THE HOUSEHOLD. THE CHOICE OF SHARED-RIDE INCREASES AS THE NUMBER OF WORKERS INCREASES. (3) AUTO AVAILABILITY - AUTOS PER WORKER. THE CHOICE OF TRANSIT DECREASES AS AUTO AVAILABILITY INCREASES. (4) AVAILABILITY IF TRANSIT SERVICE WITHIN WALKING DISTANCE OF THE RESIDENTIAL LOCATION. THE CHOICE OF TRANSIT DECREASES WHEN IT IS ONLY AVAILABLE BY AUTO ACCESS. (5) WHETHER OR NOT THE ATTRACTION ZONE IS IN A CENTRAL BUSINESS DISTRICT. THE CHOICE OF TRANSIT INCREASES FOR TRIPS TO THE CBD. (6) TOTAL WALK TIME BY MODE. THE CHOICE OF EACH MODE DECREASES AS ITS WALK TIME INCREASES. 9.8-2 (7) TOTAL IN-VEHICLE TIME BY MODE. THE CHOICE OF EACH MODE DECREASES AS ITS IN-VEHICLE TIME INCREASES. (8) TRANSIT HEADWAY. THE CHOICE IF TRANSIT DECREASES AS ITS HEADWAY INCREASES. (9) TRANSIT TRANSFER WAIT TIME. THE CHOICE OF TRANSIT DECREASES AS ITS TRANSFER TIME INCREASES. (10) OUT-OF-POCKET COST BY MODE. THE CHOICE OF EACH MODE DECREASES AS ITS COST INCREASES. 9.8.2 DISAGGREGATE MODEL FORMULATION THIS LOGIT MODEL HAS THE FOLLOWING UTILITY EQUATIONS: U(A,Q) :: = C(1)+B(1)*#DINC(H(Q))+B(2)*(WTERMP(I(Q))+WATERMA(J(G))) +B(3)*ATIMET(I(Q),J(Q))+B(4)*#ACOSTI(Q) +B(5)*CBD(J(Q))+B(6)*#CARS(H(Q))/#WHH(H(Q)) U(S.Q) :: = C(20+B(1)*#DINC(H(Q))+B(2)*(WTERMP(I(Q))+WTERMA(J(Q))) +B(3)*STIMET(I(Q),J(Q))+B(4)*#ACOSTI(Q) +B(7)*CBD(J(Q))+B(8)*#CARS(H(Q))/#WHH(H(Q)) +B(9)*#WHH(H(Q)) U(T,Q) :: = B(2)*TWALK(I(Q),J(Q))+B()3)*TTIME(I(Q),J(Q)) +B(4)*TCOSTi(Q)+B(10)*ACCDUM(I(Q),J(Q)) +B(11)*#ACCDUM2(Q)+B(12)*TWAIT1(I(Q),J(Q)) +B(13)*TWAIT2(I(Q),J(Q))=B(14)*TXFER(I(Q,J(Q)) WHERE: #PHHH(H(Q)) IS THE NUMBER OF PERSONS IN PRIMARY WORKER Q'S HOUSEHOLD, H(Q). WTERMP(I(Q)) IS THE 1-WAY PEAK PERIOD AUTO PRODUCTION WALK TERMINAL TIME IN Q'S HOME ZONE, I(Q) (MINUTES). WTERMA(J(Q)) IS THE 1-WAY PEAK PERIOD AUTO ATTRACTION WALK TERMINAL TIME IN Q'S WORK ZONE, J(Q) (MINUTES). TIMET(I(Q),J(Q)) IS THE 1-WAY PEAK PERIOD TRAVEL COST OF PRIMARY WORKER Q'S WORK TRIP BY MODE M, DIVIDED BY THE GROSS ANNUAL INCOME OF PRIMARY WORKER Q'S HOUSEHOLD, H(Q). CBD(J(Q)) IS A CBD STATUS SWITCH FOR WORK ZONE J(Q): =1 IF PHBWM 9.8-3 J(Q) IS IN THE CENTRAL BUSINESS DISTRICT; = 0 OTHERWISE. #CARS(H(Q)) IS THE NUMBER OF AUTOS AVAILABLE TO PRIMARY WORKER Q'S HOUSEHOLD, H(Q) (AUTOS). #WHH(H(Q)) IS THE NUMBER OF WORKERS IN PRIMARY WORKER Q'S HOUSEHOLD, H(Q) (WORKERS). TWALK(I(Q),J(Q)) IS THE 1-WAY PEAK PERIOD WALK TIME FOR WORKER Q'S WORK TRIP, IF HE USES TRANSIT (MINUTES). ACCDUM(I(Q), J(Q)) IS AN AUTO ACCESS DUMMY FOR Q'S TRIP TO WORK: = 1 IF AUTO ACCESS TO TRANSIT IS REQUIRED, = 0 OTHERWISE; WHERE AVAILABILITY IS DETERMINED ON A ZONAL BASIS. #ACCDUM2(Q) IS = #CARS(H(Q))/#WHH(H(Q)) IF WALK ACCESS TO TRANSIT IS NOT AVAILABLE TO PRIMARY WORKER Q FOR HIS WORK TRIP, = 0 OTHERWISE; WHERE AVAILABILITY IS DETERMINED ON A ZONAL BASIS (AUTOS/WORKER). TWAIT1(I(Q),J,(Q)) IS THE 1-WAY PEAK PERIOD TRANSIT WAIT TIME FOR PRIMARY WORKER Q'S WORK TRIP, UP TO 4 (MINUTES). TWAIT2(I(Q),J(Q)) IS THE 1-WAY PEAK PERIOD TRANSIT WAIT TIME FOR PRIMARY WORKER Q'S WORK TRIP, OVER 4 (MINUTES). TXFER(I(Q),J(Q)) IS THE 1-WAY PEAK PERIOD TRANSIT TRANSFER TIME FOR PRIMARY WORKER Q'S WORK TRIP (MINUTES). C(1),C(2),B(1)-B(14) ARE MODEL COEFFICIENTS: VARIABLE COEFF VALUE ------- C(1) -1.834 ------- C(2) -3.473 #DINC B(1) 1.817E - 5 TWALK, B(2) -.0670 WTERMP, WTERMA TIMET B(3) -.0244 # COSTI B(4) -4286 CBD B(5) -1.067 #CARS/#WHH B(6) 1.958 CBD B(7) -.347 #CARS/#WHH B(8) 1.763 #WHH B(9) .327 ACCDUM B(10) -1.237 #ACCDUM2 B(11) 1.389 TWAIT1 B(12) -.620 TWAIT2 B(13) -.0428 TXFER B(14) -.0604 PHWBM 9.8-4 9.8.3 AGGREGATE MODEL FORMULATION U(A,I,J,AO) :: = C(1)+B(1)*(($INCA(AC)-$750*$PHHA(AC))/($INCOME -$750*$PHH))*DINC(I)+B(2)*(WTERMP(I)+WTERMA(J)) +B(3)*ATIMET(I,J)+B(4)*($INCOME/$INCA(AO)) *ACOSTI(I,J)+B(5)*CBD(J)+B(6)*($WHH/$WHHA(AO)) *$MAO(AO)/WHH(I)+B(15)*$HBWMAMCF(1) U(S,I,J,AO) :: = C(2)+B(1)*(($INCA(AO)-$750*$PHHA(AO))/($INCOME -$750*$PHH))*DINC(I)+B(2)*(WTERMP(I)+WTERMA(J)) +B(3)*STIMET(I,J)+B(4)*($INCOME/$INCA(AO)) *SCOSTI(I,J)+B(7)*CBD(J)+B(8)*($WHH/$WHHA(AO)) *$MAO(AO)/WHH(I)+B(9)*($WHHA(AO)/$WHH)*WHH(I) +B(15)*$HBWMSMCF(1) U(T,I,J,AO) :: = B(2)*TWALK(I,J)+B(3)*TTIMET(I,J)+B(4)*$INCOME /$INCA(AO)*TCOSTI(I,J)+B(10)*ACCUM(I,J) +B(11)*($WHH/$WHHA(AO))*$MAO(AO)*ACCDUM2(I,J) +B(12)*TWAIT1(I,J)+B(13)*TWAIT2(I.J)+B(14) *TXFER(I,J) WHERE: U( ,I,J,AO) IS THE UTILITY OF CHOOSING MODE M (A = DRIVE ALONE, S = SHARED-RIDE, T = TRANSIT) TO PRIMARY WORKERS RESIDING IN ZONE I, WORKING IN ZONE J, AND BELONGING TO AUTO OWNERSHIP LEVEL AO. DINC(I) IS THE AVERAGE ANNUAL DISPOSABLE INCOME CALCULATED FOR HOUSEHOLDS IN ZONE I (1965 DOLLARS), COMPUTED AS: DINC(I) :: = INCOME(I)-$750*PHH(I) WHERE: INCOME(I) IS THE AVERAGE ANNUAL HOUSEHOLD INCOME IN ZONE I (1965 DOLLARS). PHH(I) IS THE AVERAGE HOUSEHOLD SIZE IN ZONE I (PERSONS). WTERMP(I) IS THE 1-=WAY PEAK PERIOD AUTO PRODUCTION WALK TERMINAL TIME FOR ZONE I (MINUTES). WTERMA(I) IS THE 1-WAY PEAK PERIOD AUTO ATTRACTION WALK TERMINAL TIME FOR ZONE I (MINUTES). TIMET(I,J) IS THE 1-WAY PEAK PERIOD IN-VEHICLE TRAVEL TIME USING MODE M FROM ZONE J (MINUTES). CBD(J) IS A CBD STATUS SWITCH FOR WORK ZONE J: = 1 IF J IS IN THE CENTRAL BUSINESS DISTRICT; = 0 OTHERWISE. WHH(I) IS THE AVERAGE NUMBER OF WORKERS PER HOUSEHOLD IN ZONE I (WORKERS). TWALK(I,J) IS THE 1-WAY PEAK PERIOD WALK TIME FOR TRIPS USING TRANSIT FROM ZONE I TO ZONE J (MINUTES). ACCCUM(I,J) IS AN AUTO ACCESS DUMMY FOR THE I TO J WORK TRIP: = 1 IF AUTO ACCESS TO TRANSIT IS REQUIRED, = 0 OTHERWISE; PHBWM 9.8-5 WHERE AVAILABILITY OF WALK ACCESS IS DETERMINED ON A ZONAL BASIS. ACCDUM2(I,J) IS = $MAO(AO)/WHH(I) IF WALK ACCESS TO TRANSIT IS NOT AVAILABLE TO PRIMARY WORKERS RESIDING/WORKING IN ZONE I/J, AND IS = 0 OTHERWISE WHERE AVAILABILITY OF WALK ACCESS IS DETERMINED IN A ZONAL BASIS (AUTOS/WORKER). TWAIT1(I,J) IS THE 1-WAY PEAK PERIOD INITIAL WAIT TIME UP TO 4 MINUTES FOR TRANSIT TRIPS FROM i TO J (MINUTES). TWAIT2(I,J) IS THE REMAINDER, IF ANY, OF THE 1-WAY PEAK PERIOD INITIAL WAIT TIME FOR TRANSIT TRIPS FROM I TO J WHICH IS OVER 4 MINUTES (MINUTES). TXFER(I,J) IS THE 1-WAY PEAK PERIOD TRANSFER WAIT TIME FOR TRANSIT TRIPS FROM ZONE I TO ZONE J (MINUTES). C(1), C(2), B(1)-B(15) ARE MODEL COEFFICIENTS: VARIABLE COEFF VALUE ------- C(1) -2.909 ------- C(2) -4.519 DINC B(1) 1.817E - 5 TWALK, B(2) -.0670 TIMET B(3) -.0244 # COSTI B(4) -4286 CBD B(5) -1.067 #CARS/#WHH B(6) 1.958 CBD B(7) -.347 #CARS/#WHH B(8) 1.763 WHH B(9) .327 ACCDUM B(10) (1) #ACCDUM2 B(11) 1.389 TWAIT1 B(12) -.620 TWAIT2 B(13) -.0428 TXFER B(14) -.0604 $HBWM MCF B(15) 1.000 (1) -160 FOR AO=0, -1.237 FOR AO=1 AND AO=2 $MAO(AO) IS THE MEAN REGIONAL AUTO OWNERSHIP LEVEL BY AO LEVEL (AUTOS); BY DEFINITION, $MAO(0)=0 AND $MAO(1)=1; 2.236 RECOMMENDED FOR $MAO(2+). $INCOME IS THE MEAN REGIONAL GROSS HOUSEHOLD INCOME (1965 DOLLARS); 10,262 RECOMMENDED. $INCA(AO) IS THE MEAN REGIONAL GROSS INCOME FOR HOUSEHOLDS IN AUTO OWNERSHIP LEVEL AO (1965 DOLLARS); 6353, 8670 AND 12,453 RECOMMENDED FOR AO=0, AO=1 AND AO=2+ RESPECTIVELY. $PHH IS THE MEAN REGIONAL HOUSEHOLD SIZE (PERSONS); 3.81 RECOMMENDED. $PHHA(AO) IS THE MEAN REGIONAL SIZE OF HOUSEHOLDS IN AUTO OWNERSHIP LEVEL AO (PERSONS); 3.96, 3.53 AND 4.22 RECOMMENDED FOR AO=0, AO=1 AND AO=2+ RESPECTIVELY. $WHH IS THE MEAN REGIONAL NUMBER OF WORKERS PER HOUSEHOLD (WORKERS); 1.71 RECOMMENDED. $WHHA(AO) IS THE MEAN REGIONAL NUMBER OF WORKERS IN PHBWM 9.8-6 HOUSEHOLDS IN AUTO OWNERSHIP LEVEL AO (WORKERS); 1.10, 1.53 AND 1.99 RECOMMENDED FOR AO=0, AO=1 AND AO=2+ RESPECTIVELY. $HBWM MCF(I) IS A MODE M ADJUSTMENT FACTOR FOR ZONE I; DERIVED FROM DISTRICT VALUES GIVEN IN TABLE 9.8-1 FOR SAN FRANCISCO BAY AREA APPLICATIONS ONLY; OTHERWISE 0.0 RECOMMENDED FOR ALL ZONES. PHBWM 9.8-7 DIST HBWMSMCF HBWMAMCF DIST HBWMSMCF HBWMAMCF 1 1.538 1.552 16 -.192 -.496 2 -.087 .080 17 -.406 -.888 3 -.015 .196 18 -.020 -.185 4 .105 .144 19 -.020 -.185 5 .161 .386 20 1.140 1.325 6 -.640 -.483 21 1.140 1.325 7 -.738 -.320 22 -1.470 -.701 8 -.015 .208 23 -1.470 -.701 9 .673 .706 24 .168 -.038 10 .790 .821 25 .168 -.038 11 .361 .140 26 .449 .323 12 .790 .821 27 .449 .323 13 .258 .117 28 .449 .323 14 .258 .117 29 .610 .762 15 .258 .117 30 .707 .348 TABLE 9.8-1 AGGREGATE VALIDATION ADJUSTMENTS $HBWMSMCF AND $HBWMAMCF DISTRICT VALUES PHBWM AND SHBWM MODAL ADJUSTMENT FACTORS SHBWG 9.9-1 9.9 SECONDARY WORKER HOME BASED WORK TRIP GENERATION MODEL SHBWG 9.9.1 GENERAL DESCRIPTION IN ITS DISAGGREGATE FORM, THE SHBWG MODEL PREDICTS THE NUMBER OF DAILY HOME BASED WORK TRIPS MADE BY A SECONDARY WORKER. IN ITS AGGREGATE FORM, THE MODEL PREDICTS THE VOLUME OF SECONDARY WORKER HOME BASED WORK TRIPS PRODUCED BY A GROUP OF SECONDARY WORKERS. IN MTCFCAST, THESE GROUPS ARE DEFINED BY RESIDENCE ZONE AND AUTO OWNERSHIP LEVEL (0, 1, 2+). THE MODEL REQUIRES, IN ITS AGGREGATE FORM, THE NUMBER OF SECONDARY WORKERS BY ZONE AND AUTO OWNERSHIP LEVEL (COMPUTED FROM OUTPUTS OF THE NWHH AND WHHAO MODELS), AS WELL AS INFORMATION ON MEAN HOUSEHOLD INCOMES AND SIZES BY ZONE AND AUTO OWNERSHIP LEVEL (COMPUTED FROM OUTPUTS OF THE WHHAO MODEL). THE MODEL RESULTS, THE SECONDARY WORKER TRIP PRODUCTION TOTALS, ARE ALLOCATED TO DESTINATIONS BY THE SHBQD MODEL. AS FOR PRIMARY WORKERS, VARIATIONS IN DAILY HOME BASED WORK TRIPS MADE BY SECONDARY WORKERS ARE DUE TO VACATIONS AND DAYS OFF EITHER ELECTED OF REQUIRED BY THE EMPLOYER), GOING HOME FOR LUNCH, COMPLEX TRIPS TO WORK THAT INCLUDE STOPS FOR NON-WORK PURPOSES, AND WALKING TO WORK. THE MODELS INDEPENDENT VARIABLES, HOUSEHOLD INCOME AND HOUSEHOLD SIZE ARE PROXIES FOR ONE OR MORE OF THESE SOURCES OF VARIATION. TRIPS PER WORKER INCREASE WITH INCREASING INCOME, AND DECREASE WITH INCREASING HOUSEHOLD SIZE. 9.9.2 DISAGGREGATE MODEL FORMULATION] THIS LINEAR MODEL HAS THE FOLLOWING EQUATION: #SHBWGV(Q) :: = MAX($MINGS,(C+B(1)8LN(#INCOME(H(Q))/1000) +B(2)*#PHH(H(Q)))) WHERE: #SHBWGV(Q) IS THE PREDICTED DAILY HOME BASED WORK TRIP PRODUCTION RATE FOR SECONDARY WORKER Q (TRIPS). #INCOME(H(Q)) IS THE ANNUAL GROSS INCOME OF Q'S HOUSEHOLD, H(Q) (1965 DOLLARS). #PPH(H(Q)) IS THE NUMBER OF PERSONS IN Q'S HOUSEHOLD, H(Q) (PERSONS). C,B(1),B(2) ARE MODEL COEFFICIENTS VARIABLE COEFF VALUE ------- C 1.246 LN(#INCOME/1000) B(1) .199 #PHH B(2) -.111 $MINGS IS A LOWER BOUND IN THE TRIP GENERATION RATE (TRIPS/WORKER); .50 RECOMMENDED. 9.9-2 9.9.3 AGGREGATE MODEL FORMULATION SHBWGVA(I,AO) :: = SWORKA(I,AO)*SHBWGF(I)*$SHBWGFR*MAX($MINGS, (C+B(1)*LN(WINCA(I,AO)/1000)+B(2)*PHHA(I,AO))) WHERE: SHBWGVA(I,AO) IS THE NUMBER OF SECONDARY WORKER HOME BASED WORK TRIPS PRODUCED IN ZONE I BY HOUSEHOLDS IN AUTO OWNERSHIP LEVEL AO. SWORK(I,AO) IS THE NUMBER IF SECONDARY WORKERS IN HOUSEHOLDS IN ZONE I HAVING AUTO OWNERSHIP LEVEL AO WINCA(IAO) IS THE MEAN GROSS ANNUAL INCOME OF HOUSEHOLDS WITH WORKERS IN ZONE I HAVING AUTO OWNERSHIP LEVEL AO. PHHA(I,AO) IS THE MEAN HOUSEHOLD SIZE FOR HOUSEHOLDS IN ZONE I HAVING AUTO OWNERSHIP LEVEL AO. C,B(1),B(2) ARE MODEL COEFFICIENTS VARIABLE COEFF VALUE ------- C 1.246 LN(WINCA/1000) B(1) .199 PHHA B(2) -.111 $MINGS IS A LOWER BOUND IN THE TRIP GENERATION RATE (TRIPS/WORKER); .50 RECOMMENDED. $SHBWGFR IS A REGIONAL ADJUSTMENT FACTOR; 1.0269 RECOMMENDED FOR SAN FRANCISCO BAY AREA APPLICATIONS ONLY; 1.0 RECOMMENDED OTHERWISE. $SHBWGF(1) IS A ZONAL ADJUSTMENT FACTOR FOR ZONE I; FACTORS DERIVED FROM THE DISTRICT VALUES GIVEN IN TABLE 9.5-1 ARE RECOMMENDED FOR SAN FRANCISCO BAY AREA APPLICATIONS ONLY; OTHERWISE 1.0 RECOMMENDED FOR ALL ZONES. 9.10-1 9.10 SECONDARY WORKER HOME BASED WORK TRIP ATTRACTION MODEL (SHBWA) 9.10.1 GENERAL DESCRIPTION THE SHBWA MODEL PREDICTS THE NUMBER OF SECONDARY WORKER HOME BASED WORK TRIPS ATTRACTED TO ZONE I. THIS IS AN AGGREGATE MODEL ONLY; NO DISAGGREGATE FORM EXISTS. THE SHBWA MODEL IS INDEPENDENT OF THE OUTPUTS OF ALL OTHER MODELS. ITS RESULTS ARE USED AS AN ATTRACTION-ZONE SIZE VARIABLE IN THE SHBWD MODEL. VARIATIONS IN TRIP ATTRACTIONS FROM THE NOMINAL VALUE OF TWO TRIPS PER EMPLOYEE PER DAY OCCUR FOR THE SAME REASONS THEY DO FOR TRIP PRODUCTIONS (SEE 9.5.1). IN ADDITION, THE AGRICULTURAL AND WHOLESALE TRADE EMPLOYMENT GROUPS WERE FOUND TO CONSISTENTLY ATTRACT THE SMALLEST NUMBER OF SECONDARY WORKER TRIPS PER JOB. THE INDEPENDENT VARIABLES ARE THE FRACTION OF ZONAL EMPLOYMENT THAT IS AGRICULTURAL AND WHOLESALE TRADE. WORKERS LIVING IN THE ZONE DIVIDED BY ZONAL JOBS, RESIDENTIAL ACREAGE AS A FRACTION OF TOTAL ACREAGE, AUTO TERMINAL TIME FOR TRIPS ATTRACTED TO THE ZONE AND WORK FORCE DENSITY. THE TWO FINAL VARIABLES ARE PROXIES THAT CAPTURE THE OBSERVED TENDENCY FOR SECONDARY WORKERS TO WORK IN AREAS THAT ARE LESS DENSELY DEVELOPED. 9.10.2 AGGREGATE MODEL FORMULATION THIS LINEAR MODEL HAS THE FOLLOWING EQUATION: SHBWA(I) :: = MAX ($SHBWAFR*$HBWAF(I)*TOTEMP(I)*C+B(1)*WORKERS(I) /TOTEMP(I)+B(2)*(AGREMP(I)+TRDEMP(I))/TOTEMP(I) +B(3)*RACRES(I)/ACRES(I)+B(4)*ATERMA(I) +B(5)*WORKERS(I)/ACRES(I)),0) WHERE: SHBWA(I) IS THE TOTAL NUMBER OF SECONDARY HOME BASED WORK TRIPS ATTRACTED TO ZONE I (TRIPS). TOTEMP(I) IS THE TOTAL EMPLOYMENT OF ZONE I EMPLOYEES). WORKERS(I) IS THE NUMBER OF WORKERS RESIDING IN ZONE I (PERSONS). AGREMP(I) IS THE GROSS AGRICULTURAL EMPLOYMENT OF ZONE I (EMPLOYEES). TRDEMP(I) IS THE GROSS TRADE EMPLOYMENT OF ZONE I (EMPLOYEES). RACRES(I) IS THE RESIDENTIAL LAND-USE AREA OF ZONE I (ACRES). ACRES(I) IS THE TOTAL ACREAGE OF ZONE I (ACRES). ATERMA(I) IS THE PEAK PERIOD AUTO ACCESS TERMINAL TIME FOR ATTRACTION ZONE I (MINUTES). C,B(1)-B(5) ARE MODEL COEFFICIENTS: 9.10-2 VARIABLE COEFF VALUE ------- C .4152 WORKERS/TOTEMP B(1) .03482 (AGREMP+TRDEMP)/TOTEMP B(2) -.2610 RACRES/ACRES B(3) .1613 ATERMA B(4) -.0490 WORKERS/ACRES B(5) -.00586 $SHBWAFR IS A REGIONAL ADJUSTMENT FACTOR; 1.043 RECOMMENDED FOR SAN FRANCISCO BAY AREA APPLICATIONS ONLY; 1.0 OTHERWISE. $HBWAF(I) IS A ZONAL ADJUSTMENT FACTOR FOR ZONE I; VALUES DERIVED FROM THE DISTRICT VALUES GIVEN IN TABLE 9.6- 1 ARE RECOMMENDED FOR SAN FRANCISCO BAY AREA APPLICATIONS ONLY; OTHERWISE 1.0 RECOMMENDED FOR ALL ZONES. SHBWD 9.11-1 9.11 SECONDARY WORKER HOME BASED WORK TRIP DISTRIBUTION MODEL (SHBWD) 9.11.1 GENERAL DESCRIPTION IN ITS DISAGGREGATE FORM THE SHBWD MODEL PREDICTS THE PROBABILITY OF A SECONDARY WORKER MAKING A HOME BASED WORK TRIP TO A PARTICULAR ZONE. IN ITS AGGREGATE FORM THE MODEL PREDICTS THE FRACTION OF A GROUP OF HOME BASED SECONDARY WORK TRIPS ATTRACTED TO EACH DESTINATION ZONE AND AUTO OWNERSHIP LEVEL (0, 1, 2+). THE MODEL REQUIRES INPUTS FROM THE SHBWA MODEL AND THE SHBWM MODEL. MOREOVER, TO OBTAIN DISTRIBUTED TRIP VOLUMES FROM THE AGGREGATE MODEL, IT MUST BE APPLIED TO ZONAL SHBW TRIP PRODUCTION VOLUMES STRATIFIED BY AUTO OWNERSHIP LEVEL, OBTAINED FROM THE SHBWG MODEL. THE MODEL RESULTS ARE USED TO PROVIDE SECONDARY WORKER TRIPS BY ZONE PAIR, WHICH ARE ALLOCATED TO MODES BY THE SHBWM MODEL. THE MODEL HAS THREE INDEPENDENT VARIABLES: (1) THE ACCESSIBILITY OF EACH ALTERNATIVE ATTRACTION ZONE. THIS VARIABLE INCORPORATED THE CONTRIBUTIONS TO ATTRACTION ZONE ACCESSIBILITY OF EACH AVAILABLE MODE, AND ALSO COLLAPSES MOST OF THE IMPEDENCE TERMS AFFECTING ATTRACTION ZONE CHOICE INTO A SINGLE VARIABLE. THE PROBABILITY OF CHOOSING A GIVEN ATTRACTION ZONE INCREASES AS ITS ACCESSIBILITY INCREASES. (2) THE TRIP LENGTH (DISTANCE) CORRECTION VARIABLE TO EACH ATTRACTION ZONE; THE AMOUNT BY WHICH THE PRODUCTION TO ATTRACTION ZONE DISTANCE EXCEEDS A THRESHOLD VALUE, THREE MILES. THIS VARIABLE REPRESENTS THE DECREASING KNOWLEDGE OF POTENTIAL WORK LOCATIONS AS THE DISTANCE FROM HOME INCREASES THE PROBABILITY IF CHOOSING A GIVEN ATTRACTION ZONE DECREASES AS TRIP DISTANCE INCREASES FOR ALL DISTANCES GREATER THAN THREE MILES. (3) THE NUMBER IF SECONDARY WORKER HBW TRIP ATTRACTIONS IN THE ATTRACTION ZONE, AS PREDICTED BY THE SHBWA MODEL. THE PROBABILITY OF CHOOSING A GIVEN ATTRACTION ZONE INCREASES AS ITS TRIP ATTRACTIONS INCREASE. 9.11.2 DISAGGREGATE MODEL FORMULATION THIS LOGIT MODEL HAS THE FOLLOWING UTILITY EQUATION: U(J,q) :: = B(1)*LNSHBWA(J)+B(2)*#SHBWM(J,G)+B(3)*PDIST3(1(Q),J) WHERE: U(J,Q) IS THE UTILITY OF ATTRACTION ZONE J FOR SECONDARY WORKER Q RESIDING IN ZONE I. LNSHBWA(J) IS THE NATURAL LOGARITHM OF THE TOTAL NUMBER OF SECONDARY WORKER HOME BASED WORK TRIPS SHBWD 9.11-2 ATTRACTED TO ZONE J (SHBWA IN TRIPS). #SHBWML(J,Q) IS THE NATURAL LOGARITHM OF THE DENOMINATOR OF THE DISAGGREGATE SHBWM MODEL FOR DESTINATION ZONE J AND WORKER Q. PDIST3(I,J) IS A DISTANCE CORRECTION VARIABLE, COMPUTED AS: PDIST3(I(Q),J) :: = MAX((&ADIST(I(Q),J)-3),0) WHERE: &ADIST(I(Q),J) IS THE 1-WAY PEAK PERIOD AUTO DISTANCE FROM Q'S HOME ZONE, I(Q), TO ZONE J (MILES). B(1)-B(3) ARE MODEL COEFFICIENTS: VARIABLE COEFF VALUE LNSHBWA B(1) 1.000 SHBWML B(2) 2.842 PDIST3 B(3) -.040 9.11.3 AGGREGATE MODEL FORMULATION U(I,J,AO) :: = B(1)*LNSHBWA(J)+B(2)*SHBWML(I,J,AO) +B(3)*SHBWDCF(I,J) WHERE: U(I,J,AO) IS THE UTILITY OF ATTRACTION ZONE J FOR SECONDARY WORKERS IN ZONE I BELONGING TO AUTO OWNERSHIP LEVEL AO. LNSHBWA(J) IS THE NATURAL LOGARITHM OF THE TOTAL NUMBER OF SECONDARY WORKER HOME BASED WORK TRIPS ATTRACTED TO ZONE J (SHBWA IN TRIPS). SHBWML(I,J,AO) IS THE NATURAL LOGARITHM IF THE DENOMINATOR OF THE SHBWM MODEL FOR TRIPS FROM I TO J BY SECONDARY WORKERS IN AUTO OWNERSHIP LEVEL AO. SHBWDCF(I,J) IS A DISTANCE CORRECTION FACTOR, COMPUTED AS: SHBWDCFi,J) :: = MIN($SHBWDC(I)*(&ADIST(I,J)-3),0) WHERE: &ADIST(I,J) IS THE 1-WAY PEAK PERIOD AUTO DISTANCE FROM ZONE I TO ZONE J (MILES). $SHBWDC(I) IS THE COEFFICIENT IF THE DISTANCE CORRECTION VARIABLE FOR ZONE I; DERIVED FROM THE DISTRICT VALUES GIVEN IN TABLE 9.11-1 FOR SAN FRANCISCO BAY AREA APPLICATIONS ONLY; OTHERWISE -.043 RECOMMENDED FOR ALL ZONES. B(1)-B(3) ARE MODEL COEFFICIENTS: SHBWD 9.11-3 VARIABLE COEFF VALUE LNSHBWA B(1) 1.000 #SHBWML B(2) 2.842 SHBWDCF B(3) 1.000 SHBWD 9.11-4 DISTRICT VALUE DISTRICT VALUE DISTRICT VALUE 1 0.0 11 -.033 21 -.066 2 -.001 12 0.0 22 -.052 3 -.034 13 -.096 23 -.556 4 -.010 14 -.079 24 -.055 5 -.045 15 -.045 25 -.050 6 -.044 16 -.026 26 -.027 7 -.012 17 -.040 27 0.0 8 -.030 18 -.078 28 0.0 9 -.005 19 -.106 29 -.046 10 -.023 20 -.046 30 -.033 TABLE 9.11-1 AHBWD AGGREGATE VALIDATION ADJUSTMENTS $SHBWDC DISTRICT VALUES COEFFICIENTS OF DISTANCE (TRIP LENGTH) CORRECTION VARIABLE SHBWM 9.12-1 9.12 SECONDARY WORKER HOME BASED WORK TROP MODE CHOICE MODEL (SHBWM) 9.12.1 GENERAL DESCRIPTION IN ITS DISAGGREGATE FORM, THE SHBWM MODEL PREDICTS THE PROBABILITIES OF A SECONDARY WORKER CHOOSING EACH OF THREE MODES (DRIVE ALONE, SHARED-RIDE, AND TRANSIT)FOR HIS WORK TRIP, GIVEN HIS WORK LOCATION. IN ITS AGGREGATE FORM, THE MODEL PREDICTS, FOR A GROUP ON SECONDARY WORKERS LIVING IN A PARTICULAR ZONE AND WORKING IN A PARTICULAR ZONE, THE FRACTIONS THAT WILL USE EACH IF THE THREE MODES IN THEIR HOME BASED WORK TRIPS. IN MTCFCAST, THE GROUPS ARE BASED ON SEGMENTATION BY AUTO OWNERSHIP LEVEL (0, 1, 2+). NONE OF THE INDEPENDENT VARIABLES IS DIRECTLY DEPENDENT IN ANY OTHER MODEL IN THE MTC SYSTEM. HOWEVER, THE SHARED-RIDE COST VARIABLES ARE COMPUTED USING AUTO OCCUPANCIES PRODUCED BY THE SROCC MODEL. MOREOVER, TO OBTAIN MODAL TRIP TABLES FROM THE AGGREGATE MODEL, IT MUST BE APPLIED TO SECONDARY WORKER TRIP TABLES OBTAINED FROM THE SHBWD MODEL. THE MODEL RESULTS, IN THE FORM OF PRODUCTION/ATTRACTION WORK TRIPS BY MODE, CAN BE CONVERTED TO ORIGIN/DESTINATION WORK TRIPS USING SPLITTING FACTORS GENERATED BY THE NHB MODELS. THE PRODUCTION/ATTRACTION WORK TRIPS, SUMMED BY ATTRACTION ZONE, ALSO SERVE AS A PORTION OF THE POPULATION OF POTENTIAL NHB TRIPS. THE SHARED-RIDE TRIPS CAN BE CONVERTED FROM PERSON TRIPS TO VEHICLE TRIPS USING THE RESULTS OF THE SROCC MODEL. IN ADDITION, ACCESSIBILITY VARIABLES FROM THIS MODEL ARE USED IN THE SHBWD MODEL. THIS MODEL AND THE PHBWM MODEL )SEE 9.8) WERE ESTIMATED JOINTLY, WITH THE ONLY DIFFERENCE BEING THE BREADWINNER DUMMY VARIABLE DEFINED TO BE 1 FOR PRIMARY WORKERS, AND 0 FOR SECONDARY WORKERS. THEREFORE, THE ONLY DIFFERENCE IN THE TWO MODELS, AS ESTIMATED, CAN BE EXPRESSED AS A DIFFERENCE IN THE DRIVE-ALONE MODE CONSTANT, AND IN THE SUBSTITUTION OF SECONDARY WORKERS FOR PRIMARY WORKERS. THEREFORE, ONLY THIS DIFFERENCE IN CONSTANT IS PRESENTED HERE; DETAILS OF THE MODEL'S INDEPENDENT VARIABLES AND MODEL FORMULATIONS ARE GIVEN IN 9.8. IN THE SHBWM MODEL, THE MODE-SPECIFIC CONSTANTS, C(1) AND C(2), HAVE THE FOLLOWING VALUES IN THE DISAGGREGATE AND AGGREGATE MODEL FORMULATIONS: DRIVE ALONE SHARED-RIDE MODEL FORMULATION C(1) C(2) DISAGGREGATE -2.511 -3.473 AGGREGATE -3.586 -4.519 HBSHG 9.13-1 9.13 HOME BASED SHOP TRIP GENERATION MODEL (HBSHG) 9.12.1 GENERAL DESCRIPTION IN ITS DISAGGREGATE FORM, THE HBSHG MODEL PREDICTS THE NUMBER OF DAILY HOME BASED SHOPPING TRIPS MADE BY MEMBERS OF A HOUSEHOLD. IN ITS AGGREGATE FORM, IT PREDICTS THE NUMBER OF HOME BASED SHOPPING TRIPS PRODUCED DAILY BY A GROUP OF HOUSEHOLDS. IN MTCFCAST, THESE GROUPS ARE DEFINED BY RESIDENCE ZONE AND AUTO OWNERSHIP LEVEL (0, 1, 2+). THE HBSHG MODEL REQUIRES INPUTS GENERATED BY THE HBSHDM MODEL. IN ADDITION, THE AGGREGATE MODEL REQUIRES SEVERAL AUTO OWNERSHIP- SPECIFIC INDEPENDENT VARIABLES, WHICH ARE OBTAINED BY PROCESSING THE OUTPUTS OF THE NWHHAP AND WHHAO MODELS. THE MODEL RESULTS, HBSH TRIP PRODUCTION TOTALS, ARE ALLOCATED TO ATTRACTION ZONES AND MODES BY THE HBSHDM MODEL. THE HBSHG MODEL HAS A NON-LINEAR FORM THAT PREVENTS THE PREDICTION OF NEGATIVE TRIP VALUES REGARDLESS OF THE VALUES OF THE INDEPENDENT VARIABLES. THE FOLLOWING INDEPENDENT VARIABLES, AND THEIR EFFECTS ON TRIP GENERATION ARE USED: (1) HOUSEHOLD SIZE AND HOUSEHOLD INCOME, REPRESENTING THE INCREASED DEMAND FOR SHOPPING TRIPS THAT EXISTS IN LARGER, WEALTHIER HOUSEHOLDS. SHOP TRIP GENERATION INCREASES AS EACH OF THESE VARIABLES INCREASES. (2) RETAIL AND SERVICE EMPLOYMENT DENSITY IN THE RESIDENCE ZONE, REPRESENTING THE PROBABILITY OF SATISFYING SHOP TRIP DEMANDS BY MAKING WALK TRIPS RATHER THAN THE VEHICULAR TRIPS PREDICTED BY THE MODEL. SHOP TRIP GENERATION DECREASES AS THIS DENSITY INCREASES. (3) THE DENOMINATOR OF THE HBSH DESTINATION/MODE CHOICE MODEL, REPRESENTING THE ACCESSIBILITY IF THE RESIDENCE ZONE TO SHOPPING ATTRACTION ZONES. SHOP TRIP GENERATION INCREASES AS ACCESSIBILITY INCREASES. 9.13.2 DISAGGREGATE MODEL FORMULATION THIS NON-LINEAR MODEL HAS THE FOLLOWING EQUATION: #HBSHGV(H) :: = C(1)/(C(2)+EXP(B(1)*#PHH(H)+B(2)*#LSUMSH(H) +B(3)*#INCOME(H)+B(4)*LN(RETSERD(I(H))+1.0))) WHERE: #HBSHGV(H) IS THE DAILY TOTAL OF HOME BASED SHOPPING TRIPS MADE BY MEMBERS OF HOUSEHOLD H (TRIPS). #PHH(H) IS THE HOUSEHOLD SIZE OF HOUSEHOLD H (PERSONS). #LSUMSH(H) IS THE NATURAL LOGARITHM OF THE DENOMINATOR OF THE DISAGGREGATE HBSHDM MODEL FOR HOUSEHOLD H. #INCOME(H) IS THE GROSS ANNUAL INCOME OF HOUSEHOLD H (1965 DOLLARS). HBSG 9.13-2 RETSERD(I(H) IS THE GROSS RETAIL AND SERVICE EMPLOYMENT DENSITY IN HOUSEHOLD H'S ZONE OF RESIDENCE, I(H) (EMPLOYEES/ACRE). C(1), C(2), B(1)-B(4) ARE MODEL COEFFICIENTS. VARIABLE COEFF VALUE ----- C(1) .8149 ----- C(2) .07766 #PHH B(1) - .3417 #LSUMSH B(2) - .0527 #INCOME B(3) -5.151E - 5 LN(RETSERD) B(4) .1146 HBSHG 9.13-3 9.13.3 AGGREGATE MODEL FORMULATION HBSHGVA(I,AO) ;; = HHSA(I,AO)*HBSHGFR*$HBSHGF(1)*C(1)/C(2) +EXP(B(1)*PHHA(I,AO)+B(2)*LSUMSHA(I,AO) +B(3)*INCOMEA(I,AO)+B(4)*LN(RETSERD(I)+1.0))) WHERE: HBSHGVA(I,AO) IS THE VOLUME OF DAILY HOME BASED SHOPPING TRIPS PRODUCED IN ZONE I BY HOUSEHOLDS IN AUTO OWNERSHIP LEVEL AO (TRIPS). HHSA(I,AO) IS THE NUMBER OF HOUSEHOLDS IN ZONE I, AUTO OWNERSHIP LEVEL AO (HOUSEHOLDS). PHHA(I,AO) IS THE MEAN HOUSEHOLD SIZE OF HOUSEHOLDS IN ZONE I, AUTO OWNERSHIP LEVEL AO (PERSONS/HOUSEHOLD). LSUMSHA(I,AO) IS THE NATURAL LOGARITHM OF THE DENOMINATOR OF THE AGGREGATE HBSHDM MODEL FOR ZONE I, AUTO OWNERSHIP LEVEL AO; IT IS A MEASURE OF SHOPPING ACCESSIBILITY AND OPPORTUNITY. INCOMEA(I,AO) IS THE MEAN GROSS ANNUAL INCOME FOR HOUSEHOLDS IN ZONE I, AUTO OWNERSHIP LEVEL AO (1965 DOLLARS). RETSERD(I) IS THE GROSS RETAIL AND SERVICE EMPLOYMENT DENSITY IN ZONE I (EMPLOYEES/ACRE). C(1), C(2), B(1)-B(4) ARE MODEL COEFFICIENTS: VARIABLE COEFF VALUE ----- C(1) .8608 ----- C(2) .07766 PHH B(1) - .3417 LSUMSHA B(2) - .0527 INCOMEA B(3) -5.151E - 5 LN(RETSERD) B(4) .4934 $HBSHGFR IS A REGIONAL ADJUSTMENT FACTOR; 1.0 RECOMMENDED. $HBSHGF(I) IS A ZONAL ADJUSTMENT FACTOR FOR ZONE I; FACTORS DERIVED FROM THE DISTRICT VALUES GIVEN IN TABLE 9.13-1 ARE RECOMMENDED FOR SAN FRANCISCO BAY AREA APPLICATIONS ONLY; OTHERWISE 1.0 RECOMMENDED FOR ALL ZONES. 9.13-4 DIST VALUE DIST VALUE DIST VALUE 1 .983 11 1.242 21 .962 2 .904 12 1.167 22 .962 3 .660 13 .795 23 .911 4 1.112 14 1.012 24 .902 5 .815 15 1.074 25 .802 6 .979 16 .981 26 .869 7 .938 17 1.078 27 .932 8 1.363 18 1.007 28 1.039 9 1.407 19 .909 29 .915 10 1.216 20 .830 30 .920 TABLE 9.13-1 HBSHG AGGREGATE VALIDATION ADJUSTMENTS $HBSHGF DISTRICT VALUES HBSHD 9.14-1 9.14 HOME BASED SHOP TRIP DESTINATION CHOICE MODEL (HBSHD) IN ITS DISAGGREGATE FORM, THE HBSHD MODEL PREDICTS THE PROBABILITY THAT A HOME BASED SHOPPING TRIPMAKER FROM A HOUSEHOLD OWNING NO AUTOS WILL CHOOSE A PARTICULAR DESTINATION ZONE, GIVEN HIS CHOICE OF MODE. IN ITS AGGREGATE FORM, THE MODEL PREDICTS THE FRACTION OF A GROUP OF HOME BASED SHOPPING TRIPS ATTRACTED TO EACH DESTINATION ZONE. IN MTCFCAST THESE TRIP GROUPS ARE DEFINED BY ORIGIN ZONE AND AUTO OWNERSHIP LEVEL (PREDICTIONS ARE ONLY NEEDED FOR THE AUTO OWNERSHIP = O GROUP IN EACH ORIGIN ZONE). THE HBSHD MODEL IS A SPECIAL FORM OF THE HBSHDM MODEL DESCRIBED IN THE NEXT SECTION. IT IS FORMED BY SUBSTITUTING INTO THE HBSHDM MODEL FOR ALL AUTO OWNERSHIP-RELATED VARIABLES THEIR SPECIFIC VALUES WHEN AO=0, AND BY CONVERTING FROM A MODEL OF JOINT CHOICE OF DESTINATION AND MODE TO A CHOICE OF DESTINATION, GIVEN THE MODE CHOSEN. BECAUSE OF THE CLOSE RELATIONSHIP OF THIS MODEL AND THE HBSHDM MODEL, ITS FORMULATION IS NOT REPEATED IN THIS SECTION. THE MODEL IS FULLY DESCRIBED IN VOLUME 2, PART II, SECTION 14. HBSHDM 9.15-1 9.15 HOME BASED SHOP TRIP DESTINATION/MODE CHOICE MODEL (HBSHDM) 9.15.1 GENERAL DESCRIPTION IN ITS DISAGGREGATE FORM, THE HBSHDM MODEL PREDICTS THE PROBABILITY THAT A HOME BASED SHOPPING TRIPMAKER WILL CHOOSE TO GO TO A PARTICULAR DESTINATION ZONE BY A PARTICULAR MODE, EITHER AUTO OR TRANSIT. IN ITS AGGREGATE FORM THE MODEL PREDICTS, FOR GROUPS OF HOUSEHOLDS, THE FRACTIONS ON HONE BASED SHOPPING TRIPS THAT WILL BE MADE TO EACH ATTRACTION ZONE BY EACH MODE. IN MTCFCAST, THESE GROUPS ARE DEFINED BY PRODUCTION ZONE AND AUTO OWNERSHIP LEVEL (0. 1. 2+). NONE OF THE MODEL'S INDEPENDENT VARIABLES DEPEND DIRECTLY ON ANY OTHER MODEL SYSTEM. HOWEVER, TO OBTAIN MODAL HBSH TRIP TABLES THE AGGREGATE MODEL MUST BE APPLIED TO ZONAL HBSH PRODUCTION VOLUMES STRATIFIED BY AUTO OWNERSHIP LEVEL, USUALLY OBTAINED FROM THE GBSHG MODEL. THE MODEL RESULTS , IN THE FORM OF PRODUCTION/ATTRACTION SHOP TRIPS BY MODE, CAN BE CONVERTED TO ORIGIN/DESTINATION SHOP TRIPS USING SPLITTING FACTORS GENERATED BY THE NHB MODELS. THE PRODUCTION/ATTRACTION SHOP TRIPS, SUMMED BY ATTRACTION ZONE, ALSO SERVE AS A PORTION OF THE POPULATION OF POTENTIAL NHB TRIPS. IN ADDITION, ACCESSIBILITY VARIABLES FROM THIS MODEL ARE USED IN EACH OF THE FOLLOWING MODELS: HBSHG, WHHAO, AND NWHHAO. THE INDEPENDENT VARIABLES OF THIS MODEL FALL INTO THREE GROUPS, DEPENDING IN WHETHER THEY AFFECT ATTRACTION ZONE CHOICE, MODE CHOICE, OR BOTH. THE VARIABLES THAT AFFECT ATTRACTION ZONE CHOICE, AND THEIR EFFECTS ON THIS CHOICE ARE: (1) NET RETAIL EMPLOYMENT DENSITY; RETAIL EMPLOYMENT IN THE ATTRACTION ZONE DIVIDED BY POPULATION-SERVING ACREAGE. SHOP TRIP ATTRACTIONS INCREASE AS THIS DENSITY INCREASES. (2) TOTAL RETAIL EMPLOYMENT IN THE ATTRACTION ZONE. SHOP TRIP ATTRACTIONS INCREASE AS RETAIL EMPLOYMENT INCREASES. (3) THE TRIP LENGTH (DISTANCE) CORRECTION VARIABLE FOR A PAIR OF ZONES; THE AMOUNT (IF ANY) BY WHICH TOTAL TRIP DISTANCE EXCEEDS A THRESHOLD VALUE OF SEVEN MILES. SHOP TRIP ATTRACTIONS DECREASE AS THIS VARIABLE INCREASES. ONE INDEPENDENT VARIABLE, THE NUMBER IF AUTOS OWNED PER HOUSEHOLD MEMBER, AFFECTS ONLY MODE CHOICE. MORE AUTO TRIPS ARE MADE AS THIS VARIABLE INCREASES. THE REMAINING INDEPENDENT VARIABLES AFFECT BOTH MODE AND ATTRACTION ZONE CHOICE: (1) WHETHER OR NOT THE ATTRACTION ZONE IS A CENTRAL BUSINESS DISTRICT. MORE SHOP TRIPS ARE MADE TO CBD ZONES, AND MORE OF THESE TRIPS ARE LIKELY TO BE MADE BY TRANSIT THAN BY AUTO. (2) TOTAL TRAVEL TIME BY MODE. SHOP TRIPS BY MODE DECREASE AS TRAVEL TOME BY MODE INCREASES. HBSHDM 9.15-2 (3) AUTO OUT-OF-POCKET COST. SHOP TRIPS BY AUTO DECREASE AS AUTO COST INCREASES. (4) TRANSIT FARE TIMES HOUSEHOLD SIZE; REFLECTING THE FACT THAT, WHEN CONTRASTED WITH INDIVIDUAL TRIPS, TRIPS BY FAMILY GROUPS ARE RELATIVELY MORE EXPENSIVE FOR THE FAMILY BY TRANSIT THAN BY AUTO. SHOP TRIPS BY TRANSIT DECREASE AS THIS VARIABLE INCREASES. 9.15.2 DISAGGREGATE MODEL FORMULATION THIS LOGIT MODEL HAS THE FOLLOWING UTILITY EQUATIONS: U(J, A, Q) :: = C+B(1)*LNRETEMP(J)+B(2)*RETEMPDP(J) +B(3)*INCATIME(I(Q),J)+B(4)CBD(J) +B(5)*HBOACOST(I(Q),J) +B(6)*#CARS(H(Q))/#PHH(H(Q)) +B(9)*ODIST7(I(Q),J) U(J, T, Q) :: = B(1)*LNRETEMP(J)+B(2)*RETEMPDP(J) +B(3)*INCTTIME(I(Q),J)+B(7)CBD(J) +B(8)*#HHTCOST(J,Q) +B(9)*ODIST7(I(Q),J) WHERE: U(J, , Q) IS THE UTILITY OF SHOPPING DESTINATION ZONE J AND MODE M (A=AUTO, T=TRANSIT), TO SHOPPER Q. LNRETEMP(J) IS THE NATURAL LOGARITHM OF THE RETAIL EMPLOYMENT IN DESTINATION ZONE J (RETAIL EMPLOYMENT IN EMPLOYEES). RETEMPDP(J) IS THE RETAIL EMPLOYMENT DENSITY FOR THE POPULATION-SERVING AREA IN ZONE J (EMPLOYEES/ACRE). INC TIME(I(Q),J) IS THE NATURAL LOGARITHM OF THE 1-WAY OFFPEAK TRAVEL TIME BY MODE M FROM Q'S RESIDENCE ZONE, I(Q), TO ZONE J, TIMES THE 1965 REGIONAL GROSS ANNUAL HOUSEHOLD INCOME (1965 DOLLARS, TIME IN MINUTES). CBD(J) IS =1 IF DESTINATION ZONE IS IN THE CBD, AND IS =0 IF J IS NOT IN THE CBD. HBOACOST(I(Q),J) IS THE TOTAL 1-WAY OFFPEAK AUTO COST FROM Q'S RESIDENCE ZONE, I(Q), TO ZONE J (1965 CENTS). #CARS(H(Q)) IS THE NUMBER OF AUTOS OWNED BY SHOPPER Q'S HOUSEHOLD, H(Q) (AUTOS). #PHH(H(Q)) IS THE SIZE OF SHOPPER Q'S HOUSEHOLD, H(Q) (PERSONS). #HHTCOST(J,Q) IS THE SIZE OF SHOPPER Q'S HOUSEHOLD TIMES THE 1-WAY OFFPEAK TRANSIT FARE FROM Q'S RESIDENCE ZONE, I(Q), TO ZONE J (1965 PERSON-CENTS). ODIST7(I(Q),J) IS A DISTANCE CORRECTION VARIABLE, COMPUTED AS: ODIST7(I(Q),J) :: = MAX((OADIST(I(Q),J)-DTH),0) HBSHDM 9.15-3 WHERE: OADIST(I(Q),J) IS THE 1-WAY OFFPEAK AUTO DISTANCE FROM Q'S RESIDENCE ZONE, I(Q), TO ZONE J (MILES). DTH IS A THRESHOLD VALUE. C, B(1)-B(9) ARE MODEL COEFFICIENTS: VARIABLE COEFFF VALUE ------ C -.0631 LNRETEMP B(1) 1.000 RETEMPDP B(2) 5.995E-4 INC TIME B(3) -2.020E-4 CBD B(4) .2575 HBOACOST B(5) -.02447 #CARS/#PHH B(6) 5.053 CBD B(7) .8912 #HHTCOST B(8) -.02299 ODIST7 B(9) -0.030 9.15.3 AGGREGATE MODEL FORMULATION U(J,A,I,AO) :: = C+B(1)*LNRETEMP(J)+B(2)*RETEMPDP(J) +B(3)*INCATIME(I,J)+B(4)*CBD(J) +B(5)*HBOACOST(I,J)+B(6)*$MAO(AO)/PHH(I) +B(9)*HBSHDCF(I,J)+B(10)*$HBSHMCF(I) U(J,I,I,AO) :: = C+B(1)*LNRETEMP(J)+B(2)*RETEMPDP(J) +B(3)*INCTTIME(I,J)+B(7)*CBD(J) +B(8)*($PHHA(AO)/$PHH)*HHTCOST(I,J) +B(9)*HBSHDCF(I,J) WHERE: U(J, ,I,AO) IS THE UTILITY OF SHOPPING DESTINATION ZONE J AND MODE M TO SHOPPERS RESIDING IN ZONE I AND BELONGING TO AUTO OWNERSHIP LEVEL AO. LNRETEMP(J) IS THE NATURAL LOGARITHM OF THE RETAIL EMPLOYMENT IN DESTINATION ZONE J (RETAIL EMPLOYMENT IN EMPLOYEES). RETEMPDP(J) IS THE RETAIL EMPLOYMENT DENSITY FOR THE POPULATION-SERVING AREA IN ZONE J (EMPLOYEES/ACRE). INC TIME(I,J) IS THE 1965 REGIONAL AVERAGE GROSS ANNUAL HOUSEHOLD INCOME TIMES THE NATURAL LOGARITHM OF THE 1-WAY OFFPEAK TRAVEL TIME FROM ZONE I TO ZONE J, USING MODE M (1965 DOLLARS, TIME IN MINUTES). CBD(J) IS =1 IF THE DESTINATION ZONE J IS IN THE CBD, AND IS =0 IF J IS NOT IN THE CBD. HBOACOST(I,J) IS THE TOTAL 1-WAY OFFPEAK COST OF AN AUTO TRIP FROM ZONE I TO ZONE J (1965 CENTS). PHH(I) IS THE MEAN HOUSEHOLD SIZE IN ZONE I (PERSONS). HHTCOST(I,J) IS THE MEAN HOUSEHOLD SIZE IN ZONE I TIMES THE 1-WAY OFFPEAK TRANSIT FARE FROM ZONE I TO ZONE J (1965 PERSON-CENTS) HBSHDM 9.15-4 HBSHDCF(I,J) IS A DISTANCE CORRECTION FACTOR FOR THE TRIP FROM ZONE I TO ZONE J, COMPUTED AS: HBSHDCF(I,J) :: = MIN($HBSHDC(I)*(OADIST(I,J)-7),0) WHERE: OADIST(I,J) IS THE 1-WAY OFFPEAK AUTO DISTANCE FROM I TO J (MILES). $HBSHDC(I) IS THE COEFFICIENT IF THE DISTANCE CORRECTION VARIABLE FOR ZONE I; VALUES DERIVED FROM THE DISTRICT FACTORS GIVEN IN TABLE 9.15-1 ARE RECOMMENDED FOR SAN FRANCISCO BAY AREA APPLICATIONS ONLY, OTHERWISE, -.90 RECOMMENDED FOR ALL ZONES. C,B(1)-B(10) ARE ALL MODEL COEFFICIENTS: VARIABLE COEFFF VALUE ------ C -1.086 LNRETEMP B(1) 1.000 RETEMPDP B(2) 5.995E-4 INC TIME B(3) -2.020E-4 CBD B(4) 0.0 HBOACOST B(5) -.02447 $MAO/PHH B(6) 5.053 CBD B(7) .0412 HHTCOST B(8) -.02299 HBSHDCF B(9) 1.000 $HBSHMCF B(10) 1.000 $MAO(AO) IS THE MEAN REGIONAL HOUSEHOLD AUTO OWNERSHIP FOR HOUSEHOLDS IN AUTO OWNERSHIP LEVEL AO (AUTOS); BY DEFINITION, $MAO(O)=0- AND $MAO(1)=1; 2.236 RECOMMENDED FOR $MAO(2+). $PHH IS THE MEAN REGIONAL HOUSEHOLD SIZE (PERSONS; 3.81 RECOMMENDED. $PHHA(AO) IS THE MEAN REGIONAL SIZE OF HOUSEHOLDS IN AUTO OWNERSHIP LEVEL AO (PERSONS); 2.96, 3.53 AND 4.22 RECOMMENDED FOR AO LEVEL 0,1 AND 2+ RESPECTIVELY. $HBSHMCF(I) IS A MODAL ADJUSTMENT FACTOR FOR ZONE I; FACTORS DERIVED FROM THE DISTRICT FACTORS GIVEN IN TABLE 9.15-2 ARE RECOMMENDED FOR SAN FRANCISCO APPLICATIONS ONLY; OTHERWISE 0.0 RECOMMENDED FOR ALL ZONES. 9.15-5 DIST VALUE DIST VALUE DIST VALUE 1 0.0 11 -.242 21 -.301 2 -.051 12 0.0 22 -.130 3 -.082 13 -.395 23 -.155 4 -.047 14 -.286 24 -.276 5 -.300 15 -.232 25 -.331 6 -.195 16 -.183 26 -.231 7 -.171 17 -.176 27 0.0 8 -.101 18 -.214 28 0.0 9 -.156 19 -.498 29 -.018 10 -.162 20 -.187 30 -.121 TABLE 9.15-1 HBSHDM AGGREGATE VALIDATION ADJUSTMENTS $HBSHDC DISTRICT VALUES COEFFICIENTS OF THE DISTANCE CORRECTION VARIABLE 9.15-6 DIST VALUE DIST VALUE DIST VALUE 1 .625 11 -.613 21 1.267 2 .026 12 1.471 22 .638 3 .037 13 .404 23 .638 4 .606 14 .404 24 1.403 5 -.141 15 .404 25 1.403 6 .309 16 .058 26 .545 7 .855 17 -.417 27 .545 8 1.471 18 -.166 28 .545 9 1.471 19 1.267 29 .799 10 1.471 20 1.267 30 .799 TABLE 9.15-2 HBSHDM AGGREGATE VALIDATION ADJUSTMENTS $HBSHMCF DISTRICT VALUES MODAL ADJUSTMENT FACTORS HBSRG 9.16-1 9.16 HOME BASED SOCIAL-RECREATIONAL TRIP GENERATION MODEL (HBSRG) 9.16.1 GENERAL DESCRIPTION IN ITS DISAGGREGATE FORM, THE HBSRG MODEL PREDICTS THE NUMBER OF DAILY HOME BASED SOCIAL-RECREATIONAL TRIPS MADE BY MEMBERS OF A HOUSEHOLD. IN ITS AGGREGATE FORM IT PREDICTS THE NUMBER OF HOME BASED SOCIAL-RECREATIONAL TRIPS PRODUCED DAILY BY A GROUP OF HOUSEHOLDS. IN MTCFCAST, THESE GROUPS ARE DEFINED BY RESIDENCE AONE AND AUTO OWNERSHIP LEVEL (0, 1, 2+). THE HBSRG MODEL REQUIRES INPUTS GENERATED BY THE HBSRDM MODEL. IN ADDITION, THE AGGREGATE MODEL REQUIRES SEVERAL AUTO OWNERSHIP- SPECIFIC INDEPENDENT VARIABLES WHICH ARE OBTAINED BY PROCESSING THE OUTPUTS OF THE NWHHAO AND WHHAO MODELS. THE MODEL RESULTS, HBSR TRIP PRODUCTION TOTALS, ARE ALLOCATED TO ATTRACTION ZONES AND MODES BY THE HBSRDM MODEL. THE MODEL'S NON-LINEAR FORM PREVENTS THE PREDICTION OF NEGATIVE TRIP VALUES REGARDLESS OF THE VALUES OF THE INDEPENDENT VARIABLES. THE INDEPENDENT VARIABLES AND THEIR EFFECTS IN TRIP GENERATION ARE: (1) HOUSEHOLD INCOME, TOTAL HOUSEHOLD SIZE, AND NUMBER OF NON- WORKERS PER HOUSEHOLD, REPRESENTING THE INCREASES DEMAND FOR SOCIAL-RECREATIONAL TRIPS THAT EXISTS IN LARGER, WEALTHIER HOUSEHOLDS. SOCIAL-RECREATIONAL TRIP GENERATION INCREASES AS EACH OF THESE VARIABLES INCREASES. (2) SERVICE EMPLOYMENT DENSITY IN THE RESIDENCE ZONE, REPRESENTING THE PROBABILITY OF SATISFYING SOCIAL-RECREATIONAL DEMANDS BY MAKING WALK TRIPS RATHER THAN THE VEHICULAR TRIPS PREDICTED BY THE MODEL. SOCIAL-RECREATIONAL TRIP GENERATION DECREASES AS THIS DENSITY INCREASES. (3) THE DENOMINATOR OF THE HBSR DESTINATION/MODE CHOICE MODEL, REPRESENTING THE ACCESSIBILITY OF THE RESIDENCE ZONE TO SOCIAL-RECREATIONAL ATTRACTION ZONES. SOCIAL-RECREATIONAL TRIP GENERATION INCREASES AS THIS ACCESSIBILITY INCREASES. 9.16.2 DISAGGREGATE MODEL FORMULATION THIS NON-LINEAR MODEL HAS THE FOLLOWING EQUATION: #HBSRGV(H) :: = C*(#PHH(H)*B(1))**((#INCOME(H)/1000)**B(2)) *((SEREMP(I(H))/ACRES(I(H))=1.0)**B(3)) *EXP(B(4D)*#NOWORK(H)+B(5)*#LSUMR(H)) WHERE: #HBSRGV(H) IS THE DAILY TOTAL OF SOCIAL/RECREATIONAL TRIPS MADE BY MEMBERS OF HOUSEHOLD H (TRIPS). #PHH(H) IS THE HOUSEHOLD SIZE OF HOUSEHOLD H (PERSONS). HBSRG 9.16-2 #INCOME(H) IS THE GROSS ANNUAL INCOME OF HOUSEHOLD H (1965 DOLLARS). #NOWORK(H) IS THE NUMBER OF NON-WORKERS IN HOUSEHOLD H (PERSONS). #LSUMR(H) IS THE NATURAL LOGARITHM IF THE DENOMINATOR OF THE DISAGGREGATE HBSRDM MODEL FOR HOUSEHOLD H. SEREMP(I(H) IS THE SERVICE EMPLOYMENT IN HOUSEHOLD H'S ZONE OF RESIDENCE, I(H) (EMPLOYEES). ACRES(I(H) IS THE AREA OF HOUSEHOLDS H'S ZONE OF RESIDENCE, I(H) (ACRES). C,B(1)-B(5) ARE MODEL COEFFICIENTS. VARIABLE COEFF/EXP VALUE -------- C .1398 #PHH B(1) .4671 #INCOME B(2) .3963 SEREMP/ACRES B(3) -.32313 #NOWORK B(4) .005055 #LSUMSR B(5) .06785 9.16.3 AGGREGATE MODEL FORMULATION HBSRGVA(I,AO) :: = HHSA(I,AO)*$HBSRGFR*$HBSRGF(I)*C *(PHHA(I,AO)**B(1)*((INCOMEA(I,AO)/1000**B(2)) *((SEREMP(I)/ACRES(I)+1.0**B(3)) *EXP(B(4)*NOWORKA(I,AO)+B(5)*LSUMSRA(I,AO)) WHERE: HBSRGVA(I,AO) IS THE NUMBER OF DAILY HOME BASED SOCIAL/RECREATIONAL TRIPS PRODUCED IN ZONE I BY HOUSEHOLD IN AUTO OWNERSHIP LEVEL AO (TRIPS). HHSA(I,AO) IS THE NUMBER OF HOUSEHOLDS IN ZONE I, AUTO OWNERSHIP LEVEL AO (HOUSEHOLDS). PHHA(I,AO) IS THE MEAN HOUSEHOLD SIZE IF HOUSEHOLDS IN ZONE I, AUTO OWNERSHIP LEVEL AO (PERSONS/HOUSEHOLD). INCOMEA(I,AO) IS THE MEAN GROSS ANNUAL INCOME OF HOUSEHOLDS IN ZONE I, AUTO OWNERSHIP LEVEL AO (1965 DOLLARS). NOWORKA(I,AO) IS THE MEAN NUMBER OF NON-WORKERS IN HOUSEHOLDS IN ZONE I, AUTO OWNERSHIP LEVEL AO (PERSONS). LSUMSRA(I,AO) IS THE NATURAL LOGARITHM OF THE DENOMINATOR OF THE AGGREGATE HBSRDM MODEL FOR AONE I, AUTO OWNERSHIP LEVEL AO; IT IS A MEASURE OF SOCIAL- RECREATIONAL ACCESSIBILITY AND OPPORTUNITY. SEREMP(I) IS THE SERVICE EMPLOYMENT IN ZONE I (EMPLOYEES). ACRES(I) IS THE AREA OF ZONE I (ACRES). C,B(1)-B(5) ARE MODEL COEFFICIENTS: HBSRG 9.16-3 VARIABLE COEFF/EXP VALUE -------- C .1398 PHHA B(1) .4671 INCOMEA B(2) .3963 SEREMP/ACRES B(3) -.32313 NOWORKA B(4) .005055 LSUMSRA B(5) .06785 $HBSRGFR IS A REGIONAL ADJUSTMENT FACTOR; 1.0 RECOMMENDED. $HBSRGF(I) IS A ZONAL ADJUSTMENT FACTOR FOR ZONE I; FACTORS DERIVED FROM THE DISTRICT VALUES GIVEN IN TABLE 9.16-1 ARE RECOMMENDED FOR SAN FRANCISCO BAY AREA APPLICATIONS ONLY; OTHERWISE 1.0 RECOMMENDED FOR ALL ZONES. HBSRG 9.16-4 DIST VALUE DIST VALUE DIST VALUE 1 .906 11 1.406 21 .991 2 .873 12 1.373 22 1.044 3 .650 13 .636 23 1.059 4 .748 14 .901 24 .878 5 .730 15 1.099 25 .798 6 .923 16 1.020 26 .880 7 .798 17 1.136 27 .931 8 1.385 18 .832 28 .995 9 1.522 19 .884 29 .666 10 1.428 20 .750 30 .630 TABLE 9.16-1 HBSRG AGGREGATE VALIDATION ADJUSTMENTS $HBSRGF DISTRICT VALUES HBSRDM 9.17-1 9.17 HOME BASED SOCIAL-RECREATIONAL DESTINATION/MODE CHOICE MODEL (HBSRDM) 9.17.1 GENERAL DESCRIPTION IN ITS DISAGGREGATE FORM, THE HBSRDM MODEL PREDICTS THE PROBABILITY THAT A HOME BASED SOCIAL-RECREATIONAL TRIPMAKER WILL CHOOSE TO GO TO A PARTICULAR DESTINATION ZONE BY A PARTICULAR MODE, EITHER AUTO OR TRANSIT. IN ITS AGGREGATE FORM, THE MODEL PREDICTS, FOR GROUPS OF HOUSEHOLDS, THE FRACTIONS OF HOME BASED SOCIAL-RECREATIONAL TRIPS THAT WILL BE MADE TO EACH DESTINATION ZONE BY EACH MODE. IN MTCFCAST, THESE GROUPS ARE DEFINED BY ORIGIN ZONE AND AUTO OWNERSHIP LEVEL (0, 1, 2+). NONE OF THE MODEL'S INDEPENDENT VARIABLE DEPEND DIRECTLY IN ANY OTHER MODEL IN THE MTC MODEL SYSTEM. HOWEVER, TO OBTAIN MODAL HBSR TRIP TABLES, THE AGGREGATE MODEL MUST BE APPLIED TO ZONAL HBSR PRODUCTION VOLUMES OBTAINED FROM THE HBSRG MODEL. THE MODEL RESULTS, IN THE FORM OF PRODUCTION/ATTRACTION SOCIAL-RECREATIONAL TRIPS BY MODE, CAN BE CONVERTED TO ORIGIN/DESTINATION TRIPS USING SPLITTING FACTORS GENERATED BY THE NHB MODELS. THE PRODUCTION/ATTRACTION SOCIAL-RECREATIONAL TRIPS, SUMMED BY ATTRACTION ZONE, ALSO SERVE AS A PORTION OF THE POPULATION OF POTENTIAL NHB TRIPS. IN ADDITION, ACCESSIBILITY VARIABLES FROM THIS MODEL ARE USED IN THE HBSRG MODEL. THE INDEPENDENT VARIABLES OF THIS MODEL FALL INTO THREE GROUPS, DEPENDING IN WHETHER THEY AFFECT ATTRACTION ZONE CHOICE, MODE CHOICE, OR BOTH. THE VARIABLES THAT AFFECT ATTRACTION ZONE CHOICE AND THEIR EFFECTS IN THIS CHOICE ARE: (1) NET RETAIL EMPLOYMENT DENSITY; RETAIL EMPLOYMENT IN THE ATTRACTION ZONE DIVIDED BY POPULATION-SERVING ACREAGE. SOCIAL- RECREATIONAL TRIP ATTRACTIONS INCREASE AS THIS DENSITY INCREASES. (2) POPULATION DENSITY IN THE ATTRACTION ZONE. SOCIAL-RECREATIONAL TRIP ATTRACTIONS INCREASE AS THIS DENSITY INCREASES. (3) TOTAL RETAIL EMPLOYMENT IN THE ATTRACTION ZONE. SOCIAL- RECREATIONAL TRIP ATTRACTION INCREASE AS RETAIL EMPLOYMENT INCREASES. (4) TOTAL POPULATION IN THE ATTRACTION ZONE. SOCIAL-RECREATIONAL TRIP ATTRACTIONS INCREASE AS POPULATION INCREASES. (5) THE TRIP LENGTH (DISTANCE) CORRECTION VARIABLE FOR A PAIR OF ZONES; THE AMOUNT (IF ANY) BY WHICH TOTAL TRIP DISTANCE EXCEEDS A THRESHOLD VALUE OF SEVEN MILES. SOCIAL-RECREATIONAL TRIP ATTRACTIONS DECREASE AS THIS VARIABLE INCREASES. NOTE THAT TO FACILITATE MODEL ESTIMATION, THE NET EFFECT OF THE RETAIL EMPLOYMENT (1 ABOVE) AND THE POPULATION (4 ABOVE) VARIABLES HBSRDM 9.17-2 IS CAPTURED IN THE TWO FOLLOWING TRANSFORMED VARIABLES: LNRETEMP AND LNPOPEMP. TWO INDEPENDENT VARIABLES AFFECT ONLY MODE CHOICE: (1) THE NUMBER OF AUTOS OWNED PER HOUSEHOLD MEMBER. MORE AUTO TRIPS ARE MADE AS THIS VARIABLE INCREASES. (2) THE NUMBER OF AUTOS REMAINING AVAILABLE TO A HOUSEHOLD AFTER ALL HOME BASED WORK TRIPS ARE MADE. MORE AUTO TRIPS ARE MADE AS THIS VARIABLE INCREASES. THE REMAINING INDEPENDENT VARIABLES AFFECT BOTH MODE AND ATTRACTION ZONE CHOICE: (1) WHETHER OR NOT THE ATTRACTION ZONE IS A CENTRAL BUSINESS DISTRICT. MORE SOCIAL-RECREATIONAL TRIPS ARE MADE TO CBD ZONES BY TRANSIT, AND FEWER SOCIAL-RECREATIONAL TRIPS ARE MADE TO CBD ZONES BY TRANSIT, AND FEWER SOCIAL-RECREATIONAL TRIPS ARE MADE TO CBD ZONES BY AUTO. (2) TOTAL TRAVEL TIME BY MODE. SOCIAL-RECREATIONAL TRIPS BY MODE DECREASE AS TRAVEL TOME BY MODE INCREASES. (4) TRANSIT FARE MULTIPLIED BY HOUSEHOLD SIZE; REFLECTING THE FACT THAT, WHEN CONTRASTED WITH INDIVIDUAL TRIPS, TRIPS BY FAMILY GROUPS ARE RELATIVELY MORE EXPENSIVE FOR THE FAMILY BY TRANSIT THAN THEY ARE BY AUTO. SOCIAL RECREATIONAL TRIPS BY TRANSIT DECREASE AS THIS VARIABLE INCREASES. 9.17.2 DISAGGREGATE MODEL FORMULATION THIS LOGIT MODEL HAS THE FOLLOWING UTILITY EQUATIONS: U(J,A,Q) :: = C+B(1)*LNRETEMP(J)+B(2)*RETEMPD(J) +B(3)*POPDEN(J)+B(4)*LNPOPEMP(J) +B(5)*INCATIME(I(Q),J)+B(6)*CBD(J) +B(7)*HBOACOST(I(Q),J) +B(8)*#CARS(H(Q))/#PHH(H(Q)) +B(9)*#AUTREM(H(Q))+B(12)*ODIST7(I(Q),J) U(J,T,G) :: = B(1)*LNRETEMP(J)+B(2)*RETEMPD(J)+B(3)*POPDEN(J) +B(4)*LNPOPEMP(J)+B(5)*INCTTIME(I(Q),J) +B(10)*CBD(J)+B(11)*#HHTCOST(J,Q) +B(12)*ODIST7(I(Q),J) WHERE: U(J, ,Q) IS THE UTILITY OF SOCIAL/RECREATIONAL DESTINATION ZONE J AND MODE M (A=AUTO, T=TRANSIT) TO TRAVELER Q. LNRETEMP(J) IS THE NATURAL LOGARITHM OF THE RETAIL EMPLOYMENT IN DESTINATION ZONE J (RETAIL EMPLOYMENT IN EMPLOYEES). RETEMPD(J) IS THE RETAIL EMPLOYMENT DENSITY IN ZONE J (EMPLOYEES/ ACRE). HBSRDM 9.17-3 POPDEN(J) IS THE POPULATION DENSITY IN ZONE J (PERSONS/ACRE). LNPOPEMP(J) IS THE NATURAL LOGARITHM OF THE RATIO OF POPULATION TO RETAIL EMPLOYMENT IN ZONE J (RATIO IN PERSONS/EMPLOYEE). INC TIME(I(Q),J) IS THE 1965 REGIONAL GROSS ANNUAL INCOME TIMES THE NATURAL LOGARITHM OF THE 1-WAY OFFPEAK TRAVEL TOME FROM Q'S RESIDENCE AONE, I(Q), TO ZONE J USING MODE M (1965 DOLLARS, TIME IN MINUTES). CBD(J) IS =1 IF DESTINATION ZONE J IS IN THE CBD, AND IS =0 IF J IS NOT IN THE CBD. HBOACOST(I(Q),J) IS THE 1-WAY OFFPEAK COST OF AN AUTO TRIP FROM Q'S RESIDENCE ZONE, I(Q), TO ZONE J (1965 CENTS). #CARS(H(Q)) IS THE NUMBER OF AUTOS OWNED BY TRAVELER Q'S HOUSEHOLD, H(Q) (AUTOS). #PHH(H(Q)) IS THE NUMBER OF PERSONS IN TRAVELER Q'S HOUSEHOLD H(Q) (PERSONS). #AUTREM(H(Q)) IS THE NUMBER OF AUTOS REMAINING TO TRAVELER Q'S HOUSEHOLD H(Q) AFTER SUBTRACTING ONE AUTO FOR EACH HOME BASED WORK TRIP MADE BY AUTO DRIVERS/WORKERS IN H()Q) (=0 IF NUMBER OF WORKERS EQUALS HOUSEHOLD SIZE( (AUTOS). #HHTCOST(J,Q) IS THE SIZE OF Q'S HOUSEHOLD TIME THE 1-WAY OFFPEAK TRANSIT FARE FROM Q'S RESIDENCE ZONE TO ZONE J(1965 PERSON-CENTS). ODIST7(I(Q),J) IS A DISTANCE CORRECTION VARIABLE, COMPUTED AS: ODIST7(I(Q),J) :: = MAX(OADIST(I(Q),J)-DTH),0) WHERE: OADIST(I(Q),J) IS THE 1-WAY OFFPEAK AUTO DISTANCE FROM Q'S RESIDENCE ZONE, I(Q) TO ZONE J (MILES). DTH IS A THRESHOLD DISTANCE (MILES). C, B(1)-B(12) ARE MODEL COEFFICIENTS: VARIABLE COEFFF VALUE ------ C 1.884 LNRETEMP B(1) 1.000 RETEMPD B(2) .0609 POPDEN B(3) .0244 LNPOPEMP B(4) .6998 INC TIME B(5) -1.097E-4 CBD B(6) -.215 HBOACOST B(7) -.0256 #CARS/#PHH B(8) 2.167 #AUTREM B(9) .3368 CBD B(10) 1.190 #HHTCOST B(11) -.0108 ODIST7 B(12) -.015 HBSRDM 9.17-4 9.17.3 AGGREGATE MODEL FORMULATION U(J,A,I,AO) :: = C+B(1)*LNRETEMP(J)+B(2)*RETEMPD(J) +B(3)*POPDEN(J)+B(4)*LNPOPEMP(J) +B(5)*INCATIME(I,J)+B(6)*CBD(J) +B(7)*HBOACOST(I,J)+B(8)*$MAO(AO)/PHH(I) +B(9)*($MAO(AO)/$MAOR)*$AUTREM +B(12)*HBSRDCF(I,J)+B(13)*$HBSRMCF(I) U(J,T,I,AO) :: = B(1)*LNRETEMP(J)+B(2)*RETEMPD(J) +B(3)*POPDEN(J)+B(4)*LNPOPEMP(J) +B(5)*INCTTIME(I,J)+B(10)*CBD(J) +B(11)*($PHHA(AO)/$PHH)*HHTCOST(I,J) +B(12)*HBSRDCF(I,J) WHERE: U(J, ,I,AO) IS THE UTILITY OF CHOOSING DESTINATION ZONE J AND MODE M TO TRAVELERS RESIDING IN ZONE I AND BELONGING TO AUTO OWNERSHIP LEVEL AO. LNRETEMP(J) IS THE NATURAL LOGARITHM OF THE RETAIL EMPLOYMENT IN DESTINATION ZONE J (RETAIL EMPLOYMENT IN EMPLOYEES). RETEMPD(J) IS THE RETAIL EMPLOYMENT DENSITY IN ZONE J (EMPLOYEES/ ACRE). POPDEN(J) IS THE POPULATION DENSITY IN ZONE J (PERSONS/ACRE). LNPOPEMP(J) IS THE NATURAL LOGARITHM OF THE RATIO OF POPULATION TO RETAIL EMPLOYMENT IN ZONE J (RATIO IN PERSONS/EMPLOYEE). INC TIME(I(Q),J) IS THE 1965 REGIONAL GROSS ANNUAL INCOME TIMES THE NATURAL LOGARITHM OF THE 1-WAY OFFPEAK TRAVEL TOME FROM Q'S RESIDENCE AONE, I(Q), TO ZONE J USING MODE M (1965 DOLLARS, TIME IN MINUTES). CBD(J) IS =1 IF DESTINATION ZONE J IS IN THE CBD, AND IS =0 IF J IS NOT IN THE CBD. HBOACOST(I,J) IS THE 1-WAY OFFPEAK COST OF AN AUTO TRIP FROM ZONE I TO ZONE J (1965 CENTS). PHH(I) IS THE MEAN HOUSEHOLD SIZE IN ZONE I (PERSONS). HHTCOST(I,J) IS THE MEAN HOUSEHOLD SIZE IN ZONE I TIMES THE 1-WAY OFFPEAK TRANSIT FARE FROM ZONE I TO ZONE J(1965 PERSON-CENTS). HBSRDCF(I,J) IS A DISTANCE CORRECTION FACTOR FOR A SOCIAL/RECREATIONAL TRIP FROM I TO J, COMPUTED AS: HBSRDCF(I,J) :: = MIN($HBSRDC(I)*(OADIST(I,J)-7),0) WHERE: OADIST(I,J) IS THE 1-WAY OFFPEAK AUTO DISTANCE FROM I TO J (MILES). $HBSRDC(I) IS THE COEFFICIENT OF THE DISTANCE CORRECTION VARIABLE FOR ZONE I; VALUES DERIVED FROM THE DISTRICT FACTORS GIVEN IN TABLE 9.17-1 ARE RECOMMENDED FOR SAN FRANCISCO BAY AREA APPLICATIONS ONLY; OTHERWISE -.047 RECOMMENDED FOR ALL ZONES. HBSRDM 9.17-5 C, B(1)-B(13) ARE MODEL COEFFICIENTS: VARIABLE COEFFF VALUE ------ C 1.376 LNRETEMP B(1) 1.000 RETEMPD B(2) .0609 POPDEN B(3) .0244 LNPOPEMP B(4) .6998 INC TIME B(5) -1.097E-4 CBD B(6) -.557 HBOACOST B(7) -.0256 $MAO/PHH B(8) 2.167 #AUTREM B(9) .3368 CBD B(10) -.760 HHTCOST B(11) -.0108 HBSRDCF B(12) 1.000 $HBSRMCF B(13) 1.000 $AUTREM(H(Q)) IS THE REGIONAL MEAN NUMBER OF AUTOS PER HOUSEHOLD AFTER SUBTRACTING ONE AUTO FOR EACH HOME BASED WORK TRIP MADE BY AUTO (AUTOS); .43 RECOMMENDED FOR 1965. $MAOR IS THE MEAN REGIONAL AUTO OWNERSHIP PER HOUSEHOLD (AUTOS); 1.5 RECOMMENDED FOR 1965. $MAO(AO) IS THE MEAN REGIONAL AUTO OWNERSHIP PER HOUSEHOLD FOR HOUSEHOLDS IN AUTO OWNERSHIP LEVEL AO (AUTOS); BY DEFINITION, $MAO(0)=0 AND $MAO(1)=1; 2.236 RECOMMENDED FOR $MAO(2+) FOR 1965. $PHH IS THE MEAN REGIONAL HOUSEHOLD SIZE (PERSONS); 3.81 RECOMMENDED. $PHHA(AO) IS THE MEAN REGIONAL SIZE OF HOUSEHOLDS IN AUTO OWNERSHIP LEVEL AO (PERSONS); 2.96, 3.53 AND 4.22 RECOMMENDED FOR AO=0, 1, AND 2+ RESPECTIVELY. $HBSRMCF(I) IS A MODAL ADJUSTMENT FACTOR FOR ZONE I; FACTORS DERIVED FROM DISTRICT VALUES GIVEN IN TABLE 9.17-2 ARE RECOMMENDED FOR SAN FRANCISCO BAY AREA APPLICATIONS ONLY; OTHERWISE 0.0 RECOMMENDED FOR ALL ZONES. HBSRDM 9.17-6 DIST VALUE DIST VALUE DIST VALUE 1 0.0 11 -.058 21 -.534 2 0.0 12 -.165 22 -.163 3 0.0 13 -.365 23 -.432 4 0.0 14 -.131 24 -.334 5 -.225 15 -.145 25 -2.759 6 -.229 16 -.175 26 -1.200 7 -.204 17 -.132 27 0.0 8 -.043 18 -.162 28 -.200 9 -.118 19 -.420 29 -.175 10 0.0 20 -.314 30 -.247 TABLE 9.17-1 HBSRDM AGGREGATE VALIDATION ADJUSTMENTS $HBSRDC DISTRICT VALUES COEFFICIENTS OF DISTANCE CORRECTION VARIABLE HBSRDM 9.17-7 DIST VALUE DIST VALUE DIST VALUE 1 -.278 11 .009 21 .604 2 -.117 12 .926 22 1.804 3 -.352 13 -.071 23 1.804 4 -.214 14 -.071 24 1.394 5 .085 15 -.071 25 1.394 6 .658 16 -.642 26 .443 7 .658 17 -.687 27 .443 8 .926 18 .604 28 .443 9 .926 19 .604 29 .640 10 .926 20 .604 30 .640 TABLE 9.17-2 HBSRDM AGGREGATE VALIDATION ADJUSTMENTS $HBSRMCF DISTRICT VALUES MODAL ADJUSTMENT FACTORS NHBAD 9.18-1 9.18 NON-HOME BASED AUTO TRIP DESTINATION MODEL (NHBAD) 9.18.1 GENERAL DESCRIPTION IN ITS DISAGGREGATE FORM THE NHBAD MODEL PREDICTS: 1) THE PROBABILITY IF A TRAVELER MAKING AN AUTO TRIP TO A DESTINATION OTHER THAN HIS HOME, GIVEN THAT HE IS AT AN ORIGIN OTHER THAN HIS HOME (I.E., GIVEN THAT THE TRAVELER IS EITHER AT THE ATTRACTION END OF A HOME BASED TRIP, OR AT THE DESTINATION END OF A NHB TRIP) AND THAT HE ARRIVED AT THIS ORIGIN BY AUTO, AND 2) THE PROBABILITY THAT, IF A NON-HOME BASED TRIP IS MADE, IT WILL BE DESTINED TO A PARTICULAR ZONE. IN ITS AGGREGATE FORM THE MODEL PREDICTS, FOR THE POOL OF POTENTIAL NON-HOME BASED TRIPMAKERS THAT ARRIVED IN A GIVEN ZONE BY AUTO, THE FRACTION THAT WILL NOT MAKE A NON-HOME BASED TRIP AND, FOR THE FRACTION MAKING A NON-HOME BASED TRIP, THE FRACTIONS DESTINED FOR EACH ZONE. THE MODEL IS NOT DEPENDENT ON ANY OTHER MODEL IN THE MTC MODEL SYSTEM. HOWEVER, TO OBTAIN A NON-HOME BASED AUTO TRIP TABLE, THE AGGREGATE FORM MUST BE APPLIED TO A POOL OF POTENTIAL NON-HOME BASED AUTO ATTRACTIONS FROM THE PHBWM, SHBWM, HBSHDM AND HBSRDM MODELS, TOGETHER WITH THE OUTPUTS OF THE NHBAO MODEL. THE RESULTS OBTAINED FROM THE JOINT APPLICATION OF THIS MODEL AND THE NHBAO MODEL INCLUDE BOTH NHB PERSON TRIPS BY AUTO (BY ORIGIN AND DESTINATION) AND SPLITTING FACTORS WHICH CAN BE USED TO CONVERT HOME BASED AUTO TRIPS FROM PRODUCTION/ATTRACTION FORMAT TO ORIGIN/DESTINATION FORMAT. THE INDEPENDENT VARIABLES IN THIS MODEL INCLUDE ONE WHICH AFFECTS ONLY THE PREDICTED FREQUENCY OF THE NHB AUTO TRIPS. THIS VARIABLE IS AN INDICATOR OF WHETHER OR NOT THE ORIGIN TRIP END IS IN A CENTRAL BUSINESS DISTRICT. THE PROBABILITY OF NHB AUTO TRIPS BEING MADE FROM CBD ORIGINS IS LESS THAN FROM NON-CBD ORIGINS. FIVE ADDITIONAL INDEPENDENT VARIABLES AFFECT THE CHOICE OF DESTINATION FOR NHB AUTO TRIPS THAT ARE MADE: (1) TOTAL TRAVEL TIME FROM ORIGIN TO DESTINATION. THE PROBABILITY OF DESTINATION CHOICE DECREASES AS TRAVEL TIME INCREASES. (2) TRAVEL COST FROM ORIGIN TO DESTINATION. THE PROBABILITY OF DESTINATION CHOICE DECREASES AS TRAVEL COST INCREASES. (3) VARIABLES REPRESENTING THE ATTRACTIVENESS IN THE ZONE AT DESTINATION END OF THE TRIP; EMPLOYMENT DENSITY, POPULATION, AND TOTAL EMPLOYMENT. THE PROBABILITY OF DESTINATION CHOICE INCREASES AS EACH OF THESE VARIABLES INCREASES. NHBAD 9.18-2 9.18.2 DISAGGREGATE MODEL FORMULATION THIS LOGIT MODEL HAS THE FOLLOWING UTILITY EQUATIONS: U(0,Q) :: = C+B(1)*CBD(I(Q) U(J,Q) :: = B(2)*LNPTEMP(J)+B(3)*TOTEMPD(J)+B(4)*LNEMP(J) +B(5)*LNATIME(I(Q),J)+B(6)*NHBACOST(I(Q),J) WHERE: U(0,Q) IS THE UTILITY TO TRIPMAKER Q OF NOT MAKING A NON- HOME BASED TRIP, GIVEN THAT HE IS AT AN ORIGIN OTHER THAN HIS HOME AND THAT HE ARRIVED THERE VIA AUTO (THIS IS THE ZERO FREQUENCY ALTERNATIVE). U(J,Q) IS THE UTILITY TO TRIPMAKER Q OF MAKING A NON-HOME BASED AUTO TRIP TO ZONE J, GIVEN THAT HE IS AT AN ORIGIN OTHER THAN HIS HOME AND THAT HE ARRIVED THERE VIA AUTO. CBD(I(Q) =1 IF Q'S ORIGIN ZONE, I(Q), IS IN THE CBD, AND =0 IF I(Q) IS NOT IN THE CBD. LNPTEMP(J) IS THE NATURAL LOGARITHM OF THE RATIO OF THE POPULATION TO THE TOTAL EMPLOYMENT OF DESTINATION ZONE J (RATIO IN PERSONS/EMPLOYEE). TOTEMPD(J) IS THE GROSS EMPLOYMENT DENSITY IN DESTINATION ZONE J (EMPLOYEES/ACRE). LNEMP(J) IS THE NATURAL LOGARITHM OF THE TOTAL EMPLOYMENT IN DESTINATION ZONE J (EMPLOYMENT IN THOUSANDS OF EMPLOYEES). LNATIME(I(Q),J) IS THE NATURAL LOGARITHM OF THE TOTAL OFFPEAK AUTO TRAVEL TIME FROM Q'S ORIGIN ZONE, I(Q), TO ZONE J (TIME IN MINUTES). NHBACOST(I(Q),J) IS THE TOTAL OFFPEAK COST OF AN AUTO TRIP FROM Q'S ORIGIN ZONE, I(Q), TO ZONE J (1965 CENTS). C,B(1)-B(6) ARE MODEL COEFFICIENTS: VARIABLE COEFF VALUE --- C -.2703 CBD B(1) 1.126 LNPTEMP B(2) .4700 TOTEMPD B(3) .002242 LNEMP B(4) 1.000 LNATIME B(5) -2.294 NHBACOST B(6) -.01518 9.18.3 AGGREGATE MODEL FORMULATION U(0,I) :: = C+B(1)*CBD(I) U(J,I) :: = B(2)*LNPTEMP(J)+B(3)*TOTEMPD(J)+B(4)*LNEMP(J) +B(5)*LNATIME(I,J)+B(6)*NHBACOST(I,J) 9.18-3 WHERE: U(0,I) IS THE UTILITY TO POTENTIAL NHB AUTO TRIPMAKERS IN ORIGIN ZONE I OF NOT MAKING A NON-HOME BASED TRIP. U(J,I) IS THE UTILITY TO POTENTIAL NHB AUTO TRIPMAKERS IN ORIGIN ZONE I OF MAKING A NON-HOME BASED TRIP TO ZONE J. CBD(I) =1 IF ORIGIN ZONE I IS IN THE CBD, AND =0 IF I IS NOT IN THE CBD. LNPTEMP(J) IS THE NATURAL LOGARITHM OF THE RATIO OF THE POPULATION TO THE TOTAL EMPLOYMENT OF DESTINATION ZONE J (RATIO IN PERSONS/EMPLOYEE). TOTEMPD(J) IS THE GROSS EMPLOYMENT DENSITY IN DESTINATION ZONE J (EMPLOYEES/ACRE). LNEMP(J) IS THE NATURAL LOGARITHM OF THE TOTAL EMPLOYMENT IN DESTINATION ZONE J (EMPLOYMENT IN THOUSANDS OF EMPLOYEES). LNATIME(I(Q),J) IS THE NATURAL LOGARITHM OF THE TOTAL OFFPEAK AUTO TRAVEL TIME FROM Q'S ORIGIN ZONE, I(Q), TO ZONE J (TIME IN MINUTES). NHBACOST(I(Q),J) IS THE TOTAL OFFPEAK COST OF AN AUTO TRIP FROM Q'S ORIGIN ZONE, I(Q), TO ZONE J (1965 CENTS). C,B(1)-B(6) ARE MODEL COEFFICIENTS: VARIABLE COEFF VALUE --- C .1903 CBD B(1) .4122 LNPTEMP B(2) .4700 TOTEMPD B(3) .002242 LNEMP B(4) 1.000 LNATIME B(5) -2.294 NHBACOST B(6) -.01518 9.19-1 9.19 NON-HOME BASED AUTO TRIP ORIGIN MODEL (NHBAO) 9.19.1 GENERAL DESCRIPTION IN ITS DISAGGREGATE FORM, THE NHBAD MODEL PREDICTS: 1) THE PROBABILITY THAT A TRAVELER HAS MADE AN AUTO TRIP FROM AN ORIGIN OTHER THAN HIS HOME, GIVEN THAT HE IS AT A DESTINATION OTHER THAN HIS HOME (I.E., GIVEN THAT THE TRAVELER IS EITHER AT THE ATTRACTION END OF A HOME BASED TRIP, OR AT THE DESTINATION END OF A NON-HOME BASED TRIP_) AND THAT HE LEAVES THIS DESTINATION BY AUTO, AND 2) THE PROBABILITY THAT, IF A NON-HOME BASED TRIP HAS BEEN MADE, IT CAME FROM A PARTICULAR ZONE. IN ITS AGGREGATE FORM THE MODEL PREDICTS, FOR THE POOL OF POTENTIAL NON-HOME BASED TRIPMAKERS THAT ARRIVED IN A GIVEN ZONE BY AUTO, THE FRACTION THAT HAS NOT MADE A NON-HOME BASED TRIP, THE FRACTION ORIGINATING IN EACH ZONE. THE MODEL REQUIRES NO INPUTS FROM ANY OTHER MODEL IN THE MTC MODEL SYSTEM. HOWEVER, TO OBTAIN A NON-HOME BASED AUTO TRIP TABLE FROM THE AGGREGATE MODEL, IT MUST BE APPLIED, FOR EACH ZONE, TO A POOL OF POTENTIAL NON-HOME BASED AUTO TRIP DESTINATIONS, OBTAINED BY PROCESSING HOME BASED ATTRACTIONS FROM THE PHBWM, SHBWM, HBSHDM AND HBSRDM MODELS, TOGETHER WITH THE OUTPUTS OF THE NHBAC MODEL. THE RESULTS OBTAINED FROM THE JOINT APPLICATION OF THIS MODEL AND NHBAD MODEL INCLUDE BOTH NHB PERSON TRIPS BY AUTO (BY ORIGIN AND DESTINATION) AND SPLITTING FACTORS THAT CAN BE USED TO CONVERT HOME BASED AUTO TRIPS FROM PRODUCTION/ATTRACTION FORMAT TO ORIGIN/DESTINATION FORMAT. OF THE INDEPENDENT VARIABLES IN THIS MODEL, ONLY ONE AFFECTS THE PREDICTED FREQUENCY OF NHB AUTO TRIPS. THIS VARIABLE IS AN INDICATOR OF WHETHER OR NOT THE DESTINATION TRIP END IS IN A CBD. THE PROBABILITY OF NHB AUTO TRIPS BEING MADE TO CBD DESTINATIONS IS LESS THAN TO NON-CBD DESTINATIONS. FIVE ADDITIONAL INDEPENDENT VARIABLES AFFECT THE CHOICE OF ORIGIN FOR NHB AUTO TRIPS WHICH ARE MADE: (1) TOTAL TRAVELTIME FROM ORIGIN TO DESTINATION.THE PROBABILITY OF ORIGIN CHOICE DECREASES AS TRAVEL TIME INCREASES. (2) TRAVEL COST FROM ORIGIN TO DESTINATION. THE PROBABILITY OF ORIGIN CHOICE DECREASES AS TRAVEL COST INCREASES. (3) VARIABLES REPRESENTING THE ATTRACTIVENESS IN THE ZONE AT THE ORIGIN END OF THE TRIP: EMPLOYMENT DENSITY, POPULATION, AND TOTAL EMPLOYMENT. THE PROBABILITY OF ORIGIN CHOICE INCREASES AS EACH OF THESE VARIABLES INCREASES. 9.19-2 9.19.2 DISAGGREGATE MODEL FORMULATION THIS LOGIT MODEL HAS THE FOLLOWING UTILITY EQUATIONS: U(0,Q) :: = C+B(1)*CBD(J(Q)) U(I,Q) :: = B(2)*LNPTEMP(I)+B(3)*TOTEMPD(I)+B(4)*LNEMP(I) +B(5)*LNATIME(I,J(Q))+B(6)*NHBCOST(I,J(Q)) WHERE: U(0,Q) IS THE UTILITY TO TRIPMAKER OF NOT HAVING MADE A NON-HOME BASED TRIP, GIVEN THAT HE IS AT A DESTINATION OTHER THAN HIS HOME AND THAT HE WILL LEAVE THERE VIA AUTO (THIS IS THE ZERO FREQUENCY ALTERNATIVE. U(I,Q) IS THE UTILITY TO TRIPMAKER Q OF HAVING MADE A NON- HOME BASED TRIP FROM ZONE I, GIVEN THAT HE IS AT A DESTINATION OTHER THAN HIS HOME AND THAT HO WILL LEAVE THERE VIA AUTO. CBD(J(Q)) =1 IF Q'S DESTINATION ZONE, J(Q), IS IN THE CBD, AND =0 IF J(Q) IS NOT IN THE CBD. LNPTEMP(I) IS THE NATURAL LOGARITHM IF THE RATIO OF THE POPULATION TO THE TOTAL EMPLOYMENT IN ORIGIN ZONE I (RATIO IN PERSONS/EMPLOYEE). TOTEMPD(I) IS THE GROSS EMPLOYMENT DENSITY IN ORIGIN ZONE I (EMPLOYEE/ACRE). LNEMP(I) IS THE NATURAL LOGARITHM OF THE TOTAL EMPLOYMENT IN ORIGIN ZONE I (EMPLOYMENT IN THOUSANDS OF EMPLOYEES). LNATIME(I,J(Q)) IS THE NATURAL LOGARITHM OF THE 1-WAY OFFPEAK AUTO TRAVELTIME FROM ZONE I TO Q'S DESTINATION ZONE, J(Q) (TIME IN MINUTES). NHBCOST(I,J(Q)) IS THE 1-WAY OFF PEAK COST OF AN AUTO TRIP FROM ZONE I TO Q'S DESTINATION ZONE, J(Q) (1965 CENTS) C,B(1)-B(6) ARE MODEL COEFFICIENTS: VARIABLE COEFF VALUE ____ C -.557 CBD B(1) 1.096 LNPTEMP B(2) .3981 TOTEMPD B(3) .001887 LNEMP B(4) 1.000 LNATIME B(5) -2.418 NHBACOST B(6) -.01338 9.19.3 AGGREGATE MODEL FORMULATION U(0,J) :: = C+B(1)*CBD(J) U(I,J) :: = B(2)*LNPTEMP(I)+B(3)*TOTEMPD(I)+B(4)*LNEMP(I) +B(5)*LNATIME(I,J)+B(6)*NHBCOST(I,J) 9.19-3 WHERE: U(0,J) IS THE UTILITY TO POTENTIAL NHB AUTO TRIPMAKERS IN DESTINATION ZONE J OF NOT HAVING MADE A NON-HOME BASED TRIP. U(I,J) IS THE UTILITY TO POTENTIAL NHB AUTO TRIPMAKERS IN DESTINATION ZONE J OF HAVING MADE A NON-HOME BASED TRIP FROM ZONE I. CBD(J) =1 DESTINATION ZONE J IS IN THE CBD, AND =0 IF J IS NOT IN THE CBD. LNPTEMP(I) IS THE NATURAL LOGARITHM IF THE RATIO OF THE POPULATION TO THE TOTAL EMPLOYMENT IN ORIGIN ZONE I (RATIO IN PERSONS/EMPLOYEE). TOTEMPD(I) IS THE GROSS EMPLOYMENT DENSITY IN ORIGIN ZONE I (EMPLOYEE/ACRE). LNEMP(I) IS THE NATURAL LOGARITHM OF THE TOTAL EMPLOYMENT IN ORIGIN ZONE I (EMPLOYMENT IN THOUSANDS OF EMPLOYEES). LNATIME(I,J) IS THE NATURAL LOGARITHM OF THE 1-WAY OFFPEAK AUTO TRAVEL TIME FROM ZONE I TO DESTINATION ZONE, J (TIME IN MINUTES). NHBCOST(I,J) IS THE 1-WAY OFF PEAK COST OF AN AUTO TRIP FROM ZONE I DESTINATION ZONE J (1965 CENTS) C,B(1)-B(6) ARE MODEL COEFFICIENTS: VARIABLE COEFF VALUE ____ C -.1350 CBD B(1) .5959 LNPTEMP B(2) .3981 TOTEMPD B(3) .001887 LNEMP B(4) 1.000 LNATIME B(5) -2.418 NHBACOST B(6) -.01338 NHBTD 9.20-1 9.20 NON-HOME BASED TRANSIT TRIP DESTINATION MODEL (NHBTD) 9.20.1 GENERAL DESCRIPTION IN ITS DISAGGREGATE FORM THE NHBTD MODEL PREDICTS: 1) THE UTILITY OF A TRAVELER MAKING A TRANSIT TRIP TO A DESTINATION OTHER THAN HIS HOME, GIVEN THAT HE IS AT AN ORIGIN OTHER THAN HIS HOME (I.E., GIVEN THAT THE TRAVELER IS EITHER AT THE ATTRACTION END OF A NON- HOME BASED TRIP) AND THAT HE ARRIVED AT THIS ORIGIN BY TRANSIT, AND 2) THE PROBABILITY THAT, IF A NON-HOME BASED TRIP IS MADE, IT WILL BE DESTINED FOR A PARTICULAR ZONE. IN ITS AGGREGATE FORM THE MODEL PREDICTS, FOR THE POOL OF POTENTIAL NON-HOME BASED TRIPMAKERS THAT ARRIVES IN A GIVEN ZONE BY TRANSIT, THE FRACTION THAT WILL NOT MAKE A NON-HOME BASED TRIP AND, FOR THE FRACTION THAT WILL MAKE A NON-HOME BASED TRIP, THE FRACTION DESTINED FOR EACH ZONE. THE MODEL REQUIRES NO INPUTS FROM ANY OTHER MODEL IN THE MTC MODEL SYSTEM. HOWEVER, TO OBTAIN A NON-HOME BASED TRANSIT TRIP TABLE, THE AGGREGATE MODEL MUST BE APPLIED FOR EACH ZONE IN THE REGION, TO A POOL OF POTENTIAL NON-=HOME BASED TRANSIT TRIP ORIGINS, OBTAINED BY PROCESSING HOME BASED ATTRACTIONS FROM THE PHBWM, SHBWM, HBSHDM AND HBSRDM MODELS, TOGETHER WITH THE OUTPUTS OF THE NHBTO MODEL. THE RESULTS OBTAINED FROM THE JOINT APPLICATION OF THIS MODEL AND THE NHBTO MODEL INCLUDE BOTH NHB PERSON TRIPS BY TRANSIT (BY ORIGIN AND DESTINATION) AND SPLITTING FACTORS WHICH CAN BE USED TO CONVERT HOME BASED TRANSIT TRIPS FROM PRODUCTION/ATTRACTION FORMAT TO ORIGIN/DESTINATION FORMAT. THE INDEPENDENT VARIABLES IN THIS MODEL INCLUDE ONE THAT AFFECTS ONLY THE PREDICTED FREQUENCY OF NHB TRANSIT TRIPS. THIS VARIABLE IS AN INDICATOR OF WHETHER OR NOT THE ORIGIN TRIP END IS IN A CENTRAL BUSINESS DISTRICT. THE PROBABILITY OF NHB TRANSIT TRIPS BEING MADE FROM CBD ORIGINS IS LESS THAN FROM NON-CBD ORIGINS. SIX ADDITIONAL INDEPENDENT VARIABLES AFFECT THE CHOICE OF DESTINATION FOR NHB TRANSIT TRIPS THAT ARE MADE: (1) TOTAL TRAVEL TIME FROM ORIGIN TO DESTINATION. THE PROBABILITY OF DESTINATION CHOICE DECREASES AS TRAVEL TIME INCREASES. (2) TRAVEL COST FROM ORIGIN TO DESTINATION. THE PROBABILITY OF DESTINATION CHOICE DECREASES AS TRAVEL COST INCREASES. (3) VARIABLES REPRESENTING THE ATTRACTIVENESS IN THE ZONE AT THE DESTINATION END OF THE TRIP: EMPLOYMENT DENSITY, POPULATION, AND TOTAL EMPLOYMENT. THE PROBABILITY OF DESTINATION CHOICE INCREASES AS EACH OF THESE VARIABLES INCREASES. (4) TRIP LENGTH (DISTANCE) CORRECTION VARIABLE FROM ORIGIN TO DESTINATION; THE AMOUNT (IF ANY) BY WHICH ORIGIN TO 9.20-2 DESTINATION TRIP LENGTH EXCEEDS SEVEN MILES. THE PROBABILITY OF DESTINATION CHOICE AS TRIP DISTANCE INCREASES BEYOND SEVEN MILES. 9.20.2 DISAGGREGATE MODEL FORMULATION THIS LOGIT MODEL HAS THE FOLLOWING UTILITY EQUATIONS: U(0,Q) :: = C+B(6)*CBD(I(Q)) U(J,Q) :: = B(1)*LNPTEMP(J)+B(2)*TOTEMPD(J)+B(3)*LNEMP(J) +B(4)*LNTTIME(I(Q),J)+B(5)*OTFARE(I(Q),J) +B(7)*ODIST7(I(Q),J) WHERE: U(0,Q) IS THE UTILITY TO TRIPMAKER Q OF NOT MAKING A NON- HOME BASED TRIP, GIVEN THAT HE IS AT AN ORIGIN OTHER THAN HIS HOME AND THAT HE ARRIVED THERE VIA TRANSIT (THIS IS THE ZERO FREQUENCY ALTERNATIVE). U(J,Q) IS THE UTILITY TO TRIPMAKER Q OF MAKING A NON-HOME BASED TRANSIT TRIP TO ZONE J, GIVEN THAT HE IS AT AN ORIGIN OTHER THAN HIS HOME AND THAT HE ARRIVED THERE VIA TRANSIT. LNPTEMP(J) IS THE NATURAL LOGARITHM IF THE RATIO OF THE POPULATION TO TOTAL EMPLOYMENT IN DESTINATION ZONE J (RATIO IN PERSONS/EMPLOYEE). TOTEMPD(J) IS THE GROSS EMPLOYMENT DENSITY IN DESTINATION ZONE J (EMPLOYEES/ACRE). LNEMP(J) IS THE NATURAL LOGARITHM IF THE TOTAL EMPLOYMENT IN DESTINATION ZONE J (EMPLOYMENT IN THOUSANDS OF EMPLOYEES). LNTTIME(I(Q),J) IS THE NATURAL LOGARITHM OF THE 1-WAY OFFPEAK TRANSIT TRAVEL TIME FROM Q'S ORIGIN ZONE, I(Q), TO ZONE J (TIME IN MINUTES). OTFARE(I(Q),J) IS THE 1-WAY OFFPEAK FARE FOR A TRANSIT TRIP FROM Q'S ORIGIN ZONE, I(Q), TO ZONE J (1965 CENTS) CBD(I(Q)) =1 IF Q'S ORIGIN ZONE, I(Q), IS IN THE CBD, AND =0 IF I(Q) IS NOT IN THE CBD. ODIST7(I(Q),J) IS A DISTANCE CORRECTION VARIABLE FOR THE TRIP FROM I TO J (MILES), COMPUTED AS: ODIST7(I,J) :: = MAX((OADIST(I,J)-7,0) WHERE: ODIST7(I,J) IS THE 1-WAY OFFPEAK AUTO DISTANCE FROM ZONE I TO ZONE J (MILES). C,B(1)-B(7) ARE MODEL COEFFICIENTS: NHBTO 9.20-3 VARIABLE COEFF VALUE ----- C 3.440 LNPTEMP B(1) .6088 TOTEMPD B(2) .003566 LNEMP B(3) 1.000 LNTTIME B(4) -1.033 OTFARE B(5) -.01608 CBD B(6) 1.213 ODIST7 B(7) -.048 9.20.3 AGGREGATE MODEL FORMULATION U(0,I) :: = C+B(6)*CBD(I) U(J,I) :: = B(1)*LNPTEMP(J)+B(2)*TOTEMPD(J)+B(3)*LNEMP(J) +B(4)*LNTTIME(I,J)+B(5)*OTFARE(I,J) +B(7)*HBSRDCF(I,J) WHERE: U(0,I) IS THE UTILITY TO POTENTIAL NHB TRANSIT TRIPMAKERS IN ORIGIN ZONE I OF NOT MAKING A NON-HOME BASED TRIP. U(J,Q) IS THE UTILITY TO POTENTIAL NHB TRANSIT TRIPMAKERS IN ORIGIN ZONE I OF MAKING A NON-HOME BASED TRANSIT TRIP TO ZONE J. LNPTEMP(J) IS THE NATURAL LOGARITHM IF THE RATIO OF THE POPULATION TO TOTAL EMPLOYMENT IN DESTINATION ZONE J (RATIO IN PERSONS/EMPLOYEE). TOTEMPD(J) IS THE GROSS EMPLOYMENT DENSITY IN DESTINATION ZONE J (EMPLOYEES/ACRE). LNEMP(J) IS THE NATURAL LOGARITHM IF THE TOTAL EMPLOYMENT IN DESTINATION ZONE J (EMPLOYMENT IN THOUSANDS OF EMPLOYEES). LNTTIME(I,J) IS THE NATURAL LOGARITHM OF THE 1-WAY OFFPEAK TRANSIT TRAVEL TIME FROM ORIGIN ZONE I TO ZONE J (TIME IN MINUTES). OTFARE(I,J) IS THE 1-WAY OFFPEAK FARE FOR A TRANSIT TRIP FROM ORIGIN ZONE I TO ZONE J (1965 CENTS) CBD(I) =1 IF ORIGIN ZONE I, IS IN THE CBD, AND =0 IF I IS NOT IN THE CBD. HBSRDCF(I,J) IS A DISTANCE CORRECTION VARIABLE FOR THE TRIP FROM I TO J (MILES), COMPUTED AS: HBSRDCF(I,J) :: = MIN($HBSRDC(I)*(OADIST(I,J)-7),0) WHERE: ODIST(I,J) IS THE 1-WAY OFFPEAK AUTO DISTANCE FROM ZONE I TO ZONE J (MILES). HBSRDCF(I,J) IS THE COEFFICIENT IF THE DISTANCE CORRECTION VARIABLE FOR ZONE I; VALUES DERIVED FROM THE DISTRICT FACTORS GIVEN IN TABLE 9.17-1 ARE RECOMMENDED FOR SAN FRANCISCO BAY AREA NHBTD 9.20- 4 APPLICATIONS ONLY; OTHERWISE -.047 RECOMMENDED FOR ALL ZONES. C,B(1)-(B(7) ARE MODEL COEFFICIENTS: VARIABLE COEFF VALUE -- C 3.440 LNPTEMP B(1) .6088 TOTEMPD B(2) .003566 LNEMP B(3) 1.000 LNTTIME B(4) -1.033 OTFARE B(5) -.01608 CBD B(6) 1.213 HBSRDCF B(7) 1.00 NHBTO 9.21- 1 9.21. NON-HOME BASED TRANSIT TRIP ORIGIN MODEL (NHBTO) 9.21.1 GENERAL DESCRIPTION IN ITS DISAGGREGATE FORM THE NHBTO MODEL PREDICTS: 1) THE PROBABILITY THAT A TRAVELER HAS MADE A TRANSIT TRIP FROM AN ORIGIN OTHER THAN HIS HOME, GIVEN THAT HE IS AT A DESTINATION OTHER THAN HIS HOME (I.E., GIVEN THAT THE TRAVELER IS EITHER AT THE ATTRACTION END OF A HOME BASED TRIP, OR AT THE DESTINATION END OF A NON-HOME BASED TRIP) AND THAT HE WILL LEAVE THIS DESTINATION BY TRANSIT, AND 2) THE PROBABILITY THAT, IF A NON-HOME BASED TRIP HAS BEEN MADE, IT CAME FROM A PARTICULAR ZONE. IN ITS AGGREGATE FORM THE MODEL PREDICTS, FOR THE POOL OF POTENTIAL NON-HOME BASED TRIPMAKERS THAT ARRIVED IN A GIVEN ZONE BY TRANSIT, THE FRACTION THAT HAS NOT MADE A NON-HOME BASED TRIP AND , FOR THE FRACTION THAT HAS MADE A NON-HOME BASED TRIP, THE FRACTION ORIGINATING IN EACH ZONE. THE MODEL REQUIRES NO INPUTS FROM ANY OTHER MODEL IN THE MTC MODEL SYSTEM. HOWEVER, TO OBTAIN A NON-HOME BASED TRANSIT TRIP TABLE, THE AGGREGATE MODEL MUST BE APPLIED, FOR EACH ZONE IN THE REGION, TO A POOL OF POTENTIAL NON-HOME BASED TRANSIT TRIPMAKERS, OBTAINED BY PROCESSING HOME BASED ATTRACTIONS FROM THE PHBWM, SHBWM, HBSHDM AND HGSRDM MODELS, TOGETHER WITH THE OUTPUTS OF THE NHBTO MODEL. THE RESULTS OBTAINED FROM THE JOINT APPLICATION OF THIS MODEL AND THE NHBTO MODEL INCLUDE BOTH NHB PERSON TRIPS BY TRANSIT (BY ORIGIN AND DESTINATION) AND SPLITTING FACTORS WHICH CAN BE USED TO CONVERT HOME BASED TRANSIT TRIPS FROM PRODUCTION/ATTRACTION FORMAT. TO ORIGIN/DESTINATION FORMAT. OF THE INDEPENDENT VARIABLES IN THIS MODEL, ONLY ONE AFFECTS THE PREDICTED FREQUENCY OF NHB TRANSIT TRIPS. THIS VARIABLE IS AN INDICATOR OF WHETHER OR NOT THE DESTINATION TRIP END IS IN A CENTRAL BUSINESS DISTRICT. THE PROBABILITY OF NHB TRANSIT TRIPS BEING MADE TO CBD DESTINATIONS IS LESS THAN TO NON-CBD DESTINATIONS. FOUR ADDITIONAL INDEPENDENT VARIABLES AFFECT THE CHOICE OF ORIGIN FOR NHB TRANSIT TRIPS THAT ARE MADE: (1) TOTAL TRAVEL TIME FROM ORIGIN TO DESTINATION. THE PROBABILITY OF ORIGIN CHOICE DECREASES AS TRAVEL TIME INCREASES. (2) TRAVEL COST FROM ORIGIN TO DESTINATION. THE PROBABILITY OF ORIGIN CHOICE DECREASES AS TRAVEL COST INCREASES. (3) THE TOTAL EMPLOYMENT OF THE ZONE AT THE ORIGIN END OF THE TRIP. THE PROBABILITY OF ORIGIN CHOICE INCREASES AS EMPLOYMENT INCREASES. (4) THE TRIP LENGTH (DISTANCE) CCRRECTION VARIABLE FROM ORIGIN TO DESTINATION; THE AMOUNT (IF ANY) BY WHICH ORIGIN TO DESTINATION TRIP LENGTH EXCEEDS SEVEN MILES. THE PROBABILITY NHBTO 9.21- 2 OF DESTINATION CHOICE DECREASES AS TRIP DISTANCE INCREASES BEYOND SEVEN MILES. 9.21.2 DISAGGREGATE MODEL FORMULATION THIS LOGIT MODEL HAS THE FOLLOWING UTILITY EQUATIONS: U(O,Q) ::= C+B(1)*CBD(J(Q)) U(I,Q) ::= B(2)*LNEMP(I)+B(3)*LNTTIME(I,J(Q)) +B(4)*OTFARE(I,J(Q)))+B95)*ODIST7(I,J9Q)) WHERE: U(C,Q) IS THE UTILITY TO TRIPMAKER Q OF NOT HAVING MADE A NON-HOME BASED TRIP, GIVEN THAT HE IS AT A DESTINATION OTHER THAN HIS HOME AND THAT HE WILL LEAVE THER VIA TRANSIT (THIS IS THE ZERO FREQUENCY ALTERNATIVE). U(I,Q) IS THE UTILITY TO TRIPMAKER Q OF HAVING MADE A NON- HOME BASED TRIP FROM ZONE I, GIVEN THAT HE IS AT A DESTINATION OTHER THAN HIS HOME AND THAT HE WILL LEAVE THERE VIZ TRANSIT. CBD(J(Q)) =1 IF Q'S DESTINATION ZONE, J(Q), IS IN THE CBD, AND =0 IF J(Q) IS NOT IN THE CBD. LNEMP(I) IS THE NATURAL LOGARITHM OF THE TOTAL EMPLOYMENT IN ORIGIN ZONE I (EMPLOYMENT IN THOUSANDS OF EMPLOYEES). LNTTIME(I,J(Q)) IS THE NATURAL LOGARITHM OF THE 1-WAY OFFPEAK TRANSIT TRAVEL TIME FROM ZONE I TO Q'S DESTINATION ZONE, J(Q) (TIME IN MINUTES). OTFARE(I,J(Q)) IS THE 1-WAY OFFPEAK FARE FOR A TRANSIT TRIP FROM ZONE I TO Q'S DESTINATION ZONE, J(Q) (1965 CENTS). ODIST7(I,J(Q)) IS A DISTANCE CORRECTION VARIABLE FOR A TRIP FROM ZONE I TO ZONE J(Q) (MILES). C,B(1)-B(5) ARE MODEL COEFFICIENTS: VARIABLE COEFF VALUE -- C 3.475 CBD B(1) 1.106 LNEMP B(2) 1.000 LNTTIME B(3) -.8982 OTFARE B(4) -.01775 ODIST7 B(5) -.048 9.21.3 AGGREGATE MODEL FORMULATION U(O,J) ::= C+B(1)*CBD(J) U(I,J) ::= B(2)*LNEMP(1)+B(3)*LNTTIME(I,J) +B(4)*OTFARE(I,J)+B(5)*HBSRDCF(I,J) WHERE: U(O,J) IS THE UTILITY TO POTENTIAL NHB TRANSIT TRIPMAKERS IN DESTINATION ZONE J OF NOT HAVING MADE A NON-HOME NHBTO 9.21-3 BASED TRIP. U(I,J) IS THE UTILITY TO POTENTIAL NHB TRANSIT TRIPPAKERS IN DESTINATION ZONE J OF HAVING MADE A NON-HOME BASED TRIP FROM ZONE I. CBD(J) =1 IF DESTINATION ZONE J IS IN THE CBD, AND =0 IF J IS NOT IN THE CBD. LNEMP(I) IS THE NATURAL LOGARITHM OF THE TOTAL EMPLOYMENT IN ORIGIN ZONE I (EMPLOYMENT IN THOUSANDS OF EMPLOYEES). LNTTIME(I,J) IS THE NATURAL LOGARITHM OF THE 1-WAY OFFPEAK TRANSIT TRAVEL TIME FROM ORIGIN ZONE I TO DESTINATION ZONE J (TIME IN MINUTES). OTFARE(I,J) IS THE 1-WAY OFFPEAK FARE OF A TRANSIT TRIP FROM ORIGIN ZONE I TO DESTINATION ZONE J (1965 CENTS). HBSRDCF(I,J) IS A DISTANCE CORRECTION FACTOR FOR THE TRIP FROM I TO J, COMPUTED AS: HBSRDCF(I,J) ::= MIN($HBSRDC(1)*(OADIST(I,J)-7),0) WHERE: OADIST(I,J) IS THE 1-WAY OFFPEAK AUTO DISTANCE FROM I TO J (MILES). $HBSRDC(I) IS THE COEFFICIENT OF THE DISTANCE CORRECTION VARIABLE FOR ZONE I; VALUES DERIVED FROM THE DISTRICT FACTORS GIVEN IN TABLE 9.17-1 ARE RECOMMENDED FOR SAN FRANCISCO BAY AREA APPLICATIONS ONLY; OTHERWISE -.047 RECOMMENDED FOR ALL ZONES. VARIABLE COEFF VALUE -- C 3.647 CBD B(1) 1.106 LNEMP B(2) 1.000 LNTTIME B(3) -.8982 OTFARE B(4) -.01775 HBSRDCF B(5) 1.000 CONCLUSION 10- 1 10. CONCLUSION THE MTC TRAVEL DEMAND MODEL SYSTEM REPRESENTS THE FIRST PRODUCTION- ORIENTED SYSTEM DEVELOPED FOR USE BY A METROPOLITAN PLANNING ORGANIZATION THAT IS BASED ON A CONSISTENT THEORY OF TRAVELER BEHAVIOR AND ON DISAGGREGATE MODEL ESTINATION. THE COMPONENT MODELS OF THE SYSTEM REPRESENT A WORKABLE COMPROMISE BETWEEN BEHAVIORAL THEORY, ABILITY TO CALIBRATE, AND ABILITY TO TOBTAIN AGGREGATE TRAVEL ESTIMATES, GIVEN REASONABLE DATA AVAILABILITY, TIME, AND COMPUTER COST CONSTRAINTS. VALIDATION OF THE MODELS HAS BEEN PERFORMED AT THE DISAGGREGATE LEVEL FOR THE COMPONENT MODELS AND AT THE AGGREGATE LEVEL FOR THE ENTIRE MODEL SYSTEM. APPLICATION PROCEDURES HAVE ALSO BEEN DEVELOPED BOTH FOR DETAILED NETWORK ANALYSES AND FOR SHOR-RANGE GENERALIZED POLICY ANALYSIS. OTHER METROPOLITAN PLANNING ORGANIZATIONS MAY BENEFIT FROM EACH ASPECT FO THE PROJECT (THE MODEL STRUCTURE, THE DISAGGREGATE MODELS THEMSELVES, AND THE APPLICATION PROGRAMS), AS THEY CONTINUE THE PROCESS OF MODEL REFINEMENT NECESSARY TO ADDRESS THE VARIED TRANSPORTATION-RELATED POLICY ISSUES REQUIRING ANALYSIS NOW AND IN THE FUTURE. HOWEVER, REAPING THESE BENEFITS WILL REQUIRE SOME ADAPTATION, RE-ESTIMATION AND/OR AGGREGATE VALIDATION TO A RECENT YEAR DATA BASE. IT SHOULD BE NOTED THAT MTC HAS AND IS CONTINUING TO REFINE THE MODELS TO IMPROVE THEIR ABILITY TO PRODUCE REASONABLE FORECASTS. THESE REFINEMENTS ARE BASED ON AGGREGATE VALIDATION OF THE MODELS FOR THE YEAR 1975 USING CONVENTIONAL MEHTODS FOR CALIBRATION AND VALIDATION ADJUSTMENTS. ALSO, DISAGGREGATE RE-ESTIMATION OF A FEW MODELS IS UNDER TESTING AND MAY RESULT IN OTHER IMPROVEMENTS. SUCH REVISIONS WILL BE DOCUMENTED IN SUPPLEMENTARY MATERIAL. THE PRESENT REPORT, IN ITS THREE VOLUMES, REFLECTS THE VERSION OF THE MODELS DELIVERED TO MTC IN JUNE, 1977. REFERENCES 11- 1 11. REFERENCES 11.1 GENERAL REFERENCES ADLER, T.J. (1976),MODELLING NON-WORK TRAVEL PATTERN, PH.D. DISSERTATION, DEPARTMENT OF CIVIL ENGINEERING, MASSACHUSETTS INSTITUTE OF TECHNOLOGY, CAMBRIDGE, MASSACHUSETTS. ATHERTON, T.J., J.H. SUHRBIER, AND W.A. JESSIMAN (1976), "USE OF DISSAGGREGATE TRAVEL DEMAND MODELS TO ANALYZE CARPOOLING POLICY INCENTIVES," TRANSPORTATION RESEARCH RECORD 599, WASHINGTON, D.C. BEN-AKIVA, M.E. (1973), STRUCTURE OF PASSENGER TRAVEL DEMAND MODELS, PH.D. DISSERTATION, DEPARTMENT OF CIVIL ENGINEERING, MASSACHUSETTS INSTITUTE OF TECHNOLOGY, CAMBRIDGE, MASSACHUSETTS. BEN-AKIVA, M.E., AND T.J. ATHERTON (1978), "CHOICE MODEL PREDICTION OF CARPOOL DEMAND: METHOD AND RESULTS," TRANSPORTATION RESEARCH RECORD 637, WASHINGTON D.C. BEN-AKIVA, M.E., AND S.R. LERMAN, (1977), "DISAGGREGATE TRAVEL AND MOBILITY CHOICE MODELS AND MEASURES OF ACCESSIBILITY," PAPER PREPARED FOR PRESENTATION AT THE THIRD INTERNATIONAL CONFERENCE ON BEHAVIORAL MODELLING, AUSTRALIA. BEN-AKIVA, M.E., S.R. LERMAN AND M.L. MANHEIM (1976), "DISAGGREGATE MODELS: AN OVERVIEW OF SOME RECENT RESEARCH RESULTS AND PRACTICAL APPLICATIONS," PROCEEDINGS OF THE PLANNING AND TRANSPORT RESEARCH AND COMPUTATION SUMMER MEETING, LONDON. BEN-AKIVA, M.E., S.R. LERMAN AND M.L. MANHEIM (1974), "SOME ESTIMATION RESULTS OF A SIMULTANEOUS MODEL OF AUTO OWNERSHIP AND MODE CHOICE TO WORK," TRANSPORTATION, VOLUME 3, NUMBER 4. CAMBRIDGE SYSTEMATICS, INC. (1976A), "LOGIT INPUT PROGRAM: PROGRAM DOCUMENTATION", SUBMITTED TO MTC, BERKELEY, CALIFORNIA. CAMBRIDGE SYSTEMATICS, INC. (1976B), "MULTINOMIAL LOGIT ESTIMATION PACKAGE: PROGRAM DOCUMENTAITON", SUBMITTED TO MTC, BERKELEY, CALIFORNIA. CAMBRIDGE SYSTEMATICS, INC. (1976C), "SRGP - A SHORT RANGE GENERALIZED POLICY ANALYSIS PROGRAM", SUBMITTED TO MTC, BERKELY, CALIFORNIA. CHARLES RIVER ASSOCIATES, INC. (CRA, 1972), "A DISAGGREGATE BEHAVIORAL MODEL OF URBAN TRAVEL DEMAND," NATIONAL TECHNICAL INFORMATION SERVICE, SPRINGFIELD, VIRGINIA. DIXON, W.J. (1973), BMD: BIOMEDICAL COMPUTER PROGRAMS, UNIVERSITY OF CALIFORNIA PRESS, BERKELEY, CALIFORNIA. FEDERAL HIGHWAY ADMINISTRATION (FHWA, 1974), "FHWA COMPUTER PROGRAMS FOR URBAN TRANSPORTATION PLANNING," WASHINGTON, D.C. REFERENCES 11-2 GOLDNER, W. (1968), PROJECTIVE LAND USE MODEL (PLUM), BATSC TECHNICAL REPORT 219, BAY AREA TRANSPORTATION STUDY COMMISSION, BERKELEY, CALIFORNIA. HOXIE, P. (1970), AUTO OWNERSHIP AND THE LONG RUN TRANSPORTATION DECISION, S.M. THESIS, DEPARTMENT OF CIVIL ENGINEERING, MASSACHUSETTS INSTITUTE FO TECHNOLOGY, CAMBRIDGE, MASSACHUSETTS. MCFADDEN, D.L. (1974), "THE MEASUREMENT OF URBAN TRAVEL DEMAND," JOURNAL OF PUBLIC ECONOMICS, PP. 303-328. METROPOLITAN WASHINGTON COUNCIL OF GOVERNMENTS (1974), "ANALYSIS OF ZONAL LEVEL TRIP GENERATION RELATIONSHIPS," TECHNICAL REPORT #10, WASHINGTON, D.C. MUDARRI, D. (1972), MODE CHOICE FOR WORK TRIPS: A KEY URBAN TRANSPORTATION PROBLEM, PH.D. DISSERTATION, DEPARTMENT OF ECONOMICS, UNIVERSITY OF MICHIGAN, ANN ARBOR, MICHIGAN. NIE, N.H., ET AL (1975), SPSS: STATISTICAL PACKAGE FOR THE SOCIAL SCIENCES, SECOND EDITION, MCGRAW HILL, NEW YORK. RICHARDS, M.G., AND M.E. BEN-AKIVA (1975), A DISAGGREGATE TRAVEL DEMAND MODEL, SAXON HOUSE/LEXINGTON BOOKS, LEXINGTON, MASSACHUSETTS. URBAN MASS TRANSPORTATION ADMINISTRATION (UMTA, 1976), "URBAN TRANSPORTATION PLANNING SYSTEMS: REFERENCE MANUAL," WASHINGTON, D.C. 11.2 OTHER TMDP REPORTS TASK 2 TECHNICAL PAPER: MODEL PURPOSE AND USAGE, CCTOBER 1975. TASK 3 TECHNICAL PAPER: DATA BASE REVIEW, OCTOBER 1975. TASK 4 TECHNICAL PAPER: MODEL REVIEW, OCTOBER 1975. PHASE 1 FINAL REPORT, NOVEMBER 1975. PHASE 2 WORK PROGRAM, NOVEMBER 1975. TASK 6 TECHNICAL PAPER: DATA BASE, AUGUST 1976. APPENDIX A VARIABLE DIRECTORY APPENDIX A-1 APPENDIX A: VARIABLE DIRECTORY THE VARIABLE DIRECTORY THAT FOLLOWS IS A TABLE CONTAINING THREE COLUMNS: (1) VARIABLE THIS COLUMN CONTAINS THE VARIABLE MNEMONIC WITH ALL STRATIFICATION LEVEL CODES REMOVED (E.G., LNRINC10, WHICH IS LOG REMAINING INCOME FOR INCOME GROUP 1, AO LEVEL O, APPEARS AS LNRINC) THE FIRST CHARACTER OF THE NAME GIVES SOME INDICATION OF ITS GENERAL CLASSIFICATION: & - AN MTCFCAST INPUT VARIABLE $ - A MODEL PARAMETER OR ADJUSTMENT # - A DISAGGREAGE VARIABLE ALL OTHERS ARE AGGREGATE VARIABLES PRODUCED BY MTCFCAST. (2) INDICES THIS COLUMN CONTAINS CODES DESCRIBING THE DIMENSIONS OF THE VARIABLE: I - SPECIFIC TO ZONE J - SPECIFIC TO DESTINATION OR ATTRACTION ZONE G - SPECIFIC TO INCOME GROUP (1, 2, OR 3) A - SPECIFIC TO AO LEVEL (0, 1, OR 2) H - SPECIFIC TO HOUSEHOLD Q - SPECIFIC TO TRIPMAKER LNRINC, FOR EXAMPLE, HAS INDICES OF I, J, G AND A, INDICATING THAT IT VARIES ACROSS THES FOUR DIMENSIONS. (3) DESCRIPTION THIS COLUMN CONTAINS A DESCRIPTION OF THE VARIABLE, INCLUDING ITS UNITS (IF ANY) AND A STATEMENT OF HOW IT IS COMPUTED IF ITS COMPUTATION IS RELATIVELY SIMPLE. A VARIABLE IS INCLUDED HERE IF IT MEETS THE FOLLOWING CONDITIONS: (1) IT IS INPUT TO OR OUTPUT FROM AT LEAST ONE OF THE PROCESSING STEPS OF THE MTCFCAST SYSTEM WHEN APPLIED IN THE STANDARD WAY (SEE VOLUME III); OR (2) IT APPEARS IN A FINAL DISAGGREGATE MODEL FORMULAITON. NOTABLY EXCLUDED ARE OPTIONAL MTCFCAST OUTPUT VARIABLES. APPENDIX A-2 VARIABLE INDICES DESCRIPTION &AACCDA I AUTO ACCESS DISTANCE, ATTRACTION ZONE (MILES) &AACCDP I AUTO ACCESS DISTANCE, PRODUCTION ZONE (MILES) &ACRES I TOTAL AREA (ACRES) &ADIST I PEAK AUTO DISTANCE, INTRAZONALS INCLUDED (MILES*100) &AGREMP I AGRICULTURAL EMPLOYMENT (EMPLOYEES) &ATERMA I PEAK ACCESS TERMINAL TIME, ATTRACTION ZONE (MIN*100) &ATERMP I PEAK ACCESS TERMINAL TIME, PRODUCTION ZONE (MIN*100) &ATIME I J PEAK AUTO TRAVEL TIME, INTRAZONALS INCLUDED (MIN*100) &ATOLL I J PEAK AUTO TOLLS (1965 CENTS) &CBDSW I CBD SWITCH: =0 IF I IS NOT IN CBD NE 0 IF I IS IN CBD &HHSG I G NUMBER OF HOUSEHOLDS IN INCOME GROUP G (HOUSEHOLDS) &INCOME I MEAN ANNUAL GROSS HH INCOME (1965 $) &MFGEMP I MANUFACTURING EMPLOYMENT (EMPLOYEES) &NMULTI I NUMBER OF MULTI-FAMILY DWELLING UNITS (D.U.'S) &NSINGL I NUMBER OF SINGLE FAMILY DWELLING UNITS (D.U.'S) &OADIST I J OFFPEAK AUTO DISTANCE, INTRAZONALS INCLUDED (MILES*100) &OATERMA I OFFPEAK ACCESS TERMINAL TIME, ATTRACTION ZONE (MIN*100) &OATERMP I OFFPEAK ACCESS TERMINAL TIME, PRODUCTION ZONE (MIN*100) &OATIME I J OFFPEAK AUTO TRAVEL TIME, INTRAZONALS INCLUDED (MIN*100) &OATOLL I J OFFPEAK AUTO TOLLS (1965 CENTS) &OPRKCST I OFFPEAK PARKING CHARGE (1965 CENTS/HOUR) &OPTERMA I OFFPEAK PARK TERMINAL TIME, ATTRACTION ZONE (MIN*100) &OPTERMP I OFFPEAK PARK TERMINAL TIME, PRODUCTION ZONE (MIN*100) &OTACCT I J OFFPEAK TRANSIT AUTO ACCESS TIME IF WALK ACCESS NOT AVAILABLE (MIN); =0 IF WALK ACCESS IS AVAILABLE &OTFARE I J OFFPEAK TRANSIT FARES, INTRAZONALS INCLUDED (1965 CENTS); =500 IF TRANSIT SERVICE UNAVAILABLE &OTHEMP I OTHER EMPLOYMENT (EMPLOYEES) &OTIVTT I J OFFPEAK TRANSIT IN-VEHICLE TIME, INTRAZONALS INCLUDED (MIN); =0 IF TRANSIT SERVICE IS UNAVAILABLE &OTWAIT I J OFFPEAK TRANSIT IST WAIT TIME, INTRAZONALS INCLUDED (MIN) APPENDIX A-3 VARIABLE INDICES DESCRIPTION &OTWALK I J OFFPEAK TRANSIT WALK TIME, INTRAZAONAL INCLUDED (MIN) &OTXFER I J OFFPEAK TRANSIT TRANSFER TIME, INTRAZONALS INCLUDED (MIN) &OWTERMA I OFFPEAK WALK TERMINAL TIME, ATTRACTION ZONE (MIN*100) &OWTERMP I OFFPEAK WALK TERMINAL TIME, PRODUCTION ZONE (MIN*100) &PACRES I POPULATION-SERVING AREA (ACRES) &RETEMP I RETAIL EMPLOYMENT (EMPLOYEES) &SEREMP I SERVICE EMPLOYMENT (EMPLOYEES) &TACCT I J PEAK TRANSIT AUTO ACCESS TIME IF WALK ACCESS IS NOT AVAILABLE (MIN); =0 IF WALK ACCESS IS AVAILABLE &TFARE I J PEAK TRANSIT FARE, INTRAZONALS INCLUDED (1965 CENTS) =5000 IF TRANSIT SERVICE IS UNAVAILABLE &TIVTT I J PEAK TRANSIT IN-VEHICLE TIME, INTRAZONALS INCLUDED (MIN); =0 IF TRANSIT SERVICE IS UNAVAILABLE &TRDEMP I TRADE EMPLOYMENT (EMPLOYEES) &TWAIT I J PEAK TRANSIT WALK TIME, INTRAZONALS INCLUDED (MIN) &TWALK I J PEAK TRANSIT WALK TIME, INTRAZONALS INCLUDED (MIN) &TXFER I J PEAK TRANSIT TRANSFER TIME, INTRAZONALS INCLUDED (MIN) &WORKERS I NUMBER OF EMPLOYED RESIDENTS (FULL OR PART TIME) (WORKERS) &WTERMA I PEAK WALK TERMINAL TIME, ATTRACTION ZONE (MIN*100) &WTERMP I PEAK WALK TERMINAL TIME, PRODUCTION ZONE (MIN*100) $ACCRATE AUTO ACCESS OPERATING COST RATE (1965 CENTS/MIN) *NOTE* 1.137 USED IN CALIBRATION $AUTREM A MEAN # AUTOS NOT USED FOR WORK TRIPS, AO LEVEL A (AUTOS/HOUSEHOLD) $COSTPYR ANNUAL AUTO OWNERSHIP COST (1965 $/VEHICLE) *NOTE* $1000 USED IN CALIBRATION $HBOPARK HBO PARKING DURATION (HOURS) *NOTE* 2.0 USED IN CALIBRATION APPENDIX A-4 VARIABLE INDICES DESCRIPTION $HBSHDC I HBSH DISTANCE CORRECTION COEFFICIENT (30 DISTRICT VALUES) $HBSHDCD HBSH DISTANCE CORRECTION COEFFICIENT DEFAULT (-.090) $HBSHGF I HBSHG ZONAL ADJUSTMENT FACTOR (30 DISTRICT VALUES) $HBSHGFD HBSHG ZONAL ADJUSTMENT FACTOR DEFAULT (1.0) $HBSHGFR HBSHG REGIONAL ADJUSTMENT FACTOR (1.0) $HBSHMCF I HBSH AUTO MODE CORRECTION FACTOR (30 DISTRICT VALUES) $HBSHM F I J HBSH MODAL TRIP TABLE ADJUSTMENT FACTOR (30 X 30 DISTRICT PAIR VALUES FOR EACH MODE) $HBSRDC I HBSR DISTANCE CORRECTION COEFFICIENT (30 DISTRICT VALUES) $HBSRDCD HBSR DISTNACE CORRECTION COEFFICIENT DEFAULT (-.047) $HBSRGF I HBSRG ZONAL ADJUSTMENT FACTOR (30 DISTRICT VALUES) $HBSRGFD HBSRG ZONAL ADJUSTMENT FACTOR DEFAULT (1.0) $HBSRGFR HBSRG REGIONAL ADJUSTMENT FACTOR $HBSRMCF I HBSR AUTO MODE CORRECTION FACTOR (30 DISTRICT VALUES) $HBSRM F I J HBSR MODAL TRIP TABLE ADJUSTMENT FACTOR (30 X 30 DISTRICT PAIR VALUES FOR EACH MODE) $HBWAF I PHBWA AND SHBWA ZONAL ADJUSTMENT FACTOR (30 DISTRICT VALUES) $HBWAFD PHBWA AND SHBWA ZONAL ADJUSTMENT FACTOR DEFAULT (1.0) $HBWM F I J HBW MODAL TRIP TABLE ADJUSTMENT FACTOR (30 X 30 DISTRICT PAIR VALUES FOR EACH MODE) $HBWM MCF I HBW MODAL CORRECTION FACTOR FOR MODES DRIVE-ALONE (A) AND SHARED-RIDE (S) (30 DISTRICT VALUES) $HBWPARK HBW PARKING DURATION (HOURS) *NOTE* 8.0 USED IN CALIBRATION $INC MEAN GROSS ANNUAL HOUSEHOLD INCOME, BASE YEAR (1965 $) $INCA A MEAN GROSS ANNUAL HOUSEHOLD INCOME, AO LEVEL A, BASE YEAR (1965 $) $INCG G MEAN GROSS ANNUAL HOUSEHOLD INCOME, INCOME GROUP G, BASE YEAR (1965 $) $INCOME MEAN GROSS ANNUAL HOUSEHOLD INCOME (1965 $) (SOMETIMES DEFAULTS TO $INC.) $LOWINC LOW INCOME ADJUSTMENT THRESHOLD (1965 $) ($4500) $MAO A MEAN HOUSEHOLD AUTO OWNERSHIP LEVEL, AO LEVEL A (AUTOS/HOUSEHOLD) (2.236 FOR AO = 2+ WAS USED IN CALIBRATION) $MAOR REGIONAL MEAN HOUSEHOLD AUTO OWNERSHIP (AUTOS/HH) $MAXRETD MAXIMUM 1965 ZONAL RETAIL EMPLOYMENT DENSITY APPENDIX A-5 VARIABLE INDICES DESCRIPTION $MAXRETD (CONT.) (EMPLOYEES/ACRE) $MINGP MINIMUM PHBWG TRIP GENERATION RATE (WORK TRIPS/PRIMARY WORKER) (0.5) $MINGS MINIMUM SHBWG TRIP GENERATION RATE (WORK TRIPS/SECONDARY WORKER) (0.5) $NHBM F I J NHB MODAL TRIP TABLE ADJUSTMENT FACTOR (30 X 30 DISTRICT PAIR VALUES FOR EACH MODE) $NHBPARK NHB PARKING DURATION (HOURS) *NOTE* 1.0 USED IN CALIBRATION $OPADJ FORECAST/1965 REAL $ AUTO OPERATING COST RATIO $PEREXP PER CAPITA LIVING EXPENSES (1965 $) *NOTE* $750 USED IN CALIBRATION $PHBWAFR PHBWA REGIONAL ADJUSTMENT FACTOR (1.0217 USED FOR 1965) $PHBWDC I PHBW DISTANCE CORRECTION COEFFICIENT (30 DISTRICT VALUES) $PHBWDCD PHBW DISTANCE CORRECTION COEFFICIENT DEFAULT (-.060) $PHBWGF I PHBWG ZONAL ADJUSTMENT FACTOR (30 DISTRICT VALUES) $PHBWGFD PHBWG ZONAL ADJUSTMENT FACTOR DEFAULT (1.0) $PHBWGFR PHBWG REGIONAL ADJUSTMENT FACTOR (.9335 USED FOR 1965) $PHH MEAN SIZE OF HOUSEHOLDS, BASE YEAR (PERSONS/HOUSEHOLD) $PHHA A MEAN SIZE OF HOUSEHOLDS IN AO LEVEL A (PERSONS/HOUSEHOLD) $PHHA G MEAN SIZE OF HOUSEHOLDS IN INCOME GROUP G (PERSONS/HOUSEHOLD) $PICKUP SHARED-RIDE PICKUP PENALTY (MIN) *NOTE* 5.0 USED IN CALIBRATION $RATRET FORECAST/1965 RETAIL EMPLOYMENT RATIO $SHBWAFR SHBWA REGIONAL ADJUSTMENT FACTOR (1.043 USED FOR 1965) $SHBWDC I SHBW DISTANCE CORRECTION COEFFICIENT (30 DISTRICT VALUES) $SHBWDCD SHBW DISTANCE CORRECTION COEFFICIENT DEFAULT (-.043) $SHBWGF I SHBWG ZONAL ADJUSTMENT FACTOR (30 DISTRICT VALUES) $SHBWGFD SHBWG ZONAL ADJUSTMENT FACTOR DEFAULT (1.0) $SHBWGFR SHBWG REGIONAL ADJUSTMENT FACTOR (1.0269 USED FOR 1965) $SROCCF SROCC REGIONAL ADJUSTMENT FACTOR (.944 USED FOR 1965) $WHH MEAN NUMBER OF WORKERS PER HOUSEHOLD, BASE YEAR (WORKERS/HOUSEHOLD) $WHHA A MEAN NUMBER OF WORKERS PER HOUSEHOLD IN AO APPENDIX A-6 VARIABLE INDICES DESCRIPTION $WHHA (CONT.) LEVEL A, BASE YEAR (WORKERS/HOUSEHOLD) #ACCDUM Q AUTO ACCESS DUMMY FOR WORK TRIP =0 IF WALK ACCESS TO TRANSIT IS AVAILABLE =I IF WALK ACCESS TO TRANSIT IS NOT AVAILABLE #ACCDUM2 Q #CARS/#WHH FOR TRIPS REQUIRING AUTO ACCESS TO TRANSIT =(#CARS(H(Q))/#WHH(H(Q)))*ACCDUM #ACOSTI Q PEAK AUTO TRAVEL COST/INCOME =(&ATOLL(I(Q),J(Q))+$OPADJ *(PRATT(&ATIME(I(Q)), *ADIST(I(Q),J(Q)))/100+PRATT(ATERMP(I(Q))) AACCDP(I(Q)))+PRATT(ATERMA(J(Q)),AACCDA(J(Q))) +.5*$HBWPARK*PRKCST(J(Q)))/(#INCOME(H(Q))*100) #AUTREM H NUMBER OF AUTOS REMAINING AFTER SUBTRACTING ONE FOR EACH WORKER USING AN AUTO FOR HIS WORK TRIP (AUTOS) #BW H BREADWINNER DUMMY: =0 IF Q IS NOT THE BREADWINNER =1 IF Q IS THE BREADWINNER #CARS H NUMBER OF AUTOS OWNED (AUTOS) #DINC H DISPOSABLE ANNUAL HH INCOME (1965 $) =#INCOME-$750*#PHH #FINC H LOW INCOME ADJUSTMENT (1965 $) =MAX((4500-#INCOME),0) #HBSHGV H DAILY HOME-BASED SHOPPING TRIPS PRODUCED BY HOUSEHOLD H (TRIPS) #HBSRGV H DAILY HOME-BASED SOCIAL/RECREATIONAL TRIPS PRODUCED BY HOUSEHOLD H (TRIPS) #HHTCOST J Q SIZE OF SHOPPER Q'S HOUSEHOLD TIMES THE 1-WAY OFFPEAK TRANSIT FARE FROM Q'S RESIDENCE ZONE TO ZONE I (1965 PERSON-CENTS) =#PHH(H(Q))*OTFARE(I(Q),J) #HOUSE H HOUSING TYPE CODE: =1 FOR SINGLE-FAMILY DWELLING =0 FOR MULTIPLE-FAMILY DWELLING #INCATIM J Q NATURAL LOG OF 1-WAY OFFPEAK AUTO TRAVEL TIME, MULTIPLIED BY HOUSEHOLD INCOME (1965 $-MINUTES) =#INCOME(H(Q))*LNATIME(I(Q),J) #INCOME H GROSS ANNUAL HOUSEHOLD INCOME (1965 $) #INCTTIM J Q NATURAL LOG OF 1-WAY OFFPEAK TRANSIT TRAVEL TIME MULTIPLIED BY HOUSEHOLD INCOME (1965 $-MINUTES) =#INCOME(H(Q))*LNTTIME(I(Q),J) #INVPHH H INVERSE OF #PHH (HOUSEHOLDS/PERSON) =1/#PHH #LNRINC H A NATURAL LOG OF REMAINING INCOME (1965 $) #LPCDINC H NATURAL LOG OF PER CAPITA DISPOSABLE INCOME (1965 $) =LN(#INCOME/#PHH - 750) APPENDIX A-7 VARIABLE INDICES DESCRIPTION #LSUMSH H NATURAL LOG OF THE DENOMINATOR OF THE DISAGGREGATE HBSHDM MODEL FOR HOUSEHOLD H #LSUMSR H NATURAL LOG OF THE DENOMINATOR OF THE DISSAGGREGATE HBSRDM MODEL FOR HOUSEHOLD H #NOWORK H NUMBER OF NON-WORKERS (PERSONS) #PHBWGV H NUMBER OF PRIMARY HOME-BASED-WORK TRIPS (TRIPS) #PHH H HOUSEHOLD SIZE (PERSONS) #REXP H A RATIO OF EXPONENTIATED UTILITIES OF PHBWM MODEL, TRANSIT/AUTO, FOR AO LEVEL A #RSUMSHA H A RATIO OF DENOMINATORS OF HBSHD MODEL, TRANSIT/AUTO, FOR AO LEVEL A #SCOSTI Q PEAK SHARED-RIDE TRAVEL COST/INCOME =#ACOSTI/#SROCC #SHBWGV Q DAILY HOME-BASED WORK TRIPS PRODUCED BY SECONDARY WORKER Q (TRIPS) #SHBWML I Q NATURAL LOG OF THE DENOMINATOR OF THE DISAGGREGATE SHBWM MODEL FOR DESTINATION ZONE I, WORKER Q #SROCC Q HBW SHARE-RIDE AUTO OCCUPANCY (PERSONS/VEHICLE) #TCOSTI Q PEAK TRANSIT TRAVEL COST/INCOME =(&TFARE(I(Q),J(Q))+$OPADJ*$ACCRATE *&TACCT(I(Q),J(Q)))/(#INCOME(H(Q))*100) #WHH H NUMBER OF WORKERS (WORKERS) #WLSUM I Q WEIGHTED PHBWM LOGSUM (SEE PHBWC MODEL) AACCDA I AUTO ACCESS DISTANCE, ATTRACTION ZONE (MILES) AACCDP I AUTO ACCESS DISTANCE, PRODUCTION ZONE (MILES) AACCDUM I J PEAK AUTO ACCESS DUMMY =1 IF &TACCT > 0 =0 OTHERWISE ACCDUM2 I J INVWHH*MAO(AO) FOR HHS REQUIRING AUTO ACCESS (AUTOS/WORKER) =ACCDUM*INVWHH(I)*$MAO(AO) ACOSTI I J PEAK AUTO TRAVEL COST/HH INCOME =(&ATOLL+$OPADJ*(PRATT(&ATIME,&ADIST)/100 +PRATT(ATERMP(I),AACCDP(I)) +PRATT(ATERMA(J),AACCDA(J)) +.5*$HBWPARK*PRKCST(J))/(INCOME(I)*100) ACRES I TOTAL AREA (ACRES) =&ACRES AGREMP I AGRICULTURAL EMPLOYMENT (EMPLOYEES) =&AGREMP AHMAV I J ANY-TO-HOME PERSON TRIPS, AUTO (TRIPS) AHMTV I J ANY-TO-HOME PERSON TRIPS, TRANSIT (TRIPS) AODFOUT I HB O/D OUTBOUND SPLITTING FACTOR, AUTO ATERMA I PEAK ACCESS TERMINAL TIME, ATTRACTION ZONE (MIN) =&ATERMA/100 ATERMP I PEAK ACCESS TERMINAL TIME, PRODUCTION ZONE (MIN) =&ATERMP/100 APPENDIX A-8 VARIABLE INDICES DESCRIPTION ATIMET I PEAK AUTO IN-VEHICLE TRAVEL TIME (MIN) =&ATIME/100+ATERMP(I)+PTERMP(I)+ATERMA(J) +PTERMA(J) CBD I CBD SWITCH: =1 IF &CBDSW NE O, I.E., IF ZONE I IS IN CBD =0 IF &CBDSW = O, I.E. IF ZONE I IS NOT IN CBD DINC I DISPOSABLE ANNUAL HH INCOME (1965 $) =MAX((INCOME-$PEREXP*PHH),0) DINCG I G DISPOSABLE ANNUAL HH INCOME, INCOME GROUP G (1965 $) =MAX((INCOMEG -$PEREXP*PHHG ),0) FINC I G LOW INCOME ADJUSTMENT (1965 $) =MAX(($LOWINC-INCOME G),0) HAMAV I J HOME-TO ANY PERSON TRIPS, AUTO (TRIPS) HAMTV I J HOME-TO ANY PERSON TRIPS, TRANSIT (TRIPS) HBMAI I HB PERSON TRIP PRODUCTIONS, AUTO (TRIPS) =PHBWMAI1+PHBWMAI2+PHBWMSIO+PHBWMSI1+PHBWMSI2 +SHBWMAI1+SHBWMAI2+SHBWMSIO+SHBWMSI1+SHBWMSI2 +HBSHMAIO+HBSHMAI1+HBSHMAI2 +HBSRMAIO+HBSRMAI1=HBSRMAI2 HMBAJ J HB PERSON TRIP ATTRACTIONS, AUTO (TRIPS) =PHBWMAJ1+PHBWMAJ2+PHBWMSIO+PHBWMSI1+PHBWMSI2 +SHBWMAJ1+SHBWMAJ2+SHBWMSIO+SHBWMSI1+SHBWMSI2 +HBSHMAJO+HBSHMAJ1+HBSHMAJ2 +HBSRMAJO+HBSRMAJI+HBSRMAJ2 HBMAV I J HB P/A PERSON TRIPS, AUTO (TRIPS) =HBWMAV+HBWMSV+HBOMAV HBMAVOD I J HB O/D PERSON TRIP PRODUCTIONS, TRANSIT (TRIPS) =HAMAV+TR(AHMAV) HBMTI I HB PERSON TRIP PRODUCTIONS, TRANSIT (TRIPS) =PHBWMTIO+PHBWMTI1+PHBWMTI2+SHBWMTIO+SHBWMTI1 +SHBWMTI2+HBSHMTIO+HBSHMTI1+HBSHMTI2+HBSRMTIO +HBSRMTI1+HBSRMTI2 HBMTV I J HB P/A PERSON TRIPS, TRANSIT (TRIPS) =HBWMTV+HBOMTV HBMTVOD I J HB O/D PERSON TRIPS, TRANSIT (TRIPS) =HAMTV+TR(AHMTV) BHOACOST I J TOTAL AUTO TRAVEL COST FOR INBOUND =&OATOLL+$OPADJ*(PRATT(&OATIME,&OADIST)/100 +PRATT(OATERMP(I),AACCDP(I)) +PRATT(OATERMA(J),AACCDA(J))) +$HBOPARK*OPRKCST(J) HBOMAV I J HBO P/A PERSON TRIPS, AUTO (TRIPS) APPENDIX A-9 VARIABLE INDICES DESCRIPTION HBOMAV (CONT.) =HBSHMAV+HBSRMAV HBOMTV I J HBO P/A PERSON TRIPS, TRANSIT (TRIPS) =HBSHMTV+HBSRMTV HBSHDCF I J HBSHDM DISTANCE CORRECTION FACTOR =$HBSHDC(I)*MAX((&OADIST-7),0) HBSHGV I HBSH PERSON TRIP PRODUCTIONS, AO LEVEL A (TRIPS) HBSHMAI I A HBSH PERSON TRIP PRODUCTIONS, AUTO, AO LEVEL A (TRIPS) HBSHMAJ I A HBSH PERSON TRIP ATTRACTIONS, AUTO, AO LEVEL A (TRIPS) HBSHMAU I A HBSHDM UTILITY W/O DIST. CORRECTION, AUTO AO LEVEL A HBSHMAV I J A HBSH P/A PERSON TRIPS, AUTO, AO LEVEL A (TRIPS) HBSHMAVT I J HBSH P/A PERSON TRIPS, AUTO (TRIPS) =HBSHMAVO+HBSHMAV1+HBSHMAV2 HBSHMTI I A HBSH PERSON TRIP PRODUCTIONS, TRN, AO LEVEL A (TRIPS) HBSHMTU I A HBSHDM UTILITY W/O DIST. CORRECTION, TRN, AO LEVEL A HBSHMTV I J A HBSH P/A PERSON TRIPS, TRN, AO LEVEL A (TRIPS) HBSHMTVT I J HBSH P/A PERSON TRIPS, TRN (TRIPS) =$HBSRDC(1)*MAX((&OADIST-7),0) HBSDCF I J HBSRDM DISTANCE CORRECTION FACTOR =$HBSRDC(1)*MAX((&OADIST-7),0) HBSRGV I HBSR PERSON TRIP PRODUCTIONS (TRIPS) =HBSRGVO+HBSRGV1+HBSRGV2 HBSRGVA I A HBSR PERSON TRIP PRODUCTIONS, AO LEVEL A (TRIPS) HBSRMAI I A HBSR PERSON TRIP PRODUCTIONS, AUTO, AO LEVEL A (TRIPS) HBSRMAJ I A HBSR PERSON TRIP ATTRACTIONS, AUTO, AO LEVEL A (TRIPS) HBSRMAU I A HBSRDM UTILITY W/O DIST. CORRECTION, AUTO, AO LEVEL A HBSRMAV I J A HBSR P/A PERSON TRIPS, AUTO, AO LEVEL A (TRIPS) HBSRMAVT I J HBSR P/A PERSON TRIPS, AUTO (TRIPS) =HBSRMAVO+HBSRPAV14HESRPAVZ HBSRMTI I A HBSR PERSON TRIP PRODUCTIONS, TRN AO LEVEL A (TRIPS) HBSRMTJ I A HBSR PERSON TRIP ATTRACTIONS, TRN AO LEVEL A (TRIPS) HBSRMTU I A HBSROM UTILITY W/O DIST. CCRRECTICK, TRN, AO LEVEL A HBSRMTV I J A HBSR P/A PERSON TRIPS, TRN, AO LEVEL A (TRIPS) HBSRMTVI I J HBSR P/A PERSON TRIPS, IRN (TRIPS) APPENDIX A-10 VARIABLE INDICES DESCRIPTION HBSRMTVT (CONT.) =HBSRMTV0+HBSRMTV1+HBSRMTV2 HBWA I HBW ATTRACTIONS (TRIPS) =PHBWA+SHBWA HBWMAV I J HBW P/A PERSON TRIPS, AUTO (TRIPS) =PHBWMAV+SHBWMAV HBWMSV I J HBW P/A PERSON TRIPS, SHARED-RIDE (TRIPS) =PHBWMSV+SHBWMSV HBWMTV I J HBW P/A PERSON TRIPS, TRANSIT (TRIPS) =PHBWMTV+SHBWMTV HHS I NUMBER OF HOUSEHOLDS (HOUSEHOLDS) =&HHSG1+&HHSG2+&HHSG3 HHSA I A NUMBER OF HOUSEHOLDS IN AO LEVEL A (HOUSEHOLDS) HHSG I G NUMBER OF HOUSEHOLDS IN INCOME GROUP G (HOUSEHOLDS) HHTCOST I J TOTAL 1-WAY OFFPEAK TRANSIT COST, MULTIPLIED BY AVERAGE HOUSEHOLD SIZE (1965 PERSON-CENTS/HOUSEHOLD) =PHH(I)*OTFARE HOUSE I FRACTION OF HOUSEHOLDS IN SINGLE FAMILY DWELLINGS =&SINGL/T IF T > 0 =0 OTHERWISE WHERE T=(&NSINGL+&NMULTI) INCATIME I J NATURAL LOG OF 1-WAY OFFPEAK AUTO TRAVEL TIME, MULTIPLIED BY 1965 REGIONAL HOUSEHOLD INCOME (1965 $-MINUTES) =$INCOME*LNATIME INCOME I MEAN GROSS ANNUAL HOUSEHOLD INCOME (1965 $) =&INCOME IF &INCOME > 0 =$INCOME OTHERWISE INCOMEA I A MEAN GROSS ANNUAL HOUSEHOLD INCOME IN AO LEVEL A (1965 $) INCOMEG I G MEAN GROSS ANNUAL HOUSEHOLD INCOME IN INCOME GROUP G (1965 $) INCTTIME I J NATURAL LOG OF 1-WAY OFFPEAK TRANSIT TRAVEL TIME, MULTIPLIED BY 1965 REGIONAL HOUSEHOLD INCOME (1965 $-MINUTES) =$INCOME*LNTTIME INVPHH I INVERSE OF MEAN HOUSEHOLD SIZE (HOUSEHOLDS/PERSON) =1/PHH IF PHH > 0 =0 OTHERWISE INVPHHA I A INVERSE OF MEAN HOUSEHOLD SIZE FOR AO LEVEL A (HOUSEHOLDS/PERSON) =1/PHHA INVSROCC I J INVERSE OF SHARED-RIDE OCCUPANCY (1000*VEHICLES/PERSON) INVWHH I INVERSE OF WORKERS PER HOUSEHOLD (HOUSEHOLDS/WORKER) APPENDIX A-11 VARIABLE INDICES DESCRIPTION INVWHH (CONT.) =1/WHH IF WHH > 0 =0 OTHERWISE LNATIME I J NATURAL LOG OF TOTAL 1-WAY OFFPEAK AUTO TRAVEL TIME (MINUTES) = LN(T) IF T > 0 =0 OTHERWISE WHERE T=&OATIME/100+OWTERMP(I)+OPTERM(I) +OATERMP(I)+OWTERMA(J) +OPTERMA(J)+OATERMA(J) LNEMP I NATURAL LOG OF TOTAL EMPLOYMENT IN 1000'S (EMPLOYEES/1000) =LN(TOTEMP/1000) IF TOTEMP > 0 =-16O OTHERWISE LNPHBWA I NATURAL LOG OF PHBW ATTRACTIONS (TRIPS) =LN(PHBWA) IF PHBWA > 0 =-160 OTHERWISE LNPHBWAX I NATURAL LOG OF INTERMEDIATE PHBW ATTRACTIONS FOR BALANCING THE PHBWD MODEL (TRIPS) LNPOPEMP I NATURAL LOG OF POPULATION/RETAIL EMPLOYMENT (PERSONS/EMPLOYEE) =LN(POP/RETEMP) IF POP > 0, RETEMP > 0 =LN(POP) IF POP > 0, RETEMP = C =LN(1/RETEMP) IF POP = 0, RETEMP > 0 =0 OTHERWISE LNPTEMP I NATURAL LOG OF POPULATION/EMPLOYMENT (PERSONS/EMPLOYEE) =LN(POP/TOTEMP) IF POP > 0, TOTEMP > 0 =LN(POP) IF POP > 0, TOTEMP = C =LN(1/TOTEMP) IF POP = C, TOTEMP > 0 =0 OTHERWISE LNRETEMP I NATURAL LOG OF RETAIL EMPLOYMENT (EMPLOYEES) =LN(RETEMP) IF RETEMP > 0 =0 OTHERWISE LNRINC I J G A NATURAL LOG OF REMAINING INCOME, INCOME GROUP G, AO LEVEL A (1965 $) =LN(DINCG(I,G)-500*INCOME(I) *(PHBWMTP(I,J,A)*TCOSTI(I,J)+PHBWMSP(I,J,A) *SCOSTI(I,J)+PHBWMAP(I,J,A) *ACOSTI(I,J))-#MAO(A)*#COSTPYR) LNSERD I NATURAL LOG OF SERVICE EMPLOYMENT DENSITY (EMPLOYEES/ACRE) =LN(SEREMP/ACRES) IF SEREMP > 0, ACRES > 0 =-160 OTHERWISE LNSHBWA I NATURAL LOG OF SECONDARY HBW ATTRACTIONS (TRIPS) =LN(SHBWA) IF SHBWA > 0 =-160 OTHERWISE LNSHBWAX I NATURAL LOG OF INTERMEDIATE SHBW ATTRACTIONS FOR BALANCING THE SHBWD MODEL (TRIPS) APPENDIX A-12 VARIABLE INDICES DESCRIPTION LNTTIME I J NATURAL LOG OF TOTAL 1-WAY OFFPEAK TRANSIT TRAVEL TIME (MINUTES) =LN(T) IF &OTIVTT ) O, &OTFARE < 5000 =255 OTHERWISE WHERE T=&OTWALK+&OTACCT+#OTIVTT+&OTWAIT+&OTXFER LPCDINC I G NATURAL LOG OF PER CAPITA DISPOSABLE INCOME FOR ZONE 1, INCOME GROUP G (1965 $/PERSON) LSUMSHA I A NATURAL LOG OF THE DENOMINATOR OF THE HBSHDM MODEL, ZONE I , AO LEVEL A LN(SUMXSHA) IF SUMXSHA > 0 = -160 OTHERWISE LSUMSHAX I NATURAL LOG OF THE DENOMINATOR OF THE HBSHD MODEL, AUTO MODE, ZONE 1, AO LEVEL 0 =LN(SUMXSHAX) LSUMSHTX I NATURAL LOG OF THE DENOMINATOR OF THE HBSHD MODEL, TRANSIT MODE ZONE 1, AO LEVEL C =LN(SUMXSHTX) IF SUMXSHTX > 0 = -160 OTHERWISE LSUMSRA I A NATURAL LOG OF THE DENOMINATOR OF THE HBSRDM MODEL, ZONE I , AO LEVEL A =LN(SUMXSRA) IF SUMXSRA > 0 = -160 OTHERWISE MFCLMP I MANUFACTURING EMPLOYMENT (EMPLOYEES) =MAX(&MFGEMP,I) NHUABASE I NHB TRIP GENERATION BASE, AUTO (TRIPS) NHBACOST I J OFFPEAK AUTO COST VARIABLE FOR NHB MODELS (1965 CENTS) =&OATOLL*$OAPDJ*(PRATT($OATIME,&OADIST)/100 +PRATT(OATERMP(I),AACCDP(I)) +PRATT(OATERMA(J),AACCDA(J))) +&4NHBPARK*OPRKCST(J) NHOADP I NHB TRIP/DESTINATION CHOICE PROBABILITY, AUTO NHBADV I J NHB O/D PERSON TRIPS, AUTO (TRIPS) NHEAFOP I J NHB F=0 PROBABILITY, AUTO NHBTBASE I NHB TRIP GENERATION BASE, TRANSIT (TRIPS) NHBTDP I J NHB TRIP/DESTINATION CHOICE PROBABILITY, TRANSIT NHBTDV I J NHB O/D PERSON TRIPS, TRANSIT (TRIPS) NHBTFOF I J NHB F=0 PROBABILITY, TRANSIT NMULTI I NUMBER OF MULTI-FAMILY DWELLING UNITS (D.U.'S) =&NMULTI NDWORKA I A NUMBER OF NON-WORKERS PER HOUSEHOLD AO LEVEL A (PERSONS/HOUSEHOLD) NSINGL I NUMBER OF SINGLE FAMILY DWELLING UNITS (D.U.'S) =&NSINGL NWAOV I G A NUMBER OF NON-WORKING HOUSEHOLDS IN INCOME GROUP G BELONGING TO AO LEVEL A (HOUSEHOLDS) NWHHSG I G NUMBER OF HOUSEHOLDS WITH NO WORKERS, INCOME GROUP G (HOUSEHOLDS) APPENDIX A-13 APPENDIX A-VARIABLE INDICES DESCRIPTION OATERMA I OFFPEAK ACCESS TERMINAL TIME ATTRACTION ZONE (MIN) =&OATERMA/100 GATERMP OFFPEAK ACCESS TERMINAL TIME PRODUCTION ZONE (MIN) =&OATERMP/100 ODIST7 I J OFFPEAK DISTANCE CORRECTION (MILES) =MAX((&OADIST-7),0) ORKCST I OFFPEAK PARKING CHARGE (1965 CENTS/HOUR) OPTERMA I OFFPEAK PARK TERMINAL TIME, ATTRACTION ZONE (MIN) =&OPTERMA/100 OPTERMP I OFFPEAK PARK TERMINAL TIME, PRODUCTION ZONE (MIN) =&OPTERMP/100 OTFARE I J OFFPEAK 1-WAY TRANSIT FARE (1965 CENTS) =&OTFARE IF &OTIVTT > 0, &OTFARE < 5000 =5000 OTHERWISE OTHLMP I OTHER EMPLOYMENT (EMPLOYEES) =&OTHEMP OWTERMA I OFFPEAK WALK TERMINAL TIME, ATTRACTION ZONE (MIN) =&OWTERMA/100 OWTERMP I OFFPEAK WALK TERMINAL TIME, PRODUCTION ZONE (MIN) =&OWTERMP/100 PACRES I POPULATION-SERVING AREA (ACRES) =&PACRES PDIST3 I J PEAK DISTANCE CORRECTION (MILES) MAX((&ADIST-3),0) PEREXP I NECESSARY ANNUAL PER CAPITA EXPENDITURES (1965 $) =&PEREXP FOR ALL ZONES PHBWA I PRIMARY WORKER HBW ATTRACTIONS (TRIPS) PHHWAV I J A PHBW P/A PERSON TRIPS, AO LEVEL A (TRIPS) PHBWAX I INTERMEDIATE PHBW ATTRACTIONS FOR BALANCING THE PHBWD MODEL (TRIPS) PHOWDCF I J PHBW DISTANCE CORRECTION FACTOR =&PHBWDC(I)*MAX((#ADIST-3),0) PHBWDV I J G PHBW P/A PERSON TRIPS, INCOME GROUP G (TRIPS) PHBWGV I PRIMARY WORKER HBW PRODUCTIONS (TRIPS) =PHBWGVGI+PHBWGVG2+PHBWGVG3 PHBWGVG I G PRIMARY WORKER HBW PRODUCTIONS, INCOME GROUP G (TRIPS) PHBWMAI I A PHBW PERSON TRIP PRODUCTIONS, AUTO, AO LEVEL A (TRIPS) *NOTE* BY DEFINITION, PHBWMAI0=0 FOR ALL ZONES PHBWMAJ I A PHBW PERSON TRIP ATTRACTIONS, AUTO, AO LEVEL A (TRIPS) *NOTE* BY DEFINITIONT PHBWMAJ0=0 FOR ALL ZONES PHBWMAP I J A AUTO PROBABILITY, PHBWM MODEL, AO LEVEL A *NOTE* BY DEFINITION, PHBWMAP0=0 FOR ALL I,J PHBWMAU I J A AUTO UTILITY, PHBWM MODEL, AO LEVEL A APPENDIX A-14 VARIABLE INDICES DESCRIPTION PHBWMAU(CONT.) *NOTE* BY DEFINITION, PHBWMAU0=-160 FOR ALL I,J PHBWMAV I J A PHBW P/A PERSON TRIPS, AUTO, AO LEVEL A (TRIPS) PHBWMAVT I J PHBW P/A PERSON TRIPS, AUTO (TRIPS) PHBWMAV1*PHBINMAV2 PHBWML I J A NATURAL LOG OF THE DENOMINATOR OF THE PHBWM MODEL, ZONE PAIR I-J, AO LEVEL A. PHBWMSI I A PHBW PERSON TRIP PRODUCTION, SHARED-RIDE, AO LEVEL A (TRIPS) PHBWMSJ I A PHBW PERSON TRIP ATTRACTIONS, SHARED-RIDE, AO LEVEL A (TRIPS) PHBWMSP I J A SHARED-RIDE PROBABILITY, PHBWM MODEL, AO LEVEL A PHBWMSU I J A SHARED-RIDE UTILITY, PHBWM MODEL, AO LEVEL A PHBWMSV I J A PHBW P/A PERSON TRIPS, SHARED-RIDE, AO LEVEL A (TRIPS) PHBWMSVI I J PHBW P/A PERSON TRIPS, SHARED-RIDE (TPIPS) =PHBWMSVO+PHBWSV1+PHBWSV2 PHBWMTI I A PHBW PERSON TRIP PROUDUCTIONS, TRANSIT, AO LEVEL A (TRIPS) PHBWMTJ I A PHBW PERSON TRIP ATTRACTIONS, TRANSIT, AO LEVEL A (TRIPS) PHBWMTP I J A TRANSIT PROBABILTY, PHBWM MODEL, AO LEVEL A PHBWMTU I J A TRANSIT UTILITY, PHBWM MODEL, AO LEVEL A PHBWMTV I J A PHBW P/A PERSON TRIPS, TRANSIT, AO LEVEL A (TRIPS) PHBWMTVT I J PHBW P/A PERSON TRIPS, TRANSIT (TRIPS) =PHBWMTV0+PHBWMTV1+PHBWMTV2 PHH I MEAN SIZE OF HOUSEHOLDS (PERSONS/HOUSEHOLD) =POP/HHS IF HHS > 0 =0 OTHERWISE PHHA I A MEAN SIZE OF HOUSEHOLDS IN AO LEVEL A (PERSONS/HH) =POPA/HHSA IF HHSA > 0 =0 OTHERWISE PHHG I G MEAN SIZE OF HOUSEHOLDS IN INCOME GROUP G (PERSONS/HH) =POPG/HHSG IF HHSG > 0 =0 OTHERWISE PICKUP I SHARED-RIDE PICKUP PENALTY TIME (MINUTES) =#PICKUP FOR ALL ZONES POP I POPULATION (PERSONS) =&POP POPA I A POPULATION IN AO LEVEL A (PERSONS) =MAX((#PHHA(A)*HHSA*POP(I)+Q)/T), (SWORKA+WHHSA)) IF T > 0 =0 OTHERWISE WHERE T=($PHHA(A)*HHSA), SUMMED OVER ALL AO LEVELS) Q=((MAX((SWORKA+WHHSA-$PHHA(A) APPENEIX A-15 VARIABLE INDICES DESCRIPTION POPA(CONT.) *HHSA/T),0)), SUMMED OVER ALL AO LEVELS) POPDEN I POPULATION DENSITY (PERSONS/ACRE) =POP/ACRES IF ACRES > 0 =0 OTHERWISE POPG I G POPULATION IN INCOME GROUP G (PERSONS) =&PHHG(G)*HHSG*PCP(I)/T IF T > 0 =0 OTHERWISE WHERE T=(($PHHG(G)*HHSG), SUMMED OVER INCOME GROUPS) PRKCST I PEAK PARKING CHARGE (1965 CENTS/HOUR) =&PRKCST PSREMP I POPULATION-SERVING EMPLOYMENT (EMPLOYEES) =MAX((&RETEMP+&SEREMP+&OTHEMP),I) PTERMA I PEAK PARK TERMINAL TIME, ATTRACTION ZONE (MIN) =&PTERMA/100 PTERMP I PEAK PARK TERMINAL TIME, PRODUCTION ZONE (MIN) =&PTERMP/100 RACRES I RESIDENTIAL AREA (ACRES) =&RACRES RETEMP I RETAIL EMPLOYMENT (EMPLOYEES) =&RETEMP RETEMPD I RETAIL EMPLOYMENT DENSITY (EMPLOYEES/ACRE) =MIN((RETEMP/ACRES),($MAXRETD*$RATRET)) IF ACRES > 0 =0 OTHERWISE RETEMPDP I RETAIL EMPLOYMENT DENSITY, POPULATION-SERVING (EMPLOYEES/ACRE) =RETEMP/PACRES IF PACRES > 0 =0 OTHERWISE RETSERD I RETAIL AND SERVICE EMPLOYMENT DENSITY (EMPLOYEES/ACRE) =(RETEMP+SEREMP)/ACRES IF ACRES > 0 =0 OTHERWISE REXP I J A RATIO OF EXPONENTS OF PHBWM UTILITIES, AO LEVEL A =PHBWMTP/T IF T > 0 =0 OTHERWISE WHERE T=PHBWMAP+PHBWMP RSUMSH I RATIO OF DENOMINATORS OF HBSHD MODEL, TRANSIT/AUTO =SUMXSHTX/SUMXSHAX If SUMXSHAX > 0 =0 OTHERWISE SCOSTI I PEAK SHARED-RICE TRAVEL COST/INCOME =ACOSTI*INVSROCC/1000 SEREMP I SERVICE EMPLOYMENT (EMPLOYEES) =&SEREMP SHBWA I SECONDARY WORKER HBW ATTRACTION (TRIPS) SHBWAX I INTERMEDIATE SHOW ATTRACTIONS FOR BALANCING THE APPENDIX A-16 VARIABLE INDICES DESCRIPTION SHBWAX (CONT.) SHBWD MODEL (TRIPS) SHHWDCF I J SHBW DISTANCE CCRRECTION FACTOR =&SHBWDC(l)*MAX((&ADIST-3),0) SHEWDV I J A SHBW P/A PERSON TRIPS, AO LEVEL A (TRIPS) SHBWGV I SECONDARY WORKER HBW PRODUCTIONS (TRIPS) =SHBWGVA0+SHBWGVA1+SHBWGVA2 SHBWGVA I A SECONDARY WORKER HBW PROOUCTION, AO LEVEL A (TRIPS) SHBWMAI I A SHBW PERSON TRIP PRODUTIONS, AUTO, AO LEVEL A (TRIPS) *NOTE* BY DEFINITION , SHBWMAI0=0 FOR ALL ZONES SHBWMAJ I A SHBW PERSON TRIP ATTRACTIONS, AUTO, AO LEVEL A (TRIPS) *NOTE* BY DEFINITION, SHBWMAJ0=0 FOR ALL ZONES SHBWMAV I J A SHBW P/A PERSON TRIPS, AUTO, AO LEVEL A (TRIPS) SHBWMAVT I J SHBW P/A PERSON TRIPS, AUTO (TRIPS) =SHBWMAV1+SHBWMAV2 SHBWML I J A NATURAL LOG OF THE DENOMINATOR OF THE SHBWM MODEL, ZONE I-J, AO LEVEL A SHBWMSI I A SHBW PERSON TRIP PRODUCTIONS, SHARED-RIDE, AO LEVEL A (TRIPS) SHBWMSJ I A SHBW PERSON TRIP ATTRACTIONS, SHARED-RIDE, AO LEVEL A (TRIPS) SHBWMSV I J A SHBW P/A PERSON TRIPS, SHARED-RIDE, AO LEVEL A (TRIPS) SHBWMSVT I J SHBW P/A PERSON TRIPS, SHARED-RIDE (TRIPS) =SHBWMSV0+SHBWMSV1+SHBWPSV2 SHBWMTI I A SHBW PERSON TRIP PRODUCTIONS, TRANSIT, AO LEVEL A (TRIPS) SHBMTJ I A SHBW PERSON TRIP ATTRACTIONS, TRANSIT, AO LEVEL A (TRIPS) SHBWMTV I J A SHBW P/A PERSON TRIPS, TRANSIT, AO LEVEL A (TRIPS) SHBWMTVT I J SHBW P/A PERSON TRIPS, TRANSIT (TRIPS) =SHBWMTV0+SHBWMTV1+SHBWMTV2 SROCC I J HOME-BASED WORK TRIP SHARED-RIDE AUTO CCCUPANCY (PERSONS/VEHICLE) STIMET I J PEAK SHARED-RIDE IN-VEHICLE TRAVEL TIME (MINUTES) =ATIMET+PICKUP(I) SUMANDA I SUM OF NHBAD ATTRACTION PROBABILITIES*200000 SUMANDT I SUM OF NHBTD ATTRACTION PROBABILIIIES*200000 SUMANOA I SUM OF NHBAO ATTRACTION PROBABILITlES*200000 SUMANOT I SUM OF NHBTO ATTRACTION PROBABILITIES*200000 SUMPNDA I SUM OF NHBAD PRODUCTION PROBABILITIES*200000 SUMPNDT I SUM OF NHBTD PRODUCTION PROBABILITIE@*200000 SUMPNOA I SUM OF NHBAU PRODUCTION PROBABILITIES*200000 SUMPNOT I SUM OF NHBTO PRODUCTION PROBABILITIES*200000 SUMXSHA I A DENOMINATOR OF HBSHDM MODEL, AO LEVEL A APPENDIX A -17 VARIABLE INDICES DESCRIPTION SUMXSHA (CONT.) *NOTE* DISTANCE CORRECTION NOT INCLUCED SUMXSHAX I DENOMINATOR OF HBSHD MODEL, AUTO, AO LEVEL 0 SUMXSHTX I DENOMINATOR OF HBSHD MODEL, TRANSIT, AO LEVEL 0 =0 IF TRANSIT SERVICE IS NOT AVAILABLE SUMXSRA I A DENOMINATOR OF HBSRDM MODEL, AO LEVEL A *NOTE* DISTANCE CORRECTIOIN NOT INCLUDED SWORKA I A NUMBER OF SECONCARY WORKERS IN AO LEVEL A (WORKERS) =SWORKER*($WHHA(A)-1)*WHHSA/T IF T > 0 =0 OTHERWISE WHERE T=((($WHHA(A)-1)*WHHSA) SUMMED OVER AO LEVELS) SWORKER I NUMBER OF SECONDARY WORKER RESIDENTS (WORKERS) TCOSTI I J PEAK TRANSIT TRAVEL COST/INCOME =(&TFARE+$OPADJ*$ACCRATE*&TACCT) /(100*INCOME(I)) IF INCOME(I) > 0, TTIMET < 255, =(5000+$OPADJ*$ACCRATE*&TACCT)/(100*INCOME(I)) IF INCOME(I) > 0, TTIMET GE 255 =0 OTHERWISE TODFOUT I HB O/D OUTBOUND SPLITTING FACTOR, TRANSIT T0TEMP I TOTAL EMPLOYMENT (EMPLOYEES) =&RETEMP+&SEREMP+&AGREMP+&TRDEMP+&MFGEMP +&OTHEMP TOTEMPD I TOTAL EMPLOYMENT DENSITY (EMPLOYEES/ACRE) =TOTEMP/ACRE IF ACRES > 0 =0 OTHERWISE TRDEMP I TRADE EMPLOYMENT (EMPLOYEES) =&TRDEMP TTIMET I J PEAK IN-VEHICLE TRANSIT TRAVEL TIME (MINUTES) =(<IVTT+C&ACCT) IF &TIVTT GT C, &TFARE < 5000 =255 OTHERWISE TWAIT1 I J PEAK TRANSIT WAIT TIME, SEGMENT 1 (MINUTES) =MIN(&TWAIT,4) TWAIT2 I J PEAK TRANSIT WAIT TIME, SEGMENT 2 (MINUTES) =&TWAIT-TWAIT1 TWALK I J PEAK TRANSIT WALK TIME (MINUTES) =&TWALK TXFER I J PEAK TRANSIT TRANSFER TIME (MINUTES) =&TXFER WAOI I G A WHHAO PERSON TRIP PRODUCTIONS, INCOME GROUP G, AO LEVEL A (TRIPS) WAOJ I J G A WHHAO PERSON TRIP ATTRACTIONS, INCOME GROUP G, AO LEVEL A (TRIPS) WADP I J G A WHHAO PROBABILITY, INCOME GROUP G, AO LEVEL A *NOTE* COMPUTED WITHOUT ADJUSTMENTS WADU I J G A WHHAO UTILITY, INCOME GROUP G, AO LEVEL A *NOTE* COMPUTED WITHOUT ADJUSTMENTS APPENDIX A-18 VARIABLE INDICES DESCRIPTION WADV I J G A WHHAO P/A PERSON TRIPS, INCOME GROUP G, AO LEVEL A (TRIPS) WHH I NUMBER OF WORKERS PER HOUSEHOLD (WORKERS/HOUSEHOLD) =WORKERS/HHS IF HHS > 0 =0 OTHERWISE WHHSA I A NUMBER OF HOUSEHOLDS WITH WORKERS, AO LEVEL (HOUSEHOLDS) WHHSG I G NUMBER OF HOUSEHOLDS WITH WORKERS, INCOME GROUP G (HOUSEHOLDS) *NOTE* EQUALS PRIMARY WORKER IN INCOME GROUP G WINCA I A MEAN INCOME OF WORKING HOUSEHOLDS IN AO LEVEL A (1965 $) WLSUM I J G WEIGHTED PHBWM LOG SUM, INCOME GROUP G =((WAOP*PHBWML), SUMMED OVER AO LEVEL) WORKERS I NUMBER OF WORKING RESIDENTS (WORKERS) =&WORKERS WTERMA I PEAK WALK TERMINAL TIME, ATTRACTION ZONE (MINUTES) =&WTERMA/100 WIERMP I PEAK WALK TERMINAL TIME, PRODUCTION ZONE (MINUTES) =&WTERMP/100
