I I I I 78-5866 KHISTY, Clement Jotindrakumar, 1928- AN EVALUATION OF ALTERNATIVE MANUAL LANDUSE FORECASTING METHODS USED IN TRANSPORTATION PLANNING. The Ohio State University, Ph.D., 1977 Engineering, civil

University Microfilms International,Ann Arbor, Michigan 48106 AN EVALUATION OF ALTERNATIVE MANUAL LANDUSE FORECASTING

METHODS USED IN TRANSPORTATION PLANNING

DISSERTATION

Presented In Partial Fulfillment of the Requirements for

the Degree Doctor of Philosophy in the Graduate

School of The Ohio State University

Hy

Clement Jotindrahumar Khisty, B.S., M.S., M.C.P.

» » * * »

The Ohio State University

1977

Reading Committee: Approved By

Dr. Zoltan A. Nemeth

Dr. Slobodan Mitric A. Nemeth, Adviser Dr. Burkhard von Rabenau Spartmenfc of Civil Engineering

Dr. Joseph Treiterer ACKNOWLEDGEMENTS

This dissertation has roots going back at least to 1970. Many people have contributed, in different ways and at different times, to its formulation. Among them are the staffs of the transportation and landuse planning agencies involved. The literature furnished by these agencies has been invaluable. The summaries of the landuse fore­ casting and distribution methods contained in this dissertation are excerpted from this literature. The author would like to express his appreciation to all these agencies.

Thanks are also extended to Dr. Michael Godfrey, Dr. Slobodan

Mitric, Dr. Zoltan Nemeth and Dr. Burkhard von Rabenau for their sin­ cere interest, encouragement and guidance throughout the course of this research.

My deepest debt is to my friend Beverly Pritchett. She not only typed the final document, but also took up the responsibility of editing and proofreading the entire dissertation. VITA

July k, 1928...... Born - Nagpur, Maharashtra, India

I9H8...... B.S., Civil Engineering, Nagpur University, Nagpur, India

19k8-1957 ...... Civil Engineer, Department of Public Works and Electricity, India

1958-1959 ...... West German Government Fellowship

1959-1969 ...... Officer in the Department of Technical Education, Ministry of Education, India

1970...... M.S., Civil Engineering, University of Cincinnati, Cincinnati, Ohio

1970-1973 ...... Transportation Engineer, Ohio-Kentucky- Indiana Regional Planning Authority, Cincinnati, Ohio

1973...... M.C.P., University of Cincinnati, Cin­ cinnati, Ohio

1973-1971* ...... Senior Transportation Engineer, Mid-Ohio Regional Planning Commission, Columbus, Ohio

197*1—1975 ...... Teaching Associate, Department of Civil Engineering, The Ohio State University, Columbus, Ohio

1975- ...... Principal Planner, Toledo Metropolitan Area Council of Governments, Toledo, Ohio

PUBLICATIONS Some Selected Publications are:

"Transportation Pooling, 11 Toledo Metropolitan Area Council of Govern­ ments, Toledo, Ohio, February, 1976.

iii "Transit Development in Central Ohio,” Traffic Engineering, Vol. U4, No. 1 0 , pp. 16-19, July, 197^. "Transportation Planning Process: Operations Plan," OKI Regional Planning Authority, Cincinnati, Ohio, April, 1972.

"Structural Steel Tables," Asia Publishing House, London, U.K., 1972.

FIELDS OF STUDY

Major Field: Civil Engineering

Studies in-Transportation Engineering. Professors Michael B. Godfrey, Slobodan Mitric and Zoltan Nemeth

Studies in Urban Economics. Professor Burkhard von Rabenau

Studies in City and Regional Planning. Professors William Sims and Daniel Czamanski

Studies in Transportation Geography. Professor Howard Gauthier

Studies in Urban Sociology. Professor Kent Schwirian TABLE OF CONTENTS

Page ACKNOWLEDGEMENTS ...... ii

VTTA ...... iii

LIST OF TABLES...... vii LIST OF FIGURES ...... ix

INTRODUCTION ...... 1

Chapter

I. THE GENERAL PROBLEM...... 2

The Planning Process...... 2 The Problem...... 3 Purpose and Scope of Research ...... 7 Research Approach ...... 7 Organization of This Research ...... 9 N o t e s ...... 10 II. THE NATURE OF PLANNING MODELS...... 11

Background ...... 11 Urban Spatial M o d e l s ...... 11 Classification of M o d e l s ...... 13 Characteristics of the Landuse Forecast and Landuse P l a n ...... 17 Classification of Landuse Models ...... 18 Critique and Conclusions...... 20 N o t e s ...... 23 III. SURVEY OF LANDUSE FORECASTING METHODS...... 25

Introduction ...... 25 General Review ...... 25 Outlines of Ten Landuse Models ...... 28 Structural Linkages of the M o d e l s ...... Ul N o t e s ...... U7

v IV. A TEST OF TWO RESIDENTIAL LANDUSE MO D E L S ...... 1*9

Introduction ...... 1*9 Methodological Problems...... 50 Data Sources and their U s e ...... 51 Model Performance and the U - T e s t ...... 51 The Density-Saturation Gradient Method (DSGM) . . . 53 The Toledo M ethod...... 65 Conclusions ..... TO N o t e s ...... 7U V. RECOMMENDATIONS AND GUIDELINES...... 75

O v e r v i e w ...... 75 General Guidelines Pertaining to Landuse Planning...... 76 Specific Guidelines ...... 8l Conclusions...... 86 N o t e s ...... 88

APPENDIX A. SUMMARIES OF LANDUSE FORECASTINGMETHODS ...... 89

Akron, O h i o ...... 89 Austin, T e x a s ...... 99 Canton, O h i o ...... 105 Chicago, Illinois ...... 110 Columbus, O h i o ...... 122 Kansas City, Missouri...... 132 Oklahoma City, Oklahoma...... ll*3 Pittsburgh, Pennsylvania...... 150 Rockford, Illinois ...... 159 Toledo, O h i o ...... 162 N o t e s ...... 175

B. DATA AND CALCULATIONS FOR THE D S G M ...... 176

C. DATA AND CALCULATIONS FOR THE TOLEDO METHOD .... 185

BIBLIOGRAPHY...... 187

vi LIST OF TABLES

Page 1. Landuse Studies Investigated ...... 26 2. Residential Development Factors ...... 6l 3. U-Test at Ring and Census Tract L e v e l ...... 6k U. Capacity Landuse Projections by Planning Areas in City of T o l e d o ...... &7 5. Development Cycle: Estimated Years Required to Achieve Given State of Development ...... 6? 6. Forecast Growth of Dwelling Units in Planning Areas. . 68 7. Forecast and Actual Population for 1 9 7 k ...... 69 8. U-Test at Planning Area L e v e l ...... 71 9. U-Test at Census Tract Level ...... 72 10. 1970 Travis County, Texas, Landuse With Projections to 1980 and 1990 (in acres)...... 102 11. Analysis of Existing Landuse Patterns Using Density- Saturation Gradient Method ...... 117 12. Allocating District Population to Zones Using Density-Saturation Gradient Method ...... 120 13. Employment by Industrial Category, Kansas City Metropolitan Region, 1970-1980-2000 ...... 138 lU. 1995 Projected Population by City, Association of Central Oklahoma Governments...... lU6 15. Percent of New Growth for 1970-1995 According to Allocation Model by Sectors, Rockford, Illinois . l6l 16. Inventory of Landuse Data, Toledo, O h i o ...... 166 17. Estimated Years Required to Achieve Given State of Development, Toledo, Ohio ...... 170 18. Distribution Worksheet, Toledo, Ohio ...... 172 19. Percentage Distribution of Remaining Growth, Toledo, O h i o ...... 173 20. Distribution Worksheet of Residential Development, Toledo, O h i o ...... 17U 21. Groupings of Census Tracts Arranged by Distance from High Value Comer (DSGM)...... 176 22. 1965 Population, Density and Holding Capacity by Census Tracts (DSGM) ...... 177 23. 1965 Residential Density by Distance Bands ...... 180 2k. 1965 Population as a Percentage of Maximum Holding Capacity by Distance B a n d s ...... l8l

vii 25. 1971* Forecast of Population “by Distance Band and Census T r a c t ...... 182 26. 197^ Actual Population by Distance Bands and Census Tracts ...... 183 27. Distribution of Population to Census Tracts (DSGM)...... 18H 2 8. Planning Area to Census Tract (i9 6 0) Equi'va­ lency Table (Toledo M e t h o d ) ...... 185 29. Distribution of 197^ Forecast Population to Census Tracts, Toledo Method ...... 186

viii LIST OF FIGURES

Page 1. The Transportation Planning Process ...... U 2. The Landuse/Transportation C y c l e ...... 12 3. Chadwick's Model Categorization ...... 13 U. Echenique's Model Classification ...... lU 5. Predictive and Planning M o d e l s ...... 16 6. Structural Linkage, Akron, O h i o ...... h2 7. Structural Linkage, Austin, Texas ...... k2 8. Structural Linkage, Canton, O h i o ...... k3 9. Structural Linkage, Chicago, Illinois ...... U3 10. Structural Linkage, Columbus, O h i o ...... UU 11. Structural Linkage, Kansas City, Missouri ...... 12. Structural Linkage, Oklahoma City, Oklahoma ...... U5 13. Structural Linkage, Pittsburgh, Pennsylvania .... U5 lU. Structural Linkage, Rockford, Illinois ...... U6 15. Structural Linkage, Toledo, Ohio ...... kS 16. Population Density by Distance Bands (1965) ...... 5^ 17. Percent Population Saturation by Distance Bands .... 57 18. Percent Population.Saturation Sketch: 1950-60-65 . . 59 19. General Flow Chart of Allocation Process, Akron, O h i o ...... 90 20. Population, Housing and Auto Distribution, Akron, O h i o ...... 92 21. Distribution of Residential Activity to Traffic Zones, Akron, Ohio...... 93 22. Landuse Plan Development Process, Austin, Texas . . . 99 23. General Flow Chart of Population and Employment Distribution, Canton, O h i o ...... 106 2k. Population Distribution, Canton, Ohio ...... 108 25. CATS Landuse Estimating Procedure, Chicago, Illinois...... Ill 26. General City Structure Showing Rings and Sectors (CATS M e t h o d ) ...... 115 27. Zone Structure of a Traffic Analysis District .... 115 28. The Forecast of District Population Using Density-Saturation Gradient Method ...... 119 29. General Flow Chart of Landuse Distribution Procedure, Kansas City, Missouri ...... 133 30. General Flow Chart of Landuse Distribution Procedure, Oklahoma City, O k l a h o m a ...... lUU

ix 31. The Cooperative Planning Process, Pittsburgh, Pennsylvania ...... 151 32. Southwestern Pennsylvania Allocation Model, Pittsburgh, Pennsylvania ...... 153 33. Equation Development Methodology, Pittsburgh, Penn­ sylvania ...... 155 3U. Employment Allocation Model, Pittsburgh, Penn­ sylvania ...... 156 35. Housing Unit Allocation Process, Pittsburgh, Pennsylvania...... 157 36. Income and Population Allocation Process, Pitts­ burgh, Pennsylvania ...... 158 37. General Flow Chart of Landuse Planning Process, Toledo, Ohio ...... 163

x INTRODUCTION

The spatial organization of urban areas lies close to the heart

of the planning process. Landuse models have been used by planners to

predict with some degree of precision the spatial, organization of

population and economic activity in the region.

The subject of this research is the examination of landuse

models or simply landuse forecasting and distribution methods. Although landuse models are used extensively in comprehensive plan­

ning, this research deals primarily with those models which are of

concern in urban transportation planning. The main thrust of this

investigation is on the evaluation of traditional manual landuse forecasting methods.

1 CHAPTER I

THE GENERAL PROBLEM

The Planning Process

Urban spatial patterns are the outcome of a process which allo­ cates activities to sites. These patterns can he observed and con­ sidered at the level of neighborhoods, metropolitan areas, regions, and the nation. They are thoroughly interrelated with patterns, systems, and modes of transportation.

Conventional wisdom has it that there is a two-way relationship between transportation and landuse. In the short term, patterns of landuse and economic activity largely determine travel demand. In the longer term, the accessibility potential offered by transportation facilities affects changes in landuse patterns.

But transportation facilities outlast the short term. This led the Bureau of Public Roads (now the Federal Highway Administration) to stimulate long-range and comprehensive planning. The passage of the Federal-Aid Highway Act of 19 ^ first provided regular federal-aid highway funds for use in urban areas, and the Federal Highway Adminis­ tration has actively promoted urban transportation planning since that time. The Federal-Aid Highway Act of 1962 required that programs for federal-aid highway projects approved after July 1, 1965, in urban areas of more than 50,000 population, must be based on a continuous, 3

comprehensive transportation planning process carried on cooperatively by states and the local communities.-^- The comprehensive character of

the planning process requires that economic, population and landuse

elements be included. Figure 1 illustrates the various elements which

comprise the four technical phases of the urban transportation planning

process.2 In the transportation systems analysis phase, many alter­

natives of both landuse plans and transportation systems can be evaluated. The objective of the transportation planning process is

to provide the information necessary for making decisions on when and

where improvements should be made in the transportation system, thus

promoting travel and land development patterns that are in keeping with community goals and objectives.3

The Problem The current conventional procedure of predicting the quantity

of travel on transportation networks is to model travel behavior as a

series of sequential, independent choices of trip generation, trip

distribution, modal split, and traffic assignment. Landuse forecasting

precedes travel forecasting as a separate step. For each travel choice,

the existing pattern of usage in the region is related to a small set of independent variables. The trend is then assumed to hold in

the future.**

In the United States and in some western societies, land is

allocated among alternative uses mainly in private markets, with more

or less public regulation.5 in such societies, urban areas result Analysis of Existing Systems - -A Analysis- - - — — Forecast ^ 1--- Conditions Inventories CNMCACTI TY Y IT IV T C A ECONOMIC CNMCACTI TY Y IT IV T C A ECONOMIC CNMCACTI TY Y IT IV T C A ECONOMIC N POPULATION AND P UAI N PULATIO PO D N A AND PO PULATIO N N PULATIO PO AND TECHNIQUES PROJECTION PROJECTION FU TU R E E R TU FU P I IES LIC PO D N A THE TRANSPORTATION TRANSPORTATION THE - PROCESS PLANNING COM M UNITY UNITY M COM COALS COALS FORECASTING TECHNIQUES ANDUSE S U D N LA T R IP GENERATION IP R T FIGURE l' FIGURE FUTURE FUTURE FEEDBACK MPL ATON TIO TA N E M LE P IM ( GENERATION IP R T CHARACTERISTICS \ Hi Tr it) s n ra -T y a w h ig (H P SRIUTON N TIO IBU ISTR D IP R T C A LIB R A T IO N O F F O N IO T A R LIB A C TffAVFT ACCURACY CHECKS CHECKS ACCURACY DEL O M P S E L B A T IP R T AND AND EWOKAJ T NT EN STM ADJU ORK NETW IN IT IA L ASSIGNMENT ASSIGNMENT L IA IT IN EWORK NETW UT E R TU FU P IT B I N TIO IBU DISTR IP R T D N A YTM NALYSIS A AN SYSTEMS TRANSPORTATION TRANSPORTATION EO ENDED M RECOM Hl t Tr lt) a n ra -T y a ltw la (H SIN ENT ASSIGNM SYSTEM • FU TURE TURE • FU RNP RTATION TRANSPO Hih yTransit) ay-T ighw (H NETW ORK AND AND ORK OF NETW N TIO LEC SE LI ES IE IT IL C A F ZONES mainly from locational decisions by large numbers of private developers and buyers, each attempting to further his own interests.

The transportation planning process requires forecasts of various socio-economic and landuse data as input for travel forecasts. This process uses socio-economic data that are ultimately dependent on landuse because landuse patterns reflect the distribution of population, income levels, employment, and other socio-economic variables. Land­ use models^ are concerned with providing small-area (usually zonal) forecasts of population, employment and other socio-economic variables, in a suitable form for input to a trip generation analysis.

The general landuse model consists of taking areawide forecasts of the several socioeconomic variables as control totals and using some procedure to allocate them to analysis zones. Land usage is then usually determined by activity land consumption rates. Generally, the activities are located in the analysis zones based on the zonal characteristics such as vacant available developable land, zoning, availability of public utilities, accessibility, etc. The allocation procedures used range from traditional manual techniques to sophisti­ cated urban development models, all falling under the rubric of landuse models.

The traditional or manual allocation technique generally consists of gathering and analyzing data and then locating (allocating) activi­ ties based on acceptable planning standards and professional judgment.

A step further is to use various mathematical formulations such as regression analysis to quantify existing relationships and then use these mathematical formulations along with planning standards and professional judgment. This "begins to make the allocation more

quantitative and requires more explicit assumptions concerning various

factors which influence growth. On the other end, urban development

models are even more analytic and include explicit relationships and

theories of urban growth processes.^ Large metropolitan study areas have in recent years made use of these sophisticated computer-oriented

landuse models. Partly" because of the over-optimistic outlook of

their creators, and partly due to the unrealistic expectations of

their potential users, many of these efforts have been partial fail- Q ures. Lack of expertise, time and money to run these large-scale models have contributed to the decision made by many metropolitan

studies to rely on simple, less expensive, and far more transparent models. These simple allocation models are the usual traditional manual techniques, and it is almost certain that many metropolitan

study areas will continue to depend on these techniques for some time

to come.

Unfortunately, these traditional manual landuse models have not been well documented, nor has much effort been expended in comparing

and evaluating them, despite the fact that a large number of trans­ portation studies in this country have made use of them. Thus, there

is an urgent need to examine, compare and evaluate traditional manual

landuse models so that promising methods can be further examined and

recommended. Suitable guidelines can then be provided to transportation planning agencies regarding their use. 7

Purpose and Scope of Research

The initial goal of this research is to examine the landuse planning process insofar as it relates to transportation planning. The subject of this investigation is the examination of landuse forecasting methods. The scope of this research has, however, been confined to meet limited objectives. While these objectives focus upon the landuse model as an operational tool in the overall transportation planning process, the basic purpose of this investigation is to explore, re­ view and evaluate alternative landuse models, oriented to basically traditional manual procedures. Since well-established procedures^ for forecasting regional area-wide totals of population, employment and economic activity have already been researched, documented, and widely used, this research deals with these issues only indirectly. The main thrust of this investigation is on methods of allocating these regional totals to small areas, such as planning areas, districts, census tracts, and traffic analysis zones. Most important, the research aims at providing guidelines to transportation planning agencies in the use of selected methods of landuse forecasting.

Research Approach

A holistic approach was taken to investigate the role of the landuse model in the overall planning framework. This investigation led to the examination of the interrelationships between transportation and landuse and the levels of sophistication of landuse models used in the planning process. 8

The nature of the research required an extensive literature review of landuse models used by planning agencies, discussions with engineers and planners, and a critical examination of the work pro­ duced by about thirty planning agencies. Somewhat narrow limits had to be defined for the range of methods of landuse forecasting and distribution for evaluation. First, only those methods are documented which fall into the basic traditional class, or at most are limited to small-scale modeling. Secondly, methods used by cities of varying populations (from 300,000 to 7,000,000) are included. And thirdly, only those methods are included which are basically non-site-specific.

Initially, about sixty transportation planning agencies were contacted with a request to provide reports, 'documents and other pertinent literature regarding their landuse models. Of these sixty agencies, a little more than half responded. Based on the material received, along with brief discussions and further correspondence with the staff of these agencies, the methods of ten study areas were selected for inclusion here. A heuristic approach was developed which provides the opportunity of learning from the method offered by each study area.

An important conclusion that surfaced from this literature review was that two of the methods used in the past appeared promising and worthy of further investigation. These two methods were then applied in one study area to ascertain what implications were involved in working with these models and also to evaluate their performance and accuracy. Based on this and other experience, a set of recommendations and guidelines has been developed for use by transportation agencies. Organization of this Research

This chapter provides a "background to the urban transportation planning process focusing on the identification of the problem of landuse forecasting. Chapter Two examines the nature of landuse models and presents a classification of such models.

A survey of traditional manual landuse forecasting methods used by transportation planning agencies is presented in Chapter Three.

First, a general review is provided, followed by comments on ten land­ use models used currently or in the past. Summaries of these ten models are placed in the Appendix for those inclined toward a more in-depth study. Two of the ten landuse models reviewed are further examined and tested in-a common setting. The procedure and results are presented in Chapter Four, while Chapter Five sets out general and specific guidelines for use by transportation planning agencies. 10

NOTES

■^U.S. Department of Transportation, Federal Highway Administra­ tion, Urban Transportation Planning, Policy & Procedure'Memorandum 50-9, Transmittal 162, Nov.-, 1962. 2U.S. Department of Transportation, Federal Highway Administra­ tion, Urban Transportation Planning, General Information and Intro­ duction to System 360, p. 1-5, March, 1965*

^Ibid., p. 1-6. I4. Daniel Brand, Theory and Method in Landuse and Travel Fore­ casting. Landuse'and Transportation Planning, Research Record No. U22 (Washington, D.C.: Highway Research Board, 1973) p. 10.

'’Edwin S. Mills, Studies in the Structure of the Urban Economy. (Baltimore: Johns Hopkins Press) 1972.

6Benjamin Reif in Models in Urban and Resional Planning (New York: Intertext Educational Publishers, 1973) points out the confusion which has arisen from names that models of spatial structure have received from different sources. Some are: Urban Spatial Structure Model; Urban Development Model; Landuse Model; Activity Model; Loca­ tion Model; Allocation Model; Model of the Ecosystem; Landuse Alloca­ tion Model. All, of course, attempt to simulate the structure of landuses and activities in cities and regions. See particularly pp. 27-38. 7 U.S. Department of Transportation, Federal Highway Administra­ tion. Computer Programs for Urban Transportation Planning: General Information, pp. 58-59. g Stephan H. Putman, Urban Landuse and Transportation Models: A State-of-the-Art Summary. Transportation Research. Vol. 9, 1975, pp. 187-202. 9 See, for instance, Robert C. Atchley, Population Pro,1 ections and Estimates for Local Areas (Oxford, Ohio: Miami University, 1970); Donald A. Krueckeberg and Arthur L. Silvers, Urban Planning Analysis, Methods and Models (New York: John Wiley and Sons, Inc. 1974). CHAPTER II THE NATURE OP PLANNING MODELS

Background

Cities, counties, townships, and other units of local government routinely prepare plans for the future development and redevelopment of their jurisdictions. Such plans, typically called comprehensive or general plans, are prepared under the auspices of local or metro­ politan planning organizations. Landuse has heen, and is, a central element in such plans. Landuse plans predetermine many of the factors that ultimately affect the content of transportation plans. The mutual inter-dependence of landuse and transportation planning activities is illustrated in figure 2.^ It may he noted, however, that this cycli­ cal process does not reflect the external factors that are brought to hear upon the system. For example, land values respond to establish­ ment of new firms based on locational decisions from outside the cycle.

Urban Spatial Models

Researchers have been making attempts to simulate the structure of landuses in cities and regions for the past thirty years. They soon recognized the interrelationships among different elements of the landuse transportation cycle. Two of these early pioneers were

Mitchell and Rapkin . In their view, trips were perceived as con­ nectors between activities through the flow of goods and people.

Transportation needs would then be based on estimates of future flows 11 12

of goods and people. Although, some techniques to deal with transpor­ tation problems were comparatively well developed, landuse forecasting,

a prerequisite for transportation planning, was, at that time, little understood. Indeed, all that was commonly done was to extrapolate

existing landuse trends. It was soon realized that it was ridiculous to base sophisticated transportation models on inputs of landuse data which were produced in a very rudimentary way.^ Systematic research was then undertaken to fulfill this need.

Trips

Landuse Transport Needs

Land Values Transport Facilities

Accessibility

FIGURE 2

THE LANDUSE/TRANSPORTATION CYCLE 13

In recent years, researchers and planners of urban systems have come to the conclusion that through model representation it is possible and advantageous to describe the process which determines the use. of urban land. In evolving an urban development model or a

forecasting technique the planner views the city as a system, that is,

the city is an organized entity displaying certain systematic patterns.

The underpinnings of a model are the theories of citydevelopments which the planner assumes to exist. Obviously, a model or technique is no better than the theory on which it is based.

Classification of Models li Chadwick has provided a general categorization of model types and their relationships and combinations (see figure 3).

Micro Macro

Static Dynamic

Descriptive Behavioral

Prescriptive Predictive

Deterministic Stochastic

(after Lowenstein, 1966) Simulation FIGURE 3

CHADWICK’S MODEL CATEGORIZATION Ik

Echenique^has also provided a model classification of particular interest to landuse planners. It classifies models according to the intentions of the planner and the purposes for which the model is designed. Four main types of models can he distinguished:

Descriptive models Predictive models Explorative models Planning or decision models

The theoretical relationships between them and reality can be observed in figure

Explorative

Descriptive Planning

A\ Predictive

Models

Reality

Past Present Future

FIGURE k

ECHENIQUE'S MODEL CLASSIFICATION 15 Descriptive models replicate relevant features of an existing urban environment and are thus useful in understanding the behavior of cities and regions. Indeed, it is not possible to predict, explore or plan without a previous description of the reality under study.

Usually a descriptive model is the first step in the model-building process, thus establishing the nature of the relationships between variables. On the other hand, predictive models forecast the future states of urban systems and are thus time-related. Explorative models are basically the same as descriptive models, with the difference that they are speculative by nature. In the planning or prescriptive model, a measure of performance is introduced with the objective of deter­ mining the optimum or near-optimum solution under certain conditions.^ Richardson^ has provided a good comparison of predictive and planning models. He says,

Although planners sometimes use predictive models aimed at forecasting the most likely course of events in the absence of intervention, they will normally wish to influence the future. In order to do this they will need to make use of a planning model. Such a model allows for specific goals and policy objectives (determined in part by the political pro­ cess), and the planners, via a manipulation of the instrumen­ tal variables under their control, try to achieve these goals.

The basic differences are shown in figure 5.

Thus planning models incorporate the objectives formulated through the political process to derive a policy. Once ratified, this policy is translated into a plan. Such plans can change events in specified directions. Predictive Model Planning Model

Facts Theories Facts Theories

Model Model

Objectives

Prediction Events Policy

Controls

Plan

Events

Inconsistent Consistent Experience

(after Richardson, 1969)

FIGURE 5

PREDICTIVE AMD PLANNING MODELS®

H ON IT

Characteristics of the Landuse Forecast and Landuse Plan

Landuse and transportation alternatives are usually prepared because it is believed that examination of different options results in the selection and adoption of a realistic plan. Certain combina­ tions of landuse patterns and transportation systems have significant advantages, in terms of efficiency. The public and their representa­ tives are thus able to look at alternatives and identify advantages. Alternatives are based on plan-form concepts such as the radial, lin­ ear, or nucleated city. Implied in these plan-forms are the policies concerning the location and density of landuse. These conceptual plans are further developed and refined using conventional planning methods, till they reach the stage where they can be used to distribute regional totals by manual or computer procedures.^

It is essential at this stage to distinguish between a landuse forecast and a landuse plan. In the past it was the common objective of transportation planners simply to develop a quantitative measure for landuse aggregated to the level of a traffic analysis zone for a future year, for feeding into trip generation equations. The arrange­ ment, design, and interactions of landuses within or between analysis zones did not and still do not directly concern the transportation planner.^ The methods used in forecasting landuse rely on regional totals. The plan alternatives usually emphasize internal consistency regarding regional totals. Today, there is a much greater demand for integrating the landuse forecasting process with the production of a landuse plan. This integration is partly due to the fact that 18 metropolitan planning agencies are now engaged in other areas of planning "besides transportation, which demand landuse plans.

Classification of Landuse Models

Landuse models have been classified in numerous ways. 11 Boyce12 discusses these models in terms of their sophistication. Level 1, the least sophisticated, involves the setting up of alternative physical patterns of end-state development similar to refined concept plans.

Such models do not get involved with the staging of facilities.

Generally, conventional manual methods are used to distribute growth.

In Level 2 models the simple concept of the urban development process for spatially allocating households and employment is introduced along with the staging of projects. However, this allocation is generally not performed by computer methods. Traditional manual methods are used. These traditional methods derive from a pre-computer technology. The most complete and widely used formulation of this approach is that of Chapin. 13 The main analytic components of this procedure are, for each landuse category, a set of location require­ ments and a set of space requirements. Specific rules for the resolu­ tion of conflicts among these competing landuses for a site are not defined in this approach. These Judgments must be made by the analyst, based on the given principles and standards, the special knowledge of local conditions, and what is considered to be in the best interest of the public. A further step is to introduce the use of some mathematical formulation such as regression equations. Level 3 programs make more sophisticated use of concepts of the development process, including a wider range of policy specifications.

They usually use computer methods. Models at this level are generally referred to as the ’’Market Simulation Approach.” The archetype of this system of models was developed in the early i9 6 01s by Ira Lowry Ik of the Rand Corporation. The general structure of the Lowry model, which has been imitated, altered, and expanded, has been used in large metropolitan studies in recent years.^ Recent improvements include more complex statistical estimating procedures, consideration of more than residential activity, stratification of residential locations into several distinct groups, and incorporation of behavioral rela­ tionships into the model formulation. Models which can be classified in this group include the EMPIRIC Activity Allocation Model and the , . l6 Projective Land Use Model (PLUM).

PLUM, for example, is designed to distribute future population, employment and landuse to small areas based on distribution data of these characteristics in some base year, coupled with a series of allocating algorithms. These allocation algorithms are based on two fundamental, concepts: first, on a distinction between "basic” and

"local-serving" employment; and second, on the notion of an- "allocation function." Both concepts are derived directly from the Lowry model.

The entire analytical process is handled in mathematical form on electronic computers.

Both the traditional "planned requirements approach" and the sophisticated computer-oriented "market simulation approach" make 20 assumptions about the role of influence of the transportation system.

However, the important difference between them lies in the fact that the traditional planned requirements approach produces a prescriptive plan. The computer-oriented "market simulation approach" disclaims any normative implications, and claims to be forecasting an economic growth process, extrapolating patterns of economic market behaviors. 17

Critique and Conclusions

The experience of planning agencies demonstrates that much difficulty and delay were experienced in operationalizing computer- oriented models. The data management problems were particularly severe. Due to these and other problems, work schedules for accom­ plishing crucial tasks had to be drastically revised, which meant that some studies lost their credibility. Alternately, agencies resorted to crude short-cuts, interim plans, or dropping the use of their computer-oriented models altogether to stay somewhat close to schedule.

While the use of computer methods of distribution was demanding of time and staff resources, many agency officials felt that the advances in metropolitan planning required the use of sophisticated computer techniques. 19 This situation was perplexing.

As to the question of data, there was no doubt that computer- oriented models required considerably more data for one or more points in time. Computer models generally demand that certain policy statements be made more explicity. As a further complication, the 21 physical design, of each alternative, most often a graphic or plan-form concept, has to he translated into quantitative terms for small areas for input to the models. Subsequently, the output of the models has in turn to he interpreted and expressed graphically for ease of communication to the public. These translations are often subjective, or at the very least they obscure the precise meaning of the fore- cast. 20 In some programs using computer models, the difficulties of making the models operational meant that no useful output was obtained until the end of a single prolonged phase of elaboration. 21

This discouraging situation has changed somewhat for the better, with the result that now there are different schools of thought regarding the proper direction for future landuse modeling. One suggests the creation of bigger, better, more integrated and complex models, while the other recommends a more modest approach. This may appear to be regressing, but it is a somewhat expected reaction when one observes that all too often in the past the over-ambitious quests for the com­ plex, comprehensive model have resulted in little more than expended 22 resources and an untested theory.

Much criticism is contained in the papers and discussions delivered at the Highway Research Board Conference, Dartmouth,2^ and by other researchers. 2b Lee 25^ criticizes these large-scale models in regard to their "Hypercomprehensiveness, Grossness, Hungriness,

Wrongheadedness, Complicatedness, Mechanicalness, and Expensiveness."

Most important, such models make heavy demands on the expertise and time of a metropolitan area study staff, as well as on a study's budget. 22

Furthermore, an important point of distinction "between the traditional manua/l versus the sophisticated computer model needs to "be noted. The former method is often dependent on intuition, reason,

and subjective judgment for allocating overall metropolitan control totals of population and activities to small areas. However, the method is thought to be capable of dealing with somewhat more subtle

questions and issues that are presently difficult to express in terms meaningful to a computer-oriented model. Most metropolitan area

studies have, for these reasons, relied on simple, less expensive, and

far more transparent urban models, and it is almost certain that they will continue to do so for some time to come. At the present time there are indications that movement is being made in both directions: towards the smaller and more simple, and towards the more integrated and complex. 26 23

NOTES

•'•Harold Marks, Protection of Highway Utility. ' NCHRP 'Report Number 121, Washington, D. C. Highway Research Board, 1971. 2 Robert B. Mitchell and Chester Rapkin, Urban Traffic: A Function O f 'Landuse. (New York: Columbia University Press, 195^). ^Benjamin Reif, Models i n 'Urban'and'Regional Planning. (New York: Intertext Educational Publishers, 1973), p. 35^ Hereafter referred to as Models in Urban and Regional Planning.

^George Chadwick, A Systems View of Planning. (Oxford, U.K., Pergamon Press, 1971), p. 1 9 6.

^M. Echenique, Models: A Discussion, Landuse and Built'Form Studies. Working Paper 6 (University of Cambridge, March, 1968).

^Reif, Models in Urban and Regional Planning, p. 55-

^Harry W. Richardson, Elements of Regional Economics. (Balti­ more: Penguin Books, 1969), pp. 111-113. 8Ibid., p. 112. g ^David E. Boyce, Norman D. Day and Chris McDonald, Metropolitan Plan Making, (Philadelphia, Pa., Regional Science Research Institute, 1970), pp. 2-U. Hereafter referred to as Boyce, Metropolitan Plan Making.

■'■^Will Terry Moore, An Introduction to Urban Development Models. (Washington, D.C., U.S. Dept, of Transportation, Oct., 1975), p. 27. Hereafter referred to as Moore, Urban Development Models.

^"See, for example, Ira S. Lowry, Seven Models of Urban Develop­ ment: A Structural Comparison, Urban Development Models. Special Report 97 (Washington, D.C.: Highway Research Board, 196 8), pp. 121- 63; Leslie J. King, Models of Urban Landuse Development, Models of Urban Structure. (Lexington, Mass., Lexington Books, 1972), pp. 3-26. •^Boyce, Metropolitan Plan Making, p. 1*3. ^•3p. Stuart Chapin, Urban Landuse Planning. (Urbana, 111.: University of Illinois Press, 1 9 6 5).

■^Ira S. Lowry, A.Model of Metropolis. (Santa Monica, Cal.: The Rand Corporation, l§6l*). 2k

-*-5see, for example, William Goldner, The Lowry Model Heritage. Journal o f 'the American Institute ~of Planners. March, 1971, pp. 100- 110; Stephan H. Putman, Urban Landuse and Transportation Models: A State-of-the-Art Summary. •Transportation Research, Vol. 9, 1975, pp. 187-202.

l^Moore, Urban Development Models, p. 32. ^Donald A. Krueckeberg and Arthur Silvers, Urban'Planning Analysis. (New York: John Wiley and Sons, Inc., 197^-), pp. 328-9.

■^Boyce, Metropolitan Plan Making, p. 25.

^Ibid., p. 25. 2^Ibid., p. UU.

21Ibid., p. k9.

22Moore, Urban Development Models, p. 22.

2%ighway Research Board, Urban Development Models. Proceedings of a Conference, June 26-30, 1967, Dartmouth College, Hanover, N.H. Special Report No. 97 (Washington, D.C.: Highway Research Board, 1968).

^HSee,PiI for example, Douglas B. Lee, Models and Techniques for Urban Planning. Report No. VY-2^7^-G-l (Buffalo: Cornell Aeronautics Lab, Inc., 1 9 6 8); and Paul 0. Roberts, Model Systems for Urban Trans­ portation planning: Where Do We Go From Here? Transportation Analy­ sis: Past and Present. Research Record No. 309 (Washington, D. C.: Highway Research Board, 1970), pp. 3^-UU.

25■'Douglas B. Lee, Requiem for Large-Scale Models. Journal of the American Institute of Planners. May, 1975, pp. 163-178. ~*~"

2^Moore, Urban Development Models, p. 22. CHAPTER III

SURVEY OF LANDUSE FORECASTING METHODS

Introduction

This chapter reviews ten landuse forecasting methods used by planning agencies across the United States. These study areas, desig­ nated "by their central cities, are listed in Table 1.

This survey is organized in three parts. First, a general critique is provided of current landuse planning practice. Second, each study is briefly outlined with comments on its strengths and weaknesses. Third, a summary of each study has been prepared and placed for reference in the Appendix.

General Review

Although the focus of this research is on procedures of dis­ tributing regional totals of population and economic activity to small areas, some observations connected with the overall landuse planning process having a direct bearing on this research are summarized in the following paragraphs.

In a large majority of the studies reviewed, it was found that goal formulation generally reflected little citizen input other than through the traditional committees. The planning documents contained goal statements which often appeared very general, perhaps even 25 26

TABLE 1

LANDUSE STUDIES INVESTIGATED

1970 Principal City Study Area 'SMSA Population

1. Akron, Ohio Akron Metropolitan Area Trans­ 679,000 portation Study (AMATS)

2. Austin, Texas City of Austin 323,000

3. Canton, Ohio Stark County Area Trans­ 39^,000 portation Study (SCATS)

Chicago, Chicago Area Transportation 6,979,000 Illinois Study (CATS)

Columbus, Ohio Mid-Ohio Regional Planning 1 ,018,000 Commission (MORPC)

6. Kansas City, Mid-America Regional Council 1 ,272,000 Missouri (MARC)

7. Oklahoma City, Association of Central Okla­ 698,000 Oklahoma homa Governments (ACOG)1

8. Pittsburgh, Southwestern Pennsylvania 2 ,1+01,000 Pennsylvania Regional Planning Com­ mission (SPRPC)

9. Rockford, Rockford Area Transportation 272,000 Illinois Study (RATS)

10. Toledo, Ohio Toledo Metropolitan Area Coun­ 763,000 cil of Governments (TMACOG) 27 superficial. Since almost all the landuse plans examined were pre­

scriptive by nature, one would have expected that goal statements would reflect to some extent the views of the community in directing

growth. This was generally observed to be the exception rather than

the rule. The temporal limits of planning generally ranged between five

and thirty years. In most planning contexts the short-range plans

are supposed to mesh into the long-range plans through time-staging

of projects. However, few of the plans examined had the time-staging

element built in with long-range planning.

The development and presentation of alternative plans for

consideration by decision makers is one indicator of the effort that

is being expended in unraveling the complexities of the plan. About

half the studies examined developed alternative plans. Some agencies

started with alternative plans but did not pursue them beyond a super­

ficial examination. There were a few instances where two sets of

conceptual plans were prepared, one predictive and the other pre­

scriptive. These sets of plans provided an excellent means of com­

parison for decision-makers and the public. One of the persistent deficiencies encountered in the planning

reports was the inadequate documentation and recording procedures of

agencies responsible for technical planning efforts. The documents were particularly deficient in describing the assumptions in making

forecasts, developing plans, and making recommendations. Also,

the documents tended to avoid descriptions of local situations 28 which influenced the development of plans.

It was observed that planning efforts often started from

’’scratch," instead of being updated from existing plans and available data. This not only indicated an unnecessary expenditure of funds, but also obliterated in many cases the rationale which was originally used in developing the plan.

According to the federal regulations1 , every planning agency is expected to review its plans in order to keep them current. It was surprising to find that very few agencies had evaluated the accuracy of their forecasting and distribution procedures. Even those that did would not readily disclose the results. Hence, no direct performance measure could be applied to check the accuracy of the methodologies reviewed. Several researchers have expressed similar disappointments.^

Outlines of Ten Landuse Models

This section briefly outlines each of the ten landuse models reviewed and provides individual comments. It also examines the structural linkages of the models. Further details of each study are to be found in the Appendix.

On the surface, some of the models show remarkable similarities.

But deeper scrutiny reveals major differences in the way the distri­ bution process is developed. For example, Austin, Texas makes use of a simple graphic technique to produce a landuse plan, while Pittsburg uses a battery of regression and time series equations to distribute regional totals to small areas. One of the oldest methods examined is 29 the one used in Chicago, more familiarly known as the CATS method, or even better, the Density-Saturation Gradient Method (DSGM). Toledo utilized a unique landuse allocation method based on the Gompertz curve and the "holding capacity" concept. The rest of the studies made varied application of the well-known Gravity Allocation Model.

The details of these studies follow.

Akron, 0hio3

The allocation process is based on an analysis of socio-economic forecasts, landuse development potential, and regional and community landuse plans. The regional totals are distributed to counties, planning analysis zones (PAZ's), and finally to traffic zones. Vacant land is carefully evaluated for residential and other uses. Resi­ dential activity is allocated to small areas by making use of "resi­ dential rating scores" and applying the typical gravity allocation model. The method is simple. The non-residential landuse allocation process, concerned with the distribution of employment, non-residential acreage, and floor space for commercial, wholesale, industrial and public landuses, is based on existing consumption rates, depending on the type of activity and the location, i.e., urban, suburban, or rural.

Comments

1. The entire forecasting and allocation methodology is clear.

Whatever blemishes it does suffer from are pointed out.

2. The distribution of regional control totals to counties and 30 subareas, and then to Planning Analysis Zones, and finally to traffic zones, has been accomplished elegantly. Factors such as attractive­ ness, accessibility, etc., have been given numerical scores for purposes of distribution.

3. The prescriptive aspects of the forecasting process have not been clearly mentioned, although this has been implicitly built into the technique. The historical trends appear to be given a good deal of emphasis.

U. Locational decisions have not been recorded, neither has any thought been given to alternate urban forms. Whether several alternatives were considered at any stage is not mentioned.

i*. Citizens' preferences in the shape of a community survey have not been recorded. This is a serious omission.

Austin, Texas**

A simple trend methodology has been used to develop the landuse plan. The area is divided into eighteen subareas. These are thor­ oughly examined, analyzed and compared. The projections are based on trends. Land absorbtion coefficients for low density residential, high density residential, commercial and industrial landuses are established, providing rates of acres/unit or units/acre as the case may be. The entire process is transparent, relying heavily on a good data base and intimate knowledge of the area. 31

Comments 1. Although the method used is simple, much effort has gone into the analysis of a set of eighteen maps used in the process. 2. Apart from the usual trend analysis, the development between

1970 and 197^ has been carefully scrutinized. This appears to be the heart of the analysis.

3. An intimate knowledge of the prevailing conditions in the region can be readily detected on the part of the landuse planner.

k. Governmental control in Texas is a major problem because zoning is restricted only to land within the corporate limits of the city. This factor has been taken into account in developing the landuse plan.

5. The forecasts represent the continuation of existing trends. 6. There is no evidence or record of any goal formulation.

Because of the predictive nature of the procedures, this omission is not altogether serious. However, the fact that no alternative plans or other urban form concepts were even superficially examined is considered a weakness.

Canton, Ohio5 County-wide control totals of population and employment are distributed to districts and traffic zones by manual methods for the years 197 5, 198 0, 1985, 1990 and 2000.

The 1970-75 dwelling unit changes are used, along with "factor score ratings" and the amount of available land, for distribution of 32 dwelling units to small areas. The factor score ratings are calcu­ lated for each traffic zone based on a diverse list of items such as water and sewer service, distance from downtown, employment centers, etc.

The use of a capability analysis technique developed by the

Ohio Department of Natural Resources helps the landuse planner in making good rational decisions.

Comments

1. The method is basically sound and well documented.

2. Good use is made of the capability analysis of land devel­ oped by the Ohio Department of Natural Resources.

3. There is no mention of alternative landuse concepts, neither has any mention been made of goal formulation and its impact on guiding the region's development.

k o Nothing is known about decisions pertaining to the location of industries. 5. The general thrust is on landuse accounting and consumption coupled with methods of ensuring internal consistency.

Chicago, Illinois^ The Chicago Area Transportation Study (CATS) has provided an excellent account of their method, which has come to be known as the

Density-Saturation Gradient Model. It is one of the oldest methods examined. The landuse forecast is accomplished by a system of land accounting. It fits generally within a set of rules developed through observation of the existing regularities in the arrays of landuse • within the study area. There are three of these rules or regulari­ ties. The first is the decline in the intensity of land development as a function of distance (or access time) from the CBD. The second is the decline in the amount of land in use as a proportion of avail­ able land, which is related also to distance from the CBD. The third is the stability in the proportional amounts of land devoted to each type of landuse. These regularities are expressions of functional requirements of people and their activities. The first two are related to accessibility to the CBD, where 30% of the labor force is enployed and where the most specialized services are found. The landuse proportions are fixed largely by the basic requirements of the people.

Urban land, by reason of varying accessibility, is a limited resource. Because urban land is scarce and valuable, urban activities tend to arrange themselves economically with relation to one another over the land.

Comments

1. CATS has provided excellent conceptual explanations of their empirical findings, and rationale for their projections.

2. This forecasting model forces the analyst to become inti­ mately familiar with the study area before attempting to make a fore­ cast. This is considered a strong feature.

3. The graphic analyses that the method is based on represent 31+ descriptions of the key spatial relations of a metropolitan area, even for small areas. U. The method of analysis is useful regardless of the fore­ casting technique used. It can serve as a check or as a backup method if more sophisticated models are used. 5. The method outlined is not site-specific. The method was applied in Greensboro, North Carolina, and the results were encouraging.

6. The analytic work is a most valuable and rewarding feature of the use of this method. During this phase the analyst gathers an understanding of the structure of the study region and the forces which change this structure.

7 Columbus, Ohio'

The allocation of regional totals to small areas is performed in two stages. The short-range (19T^— 85) allocation is based on extending the present growth trend. The long-range (1985-2000) distribution, on the other hand, is tailored to respond to the new goals and objec­ tives formulated for the region. Good use is made of the graphic overlays indicating areas suitable for development. Careful use has also been made of a ranking system which utilizes a set of criteria.

Comments

1. The goals and objectives for the study area have been care­ fully worked out. This task helped the study in fixing criteria for allocating activities across the region. 35

2. There is no indication that alternative landuse concept plans for the year 2000 were prepared. It appears that since the 1985 concept plan was in existence, the question of having alternative plans did not arise. However, there is evidence that alternative trans­ portation plans will "be "based on the adopted landuse plan.

3. The two-stage allocation procedure is commendable. It demonstrates a certain honesty in recognizing that present trends cannot be halted all of a sudden to give way to the prescriptive plan­ ning process in the wake of fresh goals and objectives.

U. The allocation process is transparent.

5. Good use is made of data storage and retrieval.

Kansas City, Missouri®

The distribution of regional totals of population and economic activity to small areas is achieved by using a form of gravity allo­ cation model. The method relies heavily on a Land Factors Atlas providing essential information in the shape of several overlays.

Basically, two alternative concept plans are considered. Concept A is a conservative trend plan depicting what the region would be if the present growth continued. Concept B is prescriptive in character, taking into account the effects of urban development caused by new modes of transit and accelerated redevelopment. 3 6

Comments 1. The method outlined, although simple to look at, has heen

complicated by unnecessary detail. One would have imagined that with

a good graphical tool such as the Land Factors Atlas and holding

capacity figures, it would be fairly easy to accomplish the allocation

exercise.

2. The Land Factors Atlas has been well utilized.

3. The use of two concept plans provides a good basis of com­ parison.

Oklahoma City, Oklahoma^

The landuse planning process is based on a prescriptive plan rather than on a predictive one. The region consists of 3^ local jurisdictions, All the local jurisdictions agreed on regional control totals used. Each city or local government submitted its future landuse plan and the proportion of population increase it could absorb.

Expected densities were worked out and applied to each city’s landuse plan. In conference with local officials, citizens and planners the landuse plans were altered to match the control totals.

Comments

1. This procedure of allocation of population is highly effi­ cient. There has to be compatibility in handling densities and deter­ mining developable land in each traffic zone. 2. It would have been easier, perhaps, to have done the 3T allocation on a "stepdown" "basis, i.e., to Planning Areas, Traffic

Districts and Traffic Zones. This may have "been done, but it is not recorded. 3. The philosophy of the study has been cleanly stated as one of planning and prescribing rather than one of predicting.

U. The procedure of allowing each jurisdiction to develop its own landuse plan appears commendable, although one wonders how this was finally coordinated into one fabric. This procedure would, however, preclude the evolution of landuse alternative plans.

5. Good use was made of the feedback from citizens in developing and refining the landuse plan.

Pittsburgh, Pennsylvania^

The cooperative planning process used by the Southwestern Penn­ sylvania Regional Planning Commission has several features worth noting. In the first place, the blending of the technical process with policy considerations at almost every stage establishes confidence in the process. The models, set up to operate at two scales, regional and small area, are well documented. The small area allocation model makes use of regression equations.

Comments

1. The results from this long-range planning process were used to plan for transportation services, housing, water and sewer facili­ ties, open space projects, other programs and facilities. This 38 process was indeed comprehensive.

2. The resulting functional plans are internally consistent with

each other, in part "because they have "been planned to serve the same

numbers of people, and in part because they conform to the policy

guidelines adopted by the study.

3. Multiple regression models possess drawbacks. If the

dependent variable is expressed as an extensive quantity, such as

increase in dwelling units, measured relations with independent vari­

ables will be influenced by peculiarities of area definition and

size, and they might not conform to logical hypotheses regarding the land development process.

U. There is no built-in provision (as there is in other studies)

to assure that the accumulated zonal estimates obtained from the

regression equations will equal the actual total regional growth.

All regression estimates have, therefore, to be factored up to sum to

the actual regional growth.

5. The cooperative planning process uses citizen input and feedback very efficiently.

6. The steps used in allocating regional totals to small areas are well described. Although regression equations derived for this

study cannot be directly applied to other metropolitan areas, the

overall process can be easily used.

7. The policy input mechanism is commendable. SPRPC is a large area and it was refreshing to note how policy decisions were blended into the technical process. 39

8. The question of working out alternatives is not specifically mentioned. This consideration is probably implicit in the policy decisions.

Rockford, Illinois^

A dwelling unit allocation model was developed and applied by the Rockford Area Transportation Study (RATS) to allocate dwelling units to Winnebago County. The model used a gravity allocation tech­ nique to distribute dwelling units, first to the sectors and then to individual districts within each sector. Factors such as transpor­ tation capacity and linkage, employment, distance from the city center, and proximity of schools, were used in the model. The final, phase of the allocation distributed district totals to traffic zones.

Comments

1. The allocation model produced poor results.

2. New growth was projected to follow old growth and thus there is significant difference when the 1970-75 growth is compared to the model output. It may, however, be noted that this inaccuracy may be for the 1970-75 period, but not necessarily for the long term. Variation during different periods of time would depend upon the timing of public utility extension and the saturation of the sector.

3. Although it is somewhat arbitrary to analyze a 1970-95 allocation model based on 1970-75 data, further research ought to be done on refining this model. Some of the discrepancies may be due to Uo short-term occurrences, such as the sewer moratorium, the economic situation in Winnebago County, and the lack of demand for multi-family homes in certain sectors.

U. There are possibilities of using this model effectively with varying landuse concepts, based, of course, on goals and objectives framed for the region. No such attempt seems to have been made.

Toledo, Ohio'1’2 The Toledo Regional Area Plan for Action (TRAPA) used a good solid method for preparing a 1985 landuse plan. The method of analysis and distribution of population and economic activity to small areas uses the principles of "holding capacity" and the Gompertz curve.

Prescriptive planning has been applied, tempered by trend analysis. Two alternative concept plans are developed and finally merged into one. A thorough knowledge of the local area has been assumed as a prerequisite for the purposes of analysis. The procedures have been fairly well documented.

Comments 1. The logic applied throughout the landuse forecasting methodology is simple.

2. The use of Gompertz curves and the concept of "holding capacity" is commendable.

3. The idea of dividing the study area into 33 planning areas, although logical, creates problems insofar as it does not match civil hi divisions in the local area.

k. No use appears to have been made of plans,forecasts and data used individually by minor civil divisions. There are areas like

Maumee and Sylvania which had development plans of their own which could have been utilized.

5. As far as residential distribution is concerned, the allo­ cation to planning areas is documented, but no details are available about how this allocation was further distributed to census tracts or traffic zones.

6. A 197^-75 survey indicates that the results of this landuse forecasting procedure were sufficiently good.

7. As a planning model the procedure used throughout could easily be replicated in any other urban area.

Structural' Linkages of the Models

The structural linkages of the ten landuse models are indicated in figures 6—15. Basically these flow charts show the process of dis­ tribution of the regional totals to small areas. U2

Totals POP 4■ DU <— VAC . •> EMP s FL SP [«_ AC

- *f >r Process FR |— H GAM RATIOS

> Zones POPDU —? AUTO — COMM Y7HQLESALE IND

AC = Acreage; COMM = Conrmercial; IND = Industrial; DU = Dwelling Units; EMP = Employment; FL SP = Floor Space; GAM = Gravity Allocation Model; FR = Factor Ratios; POP = Population; VAC = Vacant Land

' FIGURE 6

STRUCTURAL LINKAGE, AKRON, OHIO

Totals POP EMP DU

Process TRENDS, LAND ABSORB COEFF, MAPS

Zones POP EMP DU

FIGURE 7

STRUCTURAL LINKAGE, AUSTIN, TEXAS Totals POP EMP

N/ Process GAM FR

N./ Zones POP — ^AUTO LAB FR ✓• FL SP

FIGURE 8

STRUCTURAL LINKAGE, CANTON, OHIO

Totals LU DATA REDEV PL POP ECON ZONING HC

DSGM Process

Zones LU, DU, POP, EMP

LU = Landuse; HC = Holding Capacity; DSGM = Density Saturation Gradient Method

FIGURE 9.

STRUCTURAL LINKAGE, CHICAGO, ILLINOIS kk

Totals POP, DU, ECON, 7^-85 POP, DU, ECON 85-2000

Process Trend Analysis Graphic Overlays Ranking Systems Criteria, FR, Consumption Rates. Prescriptive Forecast

Zones POP, EMP POP, EMP

FIGURE 10

STRUCTURAL LINKAGE, COLUMBUS, OHIO

Totals POP LANDUSE ALTERNATIVES

\ t Process LF'ATLAS GAM DU ANALYSIS

Zones POP EMP LANDUSE

LF ATLAS = Land Factors Atlas; GAM = Gravity Allocation Model

FIGURE 11

STRUCTURAL LINKAGE, KANSAS CITY, MISSOURI U5

Totals POP DU EMP LU PLANS VAC LAND

FIXED DENSITIES & RATIOS Process

Zones SOCIO-ECON VARIABLES

FIGURE 12

STRUCTURAL LINKAGE, OKLAHOMA CITY, OKLAHOMA

Totals POP DU EMP INCOME - [

\/ ■ Process Regression & Trend Eq.

\ / Zones POP DU EMP INCOME |

FIGURE 13

STRUCTURAL LINKAGE, PITTSBURGH, PENNSYLVANIA U6 Totals POP DU

V / Process GAM |

\f Zones POP DU |

FIGURE lU

STRUCTURAL LINKAGE, ROCKFORD, ILLINOIS

\

Totals POP EMP DU LANDUSE

Vr Process GOMPERTZ CURVES, HC, LAND CONS RATES J

/ Zones SOCIO-ECON VARIABLES

FIGURE 15

STRUCTURAL LINKAGE, TOLEDO, OHIO NOTES 1 Federal Highway Administration, Instructional Memorandum 50-1^-68: Operations Plan for Continuing Urhan Transportation Plan­ ning, 196 8. 2 For example see Boyce, Metropolitan Plan Making; W. L. Grecco, et al., Transportation Planning for Small Urhan Areas,'NCHRP Report No. 167 (Washington, D. C.: Transportation Research Board, 1976); R. G. McNulty, An Analysis of the Procedures for Forecasting and Distributing Future Landuse in Small Urhan Areas (Unpublished Master's Thesis, University of Tennessee, Knoxville, 197^)• 3 Akron Metropolitan Area Transportation Study (AMATS), Landuse and Development Forecast. 1970-2000, Technical Report (Akron, Ohio, AMATS, Jan., 1976).

City of Austin, 'Austin' Tomorrow: ‘ Landuse,'An' Analysis ~Of Urban Development in Austin (Austin, Texas: City of Austin, June, 1975). 5 Stark County Area Transportation Study (SCATS), Population and Economic Study, Small Unit Forecast (Canton, Ohio: SCATS, 1977)» Ohio Department of Natural Resources (0DNR),'Land‘Capability'Analysis, County Report No. 3, Stark County (Columbus, Ohio: 0DNR, June, 1975)» 6 Chicago Area Transportation Study (CATS), D At A Pro .lections, Vol. II (Chicago, 111.: CATS, i9 6 0); John R. Hamburg and Robert H. Sharkey, Landuse FOrCCast (Chicago, 111.: Chicago Area Transportation Study, 193171------7 Mid Ohio Regional Planning Commission (M0RPC), Franklin County Regional Transportation Plan (Columbus, Ohio: M0RPC, 1972); Mid Ohio Regional Planning Commission (M0RPC), A Technical Report on the Year 2000 Landuse and Trip Generation Variables (Columbus. Ohio: M0RPC, 1977).

®Mid America Regional Council (MARC), Methodology for Landuse, Population and Employment Distribution. Kansas City Metropolitan Region (Kansas Citv. Mo.: MARC, 1973). 9 Association of Central Oklahoma Governments (AC0G), Transporta­ tion Technical Reports, 1971-7^ (Oklahoma City, 0k.: AC0G, 1974).

^Southwestern Pennsylvania Regional Planning Commission (SPRPC), Forecasting Framework: ' Jobs, P66tile and Land. (Pittsburgh, Pa.: SPRPC, 1974). U8

"Rockford Area Transportation Study (RATS), A-Dwdllihg' Uhit Allocatioil Technique for Wihhebago County (Rockford, 111.: RATS, 1971).

^Toledo Regional Area Plan for Action (TRAPA), 1<}65 Regidrlal Landuse Inventory/Analysis and 1985 Forecasts (Toledo, Ohio: TRAP A, 1973). CHAPTER IV

A TEST OF TWO RESIDENTIAL LANDUSE MODELS

Introduction

As a result of evaluating ten landuse models in Chapter III, it was evident that some of them could profitably be investigated even further. The following criteria were established for selecting land­ use models for further examination: first, they must be based on a logically consistent organizing concept or theory; second, they must have some relationship to a phenomenon or process as it actually occurs or functions in the real world; and third, they must not be site-specific.^" The two techniques which best met these criteria are described below.

The first technique tested is the Chicago Area Transportation

Study method, familiarly known as the CATS method or the Density-

Saturation Gradient Method (DSGM), which is a simplification of the CATS method. The second technique tested is a modified form of the landuse forecasting method used by the Toledo Regional Area Plan for

Action (TRAPA). A summary of each of these methods is to be found in the Appendix.

The primary objective of this investigation and testing was to apply these two models in a common setting and to examine the results

k9 50 impartially. The research was not prompted by the desire to proclaim a winner from among the models tested.

This chapter records the results of an evaluation of these two operational residential forecasting methods applied to the City of

Toledo. This area was chosen primarily because it was felt that

Toledo was representative of the kind and size of city for which forecasting techniques of the kind being examined in this research would be appropriate. Toledo was also chosen because the author knows the city and its environs well. This acquaintance with the area is almost a prerequisite for applying many of the manual landuse forecasting techniques described in this research. Another reason for choosing Toledo was that a rather extensive information file on a small-area basis is available for two time periods— 1965 and 197^.

Thus, the two techniques were used to forecast landuse for 197^» given the 1965 base.

Methodological Problems

In a previous chapter it has been pointed out that traditional manual forecasting techniques are theoretically simple and operation­ ally straightforward. At the same time, it is worth repeating that although the allocation process in such models is based on acceptable planning standards, it is also dependent on professional judgment and can be subjective. In some cases, however, subjective judgment can be buttressed by principles, standards and techniques of planning to reduce the chances of making wrong decisions. Although the need to record and justify all of one's subjective decisions appears impor­ tant, it is virtually impossible to document all aspects of a landuse forecasting and distribution procedure which depends to a greater or lesser degree on subjective judgments. This poses a problem which has not yet been solved successfully.

The author's method of dealing with this problem has been to attempt to document not only the regular procedure of allocation, but also the subjective decisions taken at different stages of the exer­ cise. Where possible, the justification for doing so has also been recorded.

Data Sources and their Use

The data for testing these two landuse forecasting techniques came from several sources, chiefly the U.S. Census data, publications of the Toledo Regional Area Plan for Action (TRAPA), and the City of 2 Toledo.

It is important to state that since these tests were conducted ex post facto, no information beyond 1965 was used. Indeed, great cane was taken to completely blank out all information and data available after 1965» except the total 197^ population.

Model Performance and the U-Test One of the key considerations of any landuse model is its per­ formance characteristics. There are, however, many problems in assessing a model's performance. For example, in the case of a 52 predictive model, a test can be made using data from a point in the past, say 19^5* as the base year input. The model then produces a forecast for the present or some point in the immediate past, say

197U, for which verifying data is available. Since the model is operating in a period in the past, its performance under those cir­ cumstances does not necessarily insure that it will perform in the same manner trader a set of circumstances occurring in the future.

Again, a simple comparison of what a model predicts and what actually occurs can also be misleading. Poor results obtained from a model can be attributed to both measurement and specification errors.

For example, certain policy or zoning inputs may have been overlooked 3 in the forecasting process. Boyce and Cote , in an investigation of ways to verify and evaluate landuse forecasting models, make the cogent remark,

These [landuse] models cannot be verified in a strict sense because their formulations do not provide a con­ fidence statement about the relationship between the observed and predicted values.

It is unlikely that any of. these problems will be resolved in the near future.

In view of the problems cited, the decision was made to use a standardized statistical test to indicate the reliability of model forecasts, by means of the U-Test\ The U-Statistic is a measure of the statistical correlation between two sets of data. The U-Test is a statistical distribution test that measures the agreement between the forecast item and observed item frequency. The accuracy of the 53 forecast is judged by the magnitude of the U-Value. The U-Statistic is calculated as:

U

where = the projected value of zone i

C^ = the actual value of zone i

N = the number of intervals in the distribution

In general, a value of U less than 0.1 is considered good; a value between 0 .1 and 0 .3, average; and a value of greater than 0 .3, poor.

Past applications of the U-Test have been satisfactory.

The Density-Saturation Gradient Method (DSGM)

The density-saturation gradient is, as the name suggests, a plot of the residential saturation as a function of distance or travel time from some convenient reference origin— usually the city center.

It is a function that can be determined experimentally for any city, although practical difficulties are encountered because of topo­ graphical and zoning peculiarities. It is sometimes necessary to define the function separately for a number of sectors of the sur­ rounding metropolis. The only known previous application of this

approach was for the Chicago area'. Swerdloff and Stowers^ later applied it in a test situation to Greensboro, North Carolina, and the procedures detailed in their paper were used by the author. Population Density (Persons/Res. Acre) 1965* 80 r 60 20 0 *See table 22 in in Appendix. 22 table *See OUAINDNIYB ITNEBNS ( BANDS DISTANCE BY DENSITY POPULATION 2 itnefo V (in miles) HVC from Distance k FIGURE FIGURE 6 1 6 65 96 1 8 ) 55

The DSGM was applied to the City of Toledo using air-line distance from the high value corner (HVC) as the key spatial variable.

The HVC is a point representative of the hypothetical activity center of the central business district. Figure 16 shows the relationship between 1965 residential density and air-line distance from the HVC. Each point on this plot represents the residential density for a ring around the HVC. The decline in density results from the operation of the competitive land market.

Each ring is defined by the boundaries of all census tracts whose centroids fall within £ 1 /2 mile of the distance of the ring from the HVC, with the exception of the first or CBD ring. The plot indicates a regular decline in residential densities with distance from the HVC. The next step is to compute the holding capacity.

Mathematically, the holding capacity of an area is defined as the existing population plus the product of vacant, available, suitable land and the expected residential density. Theoretically, this den­ sity is the anticipated average density at which all future residential development will occur.

These values can be developed from an intensive analysis of trends in residential density patterns and zoning policies. For pur­ poses of this investigation, future densities for each census tract are assumed to be those given by the smooth hand-fitted curve of figure 1 6. Vacant, suitable land for residential development was estimated from figures available from data for planning areas. Also, from zoning ordinances and zoning plans, it was possible to estimate land avail­ able or earmarked for residential use. Having future residential densities and vacant available land for residential use, it was possible to compute residential holding capacities and residential saturations (1965 population/residential holding capacity) for each census tract. The latter values were then used to construct the percentage saturation gradient for 196 5, shown as curve "a" in figure

17. Tables showing the necessary calculations for setting up the graphs are placed in the Appendix (see tables 21-27). No particular judgmental decisions were needed or used up to this point in the forecasting procedure.

The plot of the 1965 population/residential holding capacity conforms very well with the expected plot for an urban area shown in figure 17. The next step involved the 19lb projection of the per­ centage saturation curve, also shown in figure 17. This is the most critical and subjective step in the forecasting process, since the only restraint on the projected curve is that the area under the new curve must account for the existing and projected regional growth. The population in the study area grew from a I96U total of 3 2 ^,3 8 5 to

332,2l<-0 in 197^, or a growth of 2.U2/S. From trends of population move­ ment it was generally known that the center city was rapidly losing population, and that areas outside the study area were gaining popula­ tion.

One can proceed in an almost infinite number of ways in estab­ lishing an acceptable projection of the percent saturation gradient % Population Saturation 80 60 UO 20 b a c 1965 9^Pp culHligCpct (see table Capacity Pop.26197^ Actual/Holding ) 197^ Pop..(see table Capacity 25Forecast/Holding ) PERCENT POPULATION SATURATION BY DISTANCE BANDS BYDISTANCE SATURATION POPULATION PERCENT Pop./ Holding Capacity (see table. Capacity Holding Pop./ FIGURE FIGURE 17 2k,) J. 7 miles 58 for 197^ population. It was, however, found useful first to develop a feel for the overall scale of the problem by making the area under the final curve commensurate with the required total population.

The following general procedures were considered in order to fix the

197^ saturation gradient:

1. Using the 1950 and i960 census figures for population, two saturation gradients for 1950 and i960 were constructed, similar to curve "a" in figure 17. From these curves it was observed that the population at the city center was declining and the slope of the saturation gradient was becoming flatter. Thus the pattern of densities prevailing in 1965 represented a kind of equilibrium between the cost of land, building costs, locational requirements, and trans­ portation costs. Thus the 1950, i960 and 1965 saturation gradients provided a rough estimate of the proportional changes one could expect for a 197^ saturation gradient (see figure 1 8). 2. Building permits obtained from the City of Toledo for the period 1959 to 196U were grouped by census tracts and rings. The increments in housing and thus population were aggregated and used as trends for the 1965-7^ period.

3. The zoning ordinances of the City of Toledo were examined and census tracts falling into various categories were aggregated by ring. This helped to set saturation gradient figures at the ring level.

Based on the results from procedure (l), a rough 197^ satura­ tion gradient was drawn. This was adjusted to reflect the trend and location of growth obtained from procedures (2) and (3). The final % Population Saturation 100 80 20 Uo 60 - - - 1965 I 960 ECN OUAINSTRTO KTH 1950-60-65 SKETCH: SATURATION POPULATION PERCENT 1950 2 5 3 k FIGURE 18 FIGURE 6 „ 7 miles from HVC0 from 7 miles 1 1965 59 60 adjustment was made to assure that the area under the 197*+ curve was commensurate with the 197*+ total population.

Several attempts were necessary to arrive at a solution which reflected the total 197*+ population. A smooth hand-fitted curve was then drawn to represent the 197*+ saturation gradient. This curve is shown as "b" in figure 17. Multiplying the appropriate ordinate value from this forecast percent saturation gradient for 197*+ by the ring saturation quantities established the forecast population totals by analysis ring. (These are shown in columns 1-*+, table 25 , in the

Appendix.) The question now remained of distributing these ring totals to individual census tracts. This distribution is dependent on several factors, such as accessibility, water and sewer facilities, etc. In order to measure the strength of the residential development poten­ tial, an activity allocation process was formulated. The Akron

Metropolitan Area Transportation Study (AMATS) and the Stark County

Area Transportation Study (SCATS) have used residential development factors in the past in their activity allocation process. This activity allocation process, as with most manual distribution tech­ niques, depends on judgment based on the understanding of the factors which promote growth. The specific values assigned to these factors were selected to reflect the comparative impacts that sewer service, water service, accessibility, etc., have on residential growth. Based on AMATS and SCATS1 residential development factor rating procedure, a similar rating procedure was developed for this study. See table 2. 6l

TABLE 2

RESIDENTIAL DEVELOPMENT FACTORS

Maximum COMMUNITY FACILITIES Points Central Sewer System Service Existing (1965) 20 20 Planned to be in operation by 1975 15 Planned to be in operation by 1980 10 Central Water Service Existing (1965) 20 20 Planned to be in operation by 1975 15 Planned to be in operation by 1980 10 School Elementary school within 1/2-mile radius 5 5 ACCESSIBILITY CBD 0 -5 minutes 5 5 6-10 minutes 3 11+ minutes 0 Major Shopping Center 0-3.^ miles 5 5 3.5-6.7 miles 3 6.7+ miles 0 Within the census tract 0 Major Employment Center 0 -5 minutes 5 5 6-10 minutes 3 10+ minutes 0 Within census tract 0 Highway System Census tract within 11/2 mile of major arterial or freeway interchange 10 10 Mass Transit System Established bus route within l/U-l/2 mile of census tract 5 5 ACTIVITY PATTERN Existing Landuse Industrial Park 0 k Subdivision U Commercial Center 0 62

TABLE 2 (Continued)

Maximum Points Population Change (1960-65) Greater than 25$ ^ 10-25$ 2 Less than 10$ 0 Major Recreational Center Park within 2 miles of census tractk Park within 5 miles of census tract 2 Park available beyond 5 miles of census tract 0

TOTAL 87 pts. 63

There is no question that the allocation of ring totals to

individual census tracts is subjective even though there is a

certain uniformity and consistency in this subjective allocation.

However, it should be pointed out that in this allocation framework,

certain basic needs of human interaction are recognized. The need of households, firms, and other activities to interact and maintain

proximity to one another sets in motion location decisions. Taken as

a whole, these decisions constitute the development process.

After the initial distribution from ring totals to census tracts was performed, two further checks were made: first, to make sure that

a particular census tract had not been allotted a population in

excess of its residential holding capacity; and second, to determine whether a given census tract was in conformity with the applicable

zoning ordinance.

The final distribution of population for 197h to census tracts is

given in table 2J in the Appendix. Table 3 shows the results of

applying the U-Test at the ring and census tract level. The accuracy

displayed is excellent, even though the forecasting method was

applied to the City of Toledo with a certain amount of crudeness.

One would therefore conclude that the DSGM is simple, straight­

forward and easy to operate even for cities the size of Toledo. The

only caution that one may observe is that this method requires inti­

mate knowledge of the city and its zoning ordinances. 6H

TABLE 3

U-TEST AT RING AND CENSUS TRACT LEVEL

Forecast Actual 2 2 ^ Dist 19TU 197^ (Si - Ci) s± Ci Si C.

0 1 ,6 0 0 1,955 126,025 2 ,560,000 3,822,025 X HU,679 1*2,1*60 H,923,961 1,996,213,oUl 1,802,851,600 2 6o,6Hl 5 6,8lH lU,61*5,929 3,677,330,881 3,227,830,596 3 35,756 36,355 358,801 1,278,1*91,536 1,321,686,025 1* 56,771* 55,386 1,926,51*1* 3,223,287,076 3,067,608,996 5 58,260 55,158 9,622,1*01* 3,39l*,227,600 3,0l*2,l*0l*,961* 6 39,787 1*7,161 5 8,890,276 1,559,223,169 2,22U,159,921 7 31*.71*3 3l*.l*7l* 3.61*0,1*61* 1,060,51*1*, 356 1.188.1*56,676 332,2H0 332,21*0 91*,131*,1*0U 1.619187766 1.56788206 x 1010 x 1010

LStSj. - Ci.)2]1/a U = "" = 0.0383 (The accuracy is thus excellent.) .2 .1/2 Pr. . .2,1/2 C£CSi) r + [£(c.) r

Dist­ ance Actual Forecast Differ- % Band______19-71*______19 71*______ence Dif f.

0 1,955 1 ,6 0 0 355 18#* 1 1*2 ,1*60 1*1*,679 -2219 - 5% 2 58,811* 6 0,6H1 -3827 - 1% 3 36,355 35,756 599 2% 1* 55,386 56,771* -1388 - 3% 5 55,158 58,260 -3102 - 6% 6 1*7 ,1 6 1 39,787 7371* 15% 7 31*, l*7l* 3H,71*3 - 269 1% *This is the worst.

Ring 0, which is the worst, contains only Census Tract 28. u = This is good. 65 The Toledo Method

The Toledo Method combines the concepts of holding capacity,

the use of Gompertz curves, and land consumption rates in the dis­

tribution of an areawide forecast of population to small areas. The method has been well documented and a summary is to be found in the

Appendix, where an illustrative example has also been included. The

Toledo method has not been applied to any other metropolitan area,

although many study areas have adopted some of the principles on which

it is based. The Toledo method, with some refinements and adjustments, has

been applied to the City of Toledo, west of the Maumee River, to an

area identical to that used in the DSG Method previously described.

The small area unit used is the census tract. However, the area under

consideration is first demarcated into twelve planning areas. Census

tracts with similar characteristics have been grouped into a single planning area (see table 28 in Appendix C).

Table U is an inventory of the total acreage, 1965 residential

acreage, additional land for residential use, and maximum land avail­

able for residential use. The number of dwelling units in 1965, the base year, and the maximum number of dwelling units at holding capacity

for every planning area are shown.

It was possible to estimate land available or earmarked for

residential use from 1965 zoning plans, the land capability analysis,

and 196U aerial photographs. 66 The holding capacity of an area, as was stated earlier in the

description of the DSG Method, is the existing population plus the product of vacant, available, suitable land and the expected density.

In the Toledo method, dwelling units are substituted for population while calculating the holding capacity. Thus the ratio of dwelling units in a planning area to the holding capacity determines the stage

in the development cycle the planning area has reached in the base year (1965). It also provides the basis for estimating at which

stage the area will be at a future date— in this case, 197*+.

The question of establishing the development cycle was taken up next. Two neighborhoods in the central part of the city were investi­

gated for estimating the average number of years required to achieve

a given stage of development. From this investigation and analysis

a typical development cycle of approximately fifty years was selected

(see table 5 ) and applied to the planning areas to estimate the percent growth of dwelling units in each planning area, depending

on the stage of development each area is presently in. This growth is shown in table 6. It shows the forecasted dwelling units for 197*+.

The forecast population for each planning area is shown in table 7.

The vacancy and occupancy rates adopted in table 7 were derived by straight-line projections of corresponding i960 and 1965 figures

to 197*+. The only information used was the total 197*+ population of

332,2*+0 .

The distribution of the 197*+ forecast population by planning areas to census tracts was perforated by making use of the residential 67

TABLE k

CAPACITY LANDUSE PROJECTIONS BY PLANNING AREAS IN CITY OF TOLEDO

Addn. Max. DU's 1965 1965 Land Land DU's at DU's as Planning Area Total Res. For Res For Res in' Capa­ % of Acres Acres ' Use ' Use 1965 city Cap.

Old Orchard 3578.1 102U .8 6 9 .2 1091+.0 11531 11531 100 West Toledo - A 1915.8 8 66 .6 1 2 .1 878.7 9096 931+9 100 West Toledo - B 202U .1 975.5 6 9 .6 101+5.1 71+52 71+52 100 Mayfair 261U .2 578.3 273.3 85 1 .6 5127 7665 66.9 Trilby 3619.8 1331.6 897.9 2229.5 6212 101+79 59.3 Talmadge 1685.1 676 .6 1 70 .0 8 U6 .6 3150 3979 79.2 Heatherdowns 5219.5 1U90.3 551.-9 201+1.9 8183 11962 6 8.1+ Reynolds Corner 5279.9 1356.6 1290.0 261+6 .6 5536 11930 1+6.1+ Airport Hwy. 278 2.2 223.7 500.0 723.7 797 2605 30.6 Point Place 2101.2 623.8 327.0 950.8 1+597 6ll6 75.2 Fort Industry 3566.9 93.5 93.5 228 228 100 Inner Core 68785 68785 100 Lagrange 21+13.3 801.9 30.3 832.2 Center City 8U6.3 131.1 131.1 Dorr 1959.8 88 9 .9 889 .9 North End 1992.5 1+01.5 1+01.5 Old West End 1358.3 700.8 700.8 South Side 2919.7 9I+O.O 103I+.7

TABLE 5

DEVELOPMENT CYCLE: ESTIMATED YEARS REQUIRED TO ACHIEVE GIVEN STATE OF DEVELOPMENT

% of Annual Approx. Capacity Growth Number Devpd. Rate of Ye Very Slow Growth 0-10 £ 156 10 Slow Growth 11-20$ 2$ 5 Moderate Growth 21-1+0$ 3$ 7 Boom Period 1+1-60$ l+$ 5 Moderate Growth 61-80$ 3$ 7 Leveling Off Fast 81-90$ 2$ 5 Leveling Off Slow 91-100$ 1$ 10 approx. 50 yrs. 68

TABLE 6

FORECAST GROWTH OF DWELLING UNITS IN PLANNING AREAS

Fore- DU's % of Growth cast Planning Area in 1% 2% 3%' k% 3% 2% 1% DU's 1965 0 -1 0 11-20 21-Uo Ul-6 0 6l-8 0 81-90 91-100 191k % of Capacity

Old Orchard 11531 no expansion 11531 West Tol. A 9096 no expansion 9096 West Tol. B 7^52 no expansion 7U52 Mayfair 5127 3/h* 2/5 6371 Trilby 6212 k/1 3/7 2 /1 810 U Talmadge 3150 3/1 2/5 1/3 3690 Heatherdowns 8183 3 A 2/5 10168 Reynolds 5536 k/3 3/6 7^36 Airpt. Hwy. 797 3/3 k/6 1102 Point Place U597 3/2 2/5 1 /2 5k93 Fort Industry 228 no expansion 228 Inner Core 68785 3/9 (decline) 51589 •denotes a 3% growth per year 'over a U-year period 69

TABLE 7

FORECAST AND ACTUAL POPULATION FOR 197^ Raw Final Fore- Fore- Fore­ cast Vac an- cast cast Actual Forecast Planning Area 197^ cy Occp 19Jh 19Jb 197k Actual ...... DUrs Rate Rate Pop. Pop. Pop. '

Old Orchard 11531 k.l 2 .6 28751 30220 27530 1 .1 0 West Tol. A 9096 3.1 2.7 23798 25014 25111 1 .0 0 West Tol. B 7^52 3.U 2 .8 20156 21186 21395 .99 Mayfair 6371 3.0 3.0 185U0 19U87 18474 1.05 Trilby 8104 2 .6 3.1 24471 25721 29935 .86 Talmadge 3690 2 .6 2 .8 10065 10579 11314 .9b Heatherdowns 10168 3.4 3.0 29U69 30975 3657^ .8 6 Reynolds 7U36 2.7 3.0 21704 22813 21151 1 .0 8 Airpt. Hwy. 1102 3.6 3.0 53187 3350 4387 .76 Point Place 5^93 2.5 3.1 16602 17^50 16652 i.o4 Fort Industry 228 2.5 3.3 733 770 51b 1 .5 0 Inner Core 51589 4.2 2 .k 118612 124675 119203 1 .0 5 316088 3322*10 332240 development factors described in the DSG Method, where the use of

such factors is duly justified. The 197^ forecast population by

census tracts is shown in table 29, which is placed in the Appendix.

It may be stated that after distributing the planning area population to census tracts using the residential development factors,

it was necessary to assure that the zoning ordinances applicable to

different areas and census tracts were not violated, and also to

verify that the residential holding capacities were not exceeded for

any census tract. Accuracy checks in the form of U-Tests are applied on all the planning areas taken together, and then on the single planning area that performed worst. The results were quite satisfactory. These

accuracy checks are shown in tables 8 and 9.

One may conclude from the above exercise that the Toledo method

is an excellent procedure for distributing regional control totals to small areas. Above all, the results are excellent, both at the planning area and the census tract level.

Conclusions

The two traditional manual landuse forecasting methods described in this chapter are sufficiently accurate to be recommended for use

in small and medium-sized cities. The results produced were quite satisfactory. Both the methods almost force the analyst to become

intimately familiar with the study area, the zoning ordinances, the

physical characteristics, and the growth trends before attempting to TABLE 8

U-TEST AT PLANNING AREA LEVEL

Ci Actual 1 9 ^ Planning Area 1971* Forecast (Si - Ci)‘

Old Orchard 27530 30220 7236100 91321*81*00 757900900 W. Toledo A 25111 25011* 9U09 625700196 630562321 W. Toledo B 21395 21186 1*3681 1*1*881*6596 U577H6025 Mayfair 181*71* 191*87 1026169 37971*3169 3^1288676 Trilby 29935 25721 17757796 66156981*1 89610U225 Talmadge 11311* 10579 51*0225 11191521*1 128006596 Heatherdowns 3657*1 30975 3131*8801 9591*50625 1337657^76 Reynolds 21151 22813 276221*1* 5201*32969 UU736U801 Airport Hwy. 1*387 3350 1075369 11222500 1921*5769 Point Place 16652 171*50 636801* 301*502500 27728910U Fort Industry 51** 770 65536 592900 26U196 Inner Core 119203 121*675 2991*2781* 1.551*385563 x 10 1.1*20935521 x 10±u 33221*0 33221*0 921*1*1*918 2.0U8108057 x 10 1.95027853 x 10lu

(SfSj - Cj)2 ]172 u = = .031* (The accuracy is thus excellent.) it (S.)2 ]1/2 + [ l ( Ci)2]1/2 TABLE 9

U-TEST AT CENSUS TRACT LEVEL

Refer table 6. The two planning areas that had the largest percentage difference are shown below.

Airport Highway-

Actual: U387 Forecast: 3350

U = .131*, which is good.

Fort Industry-

Actual: 511* Forecast: 770

U = .199* which is average. forecast. This is considered a positive feature of the methods. In any case, the methods of analysis described here are useful tools which can stand on their own or even serve as checks on the reason­ ableness of forecasts produced by the more sophisticated computer- oriented models. NOTES

•^F. Stuart Chapin, Urban Landuse Planning (Urbana, 111.: Uni­ versity of Illinois Press, 1965), p. ^75. 2 A list of sources used 1 as follows:

U.S. Census of Population: 1950, Toledo, Ohio U.S. Censuses of Population & Housing: I960, Toledo, Ohio Toledo Regional Area Independent Variables, TRAPA, 1965 Economic Forecast, TRAPA (1965-1 9 8 5) Population Projections, TRAPA (1965-2015) Toledo Total Population, 197^ Zoning Ordinances, 1950, i960, I96U, 196 5, City of Toledo Landuse Inventory, City of Toledo, 1965 Sewerage and Water Services Data, City of Toledo, 1965 Physical Features, Toledo, 1965 Landuse Concept Plan for TRAPA, based on Goals and Objec­ tives of the Region, 1965 Goals and Policies of TRAPA, 1965 Building Permits, City of Toledo (1960-6U) Aerial Photographs, City of Toledo, i9 6 0, 1962, 196b Maps showing Census Tracts, Toledo, Ohio, i960 Maps showing Major Shopping Centers, Toledo, Ohio, 196U Maps showing Transportation System, Toledo, Ohio, I96U Maps showing Bus Routes in Toledo, 196k Maps showing Skim Trees, Isochrones, TRAPA, 196U Lucas County map showing Highways, 1965 3 David E. Boyce and Roger W. Cote, "Verification of Landuse Forecasting Models: Procedures and Data Requirements." Landuse Fore­ casting Concepts, Research Record No. 126 (Washington, D.C.: Highway Research Board, 196 6) p. 60. U Informational Report, Landuse and Demography: Growth versus Forecast, ITE Technical Committee 6F6, Traffic Engineering. March, 1977, pp. U2-UU. ^Carl N. Swerdloff and Joseph R. Stowers, A Test of Some First Generation Residential Landuse Models, Landuse Forecasting Concepts. Research Record No. 126 (Washington, D.C.: Highway Research Board, 1966) pp. 38-59. CHAPTER V

RECOMMENDATIONS AND GUIDELINES

Overview

In urban transportation planning studies, landuse forecasting

(or activity allocation) refers to estimating future amounts of

development for small areas (usually traffic analysis zones). The

development includes socioeconomic variables, such as several strati­

fications of population and employment (which are normally vised in trip generation), in addition to land usage.

Reference has been made in previous chapters to the 1962

Federal-Aid Highway Act, which required that transportation invest­ ments in urban areas with a population of 50,000 or more be predi­

cated on a continuing, comprehensive and coordinated (3-C) trans­ portation planning process. Concern for the need to recognize the mutual impacts of landuse and transportation was clearly expressed through the requirements contained in Policy and Procedure Memorandum

50-9.1 Although this memorandum provided broad recommendations and

guidelines, it did not specify how the landuse planning process should be accomplished. The method of forecasting and distributing regional totals of population, economic activity and landuse to small areas to result in some form of a descriptive future state of the region was left to the discretion of the planner. 75 76

This chapter attempts to provide guidelines regarding landuse planning in general and the distribution of regional totals to small

areas in particular. While these recommendations and guidelines are based on a critical survey of landuse forecasting and distributing methods used in the ten study areas researched, it also draws on the

experience of similar studies done in many other metropolitan areas.

General Guidelines Pertaining to Landuse Planning Basically, there are two approaches to determine what the future landuse pattern will be. One is to plan or design this future pattern and the other is to forecast the amount, type, and location of the activities that form the pattern.

In actual practice, landuse forecasting is a combination of planning and forecasting. Planning implies that urban development

controls are utilized in order to achieve a desirable future plan (i.e., landuse pattern), while forecasting implies an extension of past relationships in development. A balanced mixture of each is required in order to predict a future development pattern that is

consistent with trends els influenced by a reasonable expectation of the exercise of various controls. Setting up control procedures to achieve a desired future state relies on policy inputs, goals and objectives. It is therefore essential to frame policy inputs in cooperation with the communities involved.

Community goals and policy inputs lead to selected future urban forms. An effective way of dealing with future urban forms is via sketch planning techniques, for future optimal arrangements of the T7 area.. This approach provides the planner with the opportunity to experiment with arrangements of landuse which might conceivably have a much higher level of performance than would ordinarily be dis­ covered.

Sketch plans prepared by planners can most profitably be exam­ ined by developers, policy and decision makers, and by the general citizenry for criticism and feedback. Indeed, this feedback should eventually lead to the development and presentation of alternative plans for consideration by decision makers. The number of alternative plans depends on the complexity and size of the region. However, at least two alternatives should be developed.

The use of citizen committees has been a standard practice for many years. Advantage should be taken of citizen involvement in developing alternative sketch plans, because such involvement moder­ ates the technical recommendations of professionals by filtering them through a group knowledge of community values and desires. It also aids the professional in understanding the problems of communities from a closer viewpoint.

Once the goals, objectives and policies have been framed and accepted, and regional forecasts of population and economic activity have been worked out, the decision must be made to select a suitable landuse model for use on a continuing basis.

One of the preliminary concerns is the desired use of the model.

Certain basic outputs produced by a landuse model for transportation planning purposes, such as small area population and employment, 78

satisfy many planning tasks in environmental planning, solid waste

disposal programs, housing, etc. However, in a comprehensive trans­ portation planning concept, the two models elaborated in Chapter IV may well serve most planning situations. Of course, no model con­

struct can be flawless and absolutely consistent. There is a con­

stant compromise between operationality and descriptiveness. However,

it must be said that whatever model is accepted should be transparent

to decision makers and the citizens. Any proposed use of a model in

a "black box" approach should be avoided, to maintain credibility.

Despite the recent increase in information sources, data

collection and analysis is still a large portion of any modeling effort

and may consume more than half the project's resources. In some transportation studies, data collection and processing absorbed about

60% of the total budget. Hence every effort should be made to mini­

mize the amount of data collection consistent with the time and

budget available. For the DSG Method (CATS) and the Toledo method

the data for only one point in time is really necessary, although

historic data is always of considerable help in establishing trends.

In those instances where the more sophisticated computer- oriented landuse models are being considered for large regions, it may be advantageous to place the initial work connected with this

sophisticated landuse model off-line, or off the critical path. In

such cases, one of the manual methods mentioned can be used as a backup strategy. 79

In general, it is recommended that planners forego computer- oriented landuse models in favor of manual, methods of distribution.

These latter are heavily dependent on the planner’s knowledge of the community and the exercise of professional judgment in an ad hoc or opportunistic fashion. This recommendation is particularly applicable to study areas of a population of one million or less. In communi­ ties under 5 0,0 0 0 , for example, a planner can make gross estimates of the amount of various landuses needed at some future date on the basis of population and economic studies. The required landuse activities are then spatially distributed, more by design principles than by projections, while taking into account the capacities of vacant land and proposed public improvements as well as the planner’s knowledge 2 of local development trends, land availability, and similar factors.

In summary, the decision to select a particular analysis technique should be based on the following conditions: 1. Sensitivity of the forecasting procedure to the underlying transportation issues, particularly of trip generation.

2. Ability to provide results meaningful to the decision maker, the citizenry, the local officials and others interested in the overall planning process.

3. Compatibility with the degree of sophistication and time requirement appropriate for the region and/or the community.

k. Availability of data and other informational and compu­ tational facilities.

5. Availability of manpower and technical expertise. 80

In recognizing the importance of the continuing nature of plan­

ning, the Federal Highway Administration has established an elaborate

procedure for the continuous monitoring and periodic review of the

adopted transportation plans in urbanized areas with 50,000 or more population. In addition to current estimates of growth annually,

the procedure requires reappraisal of the transportation plan at three levels of intensity: routine annual review, major review every five

years, and plan reevaluation at least every ten years. The routine

review is based primarily on local traffic counts and a growth analysis

of urban development. Similarly, the major review does not require

any major data collection or adjustments. However, the plan reevalu­

ation requires a large investment in new data collection and model

validation. If the surveillance and reappraisal procedures are honestly adhered to, it is certain that large savings could be

effected. Moreover, this procedure provides the opportunity of per­

forming accuracy checks which are of profound significance to the

study. If a landuse forecasting methodology is not providing good

results, the administration can make alternate arrangements to use

other methods. In some cases the inaccuracies may be due to local

conditions such as changes in policies or alterations in zoning.

This brings up the importance of documentation. It behooves

every planning agency to record local value systems and physical or

economic peculiarities which influence planning. Assumptions and

methods of making forecasts and developing alternatives ought to be

recorded meticulously. 81

In study areas made up of several constituent local governments, the larger units sometimes prepare landuse comprehensive plans on their own. It is recommended that full use he made of such plans hy set­ ting up a suitable mechanism of integration.

Specific Guidelines

The first step in organizing an urban landuse planning program is to obtain agreement on the procedure for conducting the study from the various jurisdictions and agencies having a responsibility for transportation and landuse in the study area. The landuse study involves a wide variety of tasks aimed at providing an accounting of the current landuse activity structure of the study area and the most probable or desirable future structure.

All landuse studies include the following items for the entire study area:

1. An inventory of the location and intensity of existing land­ use activities, including vacant land.

2. An analysis of past trends to aid in determining land con­ sumption rates, and the most likely location patterns of households and business firms.

3. The distribution of an area-wide forecast of population and economic activity to small areas (zones).

The scope of the landuse inventory may vary with each procedure, k but a list of commonly employed inputs is developed to serve as a good reference point: 82

1. Graphics: aerial photographs, hase maps, identification, measurement and location of landuse.

2. Regional activity totals for future years: future popula­ tion, economic activity or employment.

3. Existing development: current economic conditions, popular tion and employment distribution, existing landuse patterns, and so on.

h. Physiographic influences on development: topography, soils, geology, climate, and other physical features which may influ­ ence growth.

5. Vacant land: the amount and location of existing vacant parcels of land and the usefulness of the land for future urban development.

6. Utilities: the extent, capacity and characteristics of utility systems as a constraint on future urban development.

T. Transportation: the existing transportation system as a determinant of future urban development, to include accessi­ bility factors.

8. Public policies: referring primarily to local government action or inaction in terms of their influence on future landuse procedures. Some of the more important public policies include zoning regulations, subdivision regulations, redevelopment policies, and master plan considerations (principles and stan­ dards for development, landuse planning concepts such as pro­ viding for compatibility of uses, existing plans for the com­ munity, and so on).

9. Social and community value factors: goals and objectives for the urban development process and other value factors which relate to community development.

10. Recent past trends of change: basically the record of the inertia of recent urban growth.

A Land Capability Analysis Study similar to the one developed by the 5 Ohio Department of Natural Resources' • should prove of immense value in this phase of the study. 83

Tlie analysis phase of the landuse study should determine:

1. Patterns of future land development

2. Patterns of population density

3. Trends in land utilization

k. Absorption of vacant land

5. Character and suitability of vacant land

6. Areas of transition to changing landuse

7. Physiographic influences on development patterns.

In this analysis the planner should make full use of the aerial photo­ graphs. The zoning ordinances and other state and local controls published from time to time should be used here also.

Landuse density measurements are of particular significance in transportation planning. Indices of residential densities may be included in modal split models in addition to the behavioral charac­ teristics of car ownership and income. The density of commercial development is also useful in traffic studies such as parking analysis.

The distribution phase of the landuse study has been illustrated - by two methods detailed in Chapter IV. The specific information to be provided by the forecast will vary with the size of the study area. However, a list of typical outputs by small area is as follows:

1. Population: the forecast and allocation of future population

2. Number of occupied dwelling units

3. Net residential density at future dates

k. Retail activities: may include future employment, future sales, floor space, etc. 8fc.

5. Amount and location of nonresidential land by major use

6. Amount of vacant land

7. School enrollment

8. Auto ownership c One approach of analysis and distribution is shown below.

Step One: Prepare Change Statements for Existing Urban Area

A. Analyze each planning area in turn: (1) Existing landuse (2) Existing landuse-activity relationships (3) Trend change in landuse (1*) "Desired" change in landuse (5) Estimate future landuse

B. Sum and check with independent activity estimate for urban area.

Step Two: Estimate Future Additional Land Requirements

A. Siam activities for all planning area (above) and subtract from estimate of total activities in design year.

B. Translate future, unallocated activities to land require­ ments .

Step Three: Prepare Change Statements for New Urban Areas

A. Analyze each planning area in turn: (1) Existing landuse (2) Constraints to development (3) Developmental pressures (U) Public policy with respect to development (5) Estimate future total probability for development B. Allocate Propoortion of total land requirements (2B above) to each planning area.

C. Check for constraints on available, developable land (1) Change market pressure, if necessary (2) Change public policy, if necessary

D. Sum and check with independent activity estimates

Step Four: Review and Adjustment 85

It is a kind of checklist which might help a planner to run through

a list of tasks. It involves a detailed review of small area charac­ teristics and a determination of optimum future conditions for each

small area modified for certain obvious constraints of practicability.

In this type of analysis, the analyst moves from consideration of the

future of one small area to the next, taking into account all of the

facets of existing, probable and desirable landuse mixes within

each area. This is accomplished, of course, by reference to existing

local conditions and to certain broad regional studies, which provide overall totals and market constraints applicable to every area.

While working with planning areas and/or census tracts the fol­

lowing information should be fully exploited:

1. Published plans of individual communities

2. Zoning ordinances

3. Reports from utility companies

H. Forecast and general reports of established industrial corporations in the region

5. Information of building permits and demolition from appro­ priate authorities

6. Water and sewer facility expansion information

7. Transportation facilities plans

Much information can also be obtained from developers and community leaders. 86

Conclusions

The principal task of this research was to examine several methods used by transportation planning agencies in forecasting landuse. More specifically, the research focused on assessing methods of distributing regional totals of population, economic activity, and other variables to small areas, such as sectors, districts, census tracts and traffic zones. The methods chosen for examination were usually "manual” or traditional. Two methods in particular were closely examined to ascertain their suitability and accuracy.

The findings range from general or policy-type at the organi­ zational level to more specific at the procedural and technical level.

Prom the research the following conclusions may be drawn:

1. Transportation and landuse planning cannot be done in iso­ lation. In most cases they are only a part of the larger comprehensive planning process. Hence planning must be performed in a variety of environments, under various institutional personnel and other resource constraints. Technical procedures must, therefore, be selected on the basis of resources available and issues to be resolved.

2. Many study areas have evolved manual landuse forecasting methods. Some of these traditional manual methods have proved to be excellent, especially for use in medium and small-sized urban areas, although their use in large metropolitan areas is not precluded. It is unfortunate that many of these operational methods have not been publicized or found their way into planning literature. 3. It is possible to reduce time and cost in organizing and operationalizing manual landuse forecasting methods by making judi­ cious use of available data and standardizing procedures.

k. In the case of large metropolitan areas where large-scale computer-oriented landuse models are proposed to be used, the availa­ bility of a manual landuse forecasting method as a back-up procedure is highly recommended.

5. The development and ultimate use of a perfect method for landuse forecasting and distribution is a utopian dream. The nature of landuse models involves too many parameters which are not in the hands of planners or policy makers.

6. The critical examination and evaluation of landuse fore­ casting methodology was a difficult task chiefly because of poor documentation. In the first place, very few agencies performed accuracy checks to find out how well their landuse forecasting method­ ology performed. Of those few that did, none pointed out weaknesses or shortcomings in their modeling procedures. It was evident that most planning agencies were under heavy pressure to produce results as quickly as possible within the constraints of slender budgets.

7. No manual allocation procedure of landuse can be solely based on systematic weighting of zonal attractions. The planner has to be intimately familiar with the study area to the point where he is capable of dealing with subtle questions and issues connected with the region. Indeed, these are the issues that are presently difficult to express and resolve in terms meaningful to a computer-based model. 88-

NOTES

-kj. S. Department of Transportation, Federal Highway Admini­ stration, Urban Transportation Planning, Policy'and Procedure'Memo­ randum 50-9» Transmittal 162, Nov., 1962. O W. L. Grecco, et al., Transportation Planning for Small Urban Areas. NCHEP Report No. 16?. (Washington, D. C.: Transportation Re­ search Board, 1976) p. 2.

%ederal Highway Administration, Instructional -Memorandum 50-U-68: • Operations Plan' for Continuing Urban Transportation Plan­ ning. May 3. 1968.

^Data Inputs, Ohio Procedure Manual: Landuse (Cincinnati: Vogt-Ivers, 19&5) pp. 3.600-3.693.

^Ohio Department of Natural Resources, Land Capability Analy­ sis, County Report #3, June, 1975.

^Data Inputs, Ohio Procedure Manual; Landuse (Cincinnati: Vogt-Ivers, 1965) p. 3.701. APPENDIX A

SUMMARIES OF LANDUSE FORECASTING METHODS

Akron, QhicJ-

Introduction

The Akron Metropolitan Area Transportation Study (AMATS) pro­ vides a forum for decisions concerning transportation needs in Portage and Summit Counties. The forecasting of future landuse is done for a twenty- to thirty-year time frame.

The allocation of regional totals of population and economic activity is based on landuse development potential and regional and community landuse plans. The process involves, first, evaluating vacant land development potential, and then distributing the regional totals to broad areas called "Planning Analysis Zones" (PAZ's), and finally to traffic zones. The general procedure is shown in the flow chart in figure 1 9.

Methodology

Evaluation of vacant land development -potential

Vacant land which is physically deterred is eliminated from the allocation process. The remaining vacant areas are considered to have development potential. In order to measure the strength of development potential, the AMATS Activity Allocation Model was formu­ lated. The model works on a points system, assigning specific numerical scores that reflect an area's relative attractiveness, based on community facilities, various measures of accessibility, and other characteristics. Each traffic zone receives a score based upon its individual characteristics. These scores become part of the evaluation procedure used in distributing forecasted activity to smaller areas within the region. This model, as with all distributional techniques, depends on judgment and the understanding of the factors which promote development.

89 Physically Deterred Vacant Land Existing 1970 Vacant Land Most Existing Land Usable Attractive Regional and Landuse ; Vacant to Future Local Plan Inventory Land Development Frameworks Tabular Existing Development Allocation and Mapped Developed Factor of Future Output of Land Ratings Activity to Forecasted Usable Vacant Activity Land 1970-2000

Recent and Existing Additional Announced Landuse Acres Needed Development Activity for Future and Regional 1970 Ratios Activity Existing Plans Socio-Econ Forecasted Activity Socio-Econ Inventory Activity 1970-2000

FIGURE 19

GENERAL FLOW CHART OF ALLOCATION PROCESS, AKRON, OHIO

vo o The residential allocation •process 91 Residential allocation is concerned with the activity measures of population, housing units, residential acreage, and automobiles. The allocation process begins with the stepping down of regional totals to PAZ's. However, initially the control totals are distributed to eight county subareas. The distribution to these county subareas is based on the present proportion of existing county population adjusted to reflect recent growth trends.

These subarea forecast totals are then further stepped down to PAZ’s. The PAZ population forecasts are adjusted to reflect the segment of population living in group quarters such as dormitories, long-term health care facilities, and other institutions.

The average family size for each PAZ is divided into the population control to determine the number of housing units to be occupied. The average family size ratios have been based upon pro­ jections for the United States as a whole, adjusted to reflect both regional and local variation.

The vacancy rate is applied to the forecasted occupied housing unit number to derive total housing units to be allocated to traffic zones within PAZ’s. A five percent vacancy rate was assumed.

Finally, the automobile ownership rate is applied to forecasted occupied housing units to determine the number of available autos located in various areas around the region. The auto ownership rate (measured as autos per household) is based on the AMATS Automobile Forecast.

The flow chart in figure 20 illustrates how PAZ forecasts of population, housing units, and automobiles are derived.

Distribution of residential activity to traffic zones

The PAZ total housing unit control is the statistic used for distributing forecasted residential activity to traffic zones. All other necessary measures of residential activity changes (population, occupied housing units, residential acres, and automobiles) are derived from the total housing unit allocation for each traffic zone.

The distribution of the PAZ total housing unit control into the appropriate traffic zones is made through the evaluation of residential traffic zone ratings, available acreage, announced development plans, and plan density frameworks. All of these factors are used as inputs for a computer program which mechanically allocates forecasted controls to traffic zones. The program also calculates the corresponding popu­ lation, acreage, occupied housing, and automobile statistics. 92

Regional. Population Forecast

PAZ Forecast Controls

Deduct PAZ Group , Quarters

Total Population "In Housing"

Apply PAZ Family Size

PAZ Autos, per Occupied PAZ Occupied Housing Housing Unit Unit Control

Apply PAZ Vacancy Rate

PAZ Total Housing Unit Control

PAZ Total Auto Control

FIGURE 20

POPULATION, HOUSING AND AUTO DISTRIBUTION, AKRON, OHIO 93

Figure 21 shows generally the various steps used in the resi­ dential allocation to traffic zones. Each step on the chart is detailed below.

Planned Residential Frameworks Traffic Zone Ratings III

Identify and Measure The Initial Available Residential Traffic Zone Acres Allocation ______II______

\ 1 Residential Allocation Program

Total Housing Units Allocated to Traffic Zones VI

FIGURE 21

DISTRIBUTION OF RESIDENTIAL ACTIVITY TO TRAFFIC ZONES, AKRON, OHIO

Planned frameworks. The housing units per acre density is derived from the Tri-County Regional Planning Commission Regional Development Plan and individual community plans for communities that had them when the allocation was completed.

The identification and measurement of residential land. Utilizing the plan frameworks as an indicator of density and use types, all vacant acreage identified as residential land is measured and tabulated for each traffic zone.

Residential traffic zone ratings. The factor ratings for all traffic zones with vacant residential acreage are listed from the lowest to highest score. 9U

The initial traffic zone allocation. The first phase in the allocation process distributes a proportion of the PAZ control based on the traffic zone residential factor ratings. The sum of the ratings, for selected traffic zones with available residential acreage, is divided into each score to derive the proportion of the control to be allocated to each zone. This numerical process is then modified with profes­ sional Judgment based upon recent trends, announced development plans, local policies, and other factors in an attempt to determine the most likely sites for future residential development.

Residential allocation program. A computerized process for allocating the traffic zone housing units uses the proportions derived from'the factor ratings. The program identifies the number of dwelling units desired for each traffic zone and applies the appropriate density value for the zone which identifies the amount of acreage needed to accom­ modate the additional 1980, 199 0, or 2000 dwelling unit increment. The program allocates to the lowest-rated zone first. This low-to-high allocating procedure is necessary to prevent an oversaturation of dwelling units in a zone. If acreage is not sufficient, the program automatically allocates the remaining units to the next highest rated zone.

Total housing units allocated to traffic zones. The residential allo­ cation program tabulates the increase in land utilization for each traffic zone that has received additional dwelling units for each decade, 1970-2000. These increases include:

1. Acreage utilized for new dwelling units. 2. Acreage available for future residentialuse. 3. Acreage for new streets to serve newdwelling units.

The allocation program also tabulates increases in socio-economic characteristics related to the dwelling unit allocation at the traf­ fic zone level. The socio-economic increases include:

1. Additional population growth. 2. Additional automobile ownership.

The non-residential allocation -process

The non-residential allocation is concerned with employment, acreage, and floor space for commercial, wholesale, industrial, and public land uses. The AMATS Economic Forecast report sets forth the necessary regional, employment controls used in the allocation. Employment is the primary variable used in distributing non-residential activity to both PAZ's and traffic zones. Acreage and floor space needed to accommodate employment increases are based on existing ratios of employees per acre and square feet per employee. These ratios are differentiated according to the type of activity and also according to the location in question (i.e., urban, suburban, or rural). 95 The formulation of PAZ commercial, wholesale, and industrial employment controls

The formulation of PAZ employment controls from forecasted regional non-residential employment figures is made through a series of formulas which consider traffic zone factor ratings, announced plans, projected population growth trends, and regional or community plan frameworks. These formulas result in a PAZ employment growth in each landuse category. The formulas used are as follows:

COMMERCIAL: WHOLESALE:

PAZ A + D + E + F PAZ A + B + C COMMERCIAL WHOLESALE ALLOCATION ALLOCATION INDEX INDEX

INDUSTRIAL:

PAZ D + G + H _ INDUSTRIAL 3 ALLOCATION INDEX

PAZ employment allocation index 1970-2000, where:

A = PAZ commercial factor rating divided hy the total of all PAZ commercial factor ratings

B = PAZ population increment, by decade, divided by the regional decade population increment

C = Announced commercial square footage increases in each PAZ divided by regional commercial square footage increase totals

D = The PAZ industrial factor rating divided by the total of all PAZ industrial factor ratings E = PAZ announced plan wholesale employment increases divided by the regional announced wholesale employment F = Existing PAZ wholesale employment total divided by the 1970 existing regional wholesale employment

G = The square footage of new industrial construction at the PAZ level divided by the total increase in industrial square footage 96'

H = Available PAZ industrial acreage divided by regional industrial acres.

Distribution of commercial, wholesale, and industrial employment to traffic zones The commercial., wholesale, and industrial employment allocation index is computed for all PAZ’s and is applied to the regional employment total. PAZ PAZ EMPLOYMENT EMPLOYMENT = ALLOCATION X REGIONAL EMPLOYMENT CONTROL INDEX TOTAL

The traffic zone allocation of PAZ commercial, wholesale, and indus­ trial employment controls utilizes the following procedure:

1. Allocate announced non-residential plan development employ­ ment figures to their respective traffic zones.

2. Allocate remaining employment increases to existing and potential non-residential land areas by:

a. Utilizing plan frameworks to identify traffic zones with commercial, wholesale, or industrial growth potential.

b. Using the traffic zone ratings, determine the relative attractiveness of the selected traffic zones with com­ mercial, wholesale or industrial potential. Modify with professional judgment based on recent trends, announced development, local policies, and other fac­ tors. Final result of traffic zone share is quoted as a percentage of total forecasted development for the specific land use. c. Multiply the PAZ employment control by each percentage to determine employee levels to be allocated.

d. Apply density ratios (employees/ acre, square feet/ employee) to derive the number of acres and square footage needed to accommodate the employment increases. 97

The formulation of PAZ public employment controls

The distribution of public employment to PAZ’s is made through:

1. The application of 1970 PAZ government, education, and institutional employment proportions to 1970-2000 employment forecast in each category.

2. The adjustment of forecasted proportions utilizing employment levels of announced development. The process also involves the further division of these PAZ category controls into centralized and non-centralized public employment facilities.

Centralized public facilities

Centralized public facilities are located in the more densely populated urban centers of Summit and Portage Counties. Included in this category are federal and state government buildings, hospitals, universities, and sanitariums.

The calculation of PAZ employment controls for these centralized facilities assumes the 1970 centralized vs. non-centralized public facility ratio constant for 1980, 1990, and 2000. The PAZ centralized employment controls derived from this ratio are allocated at the traf­ fic zone level to existing facilities utilizing 1970 employment pro­ portions and announced plan employment information.

Non-centralized public facilities

Non-centralized public facilities are located in every community in the two-county area. These facilities include local government administrative and service agencies, and schools.

Increases in non-central public enployment facilities are directly related to PAZ population increases. A 1970 employment allocation factor was generated for each PAZ. This factor identifies the number of additional local government and school employees needed to serve increasing population increases in each community or PAZ. LOCAL GOVERNMENT OR SCHOOL EMPLOYMENT/ 1970 EMPLOYMENT PAZ POP. 1970...... ALLOCATION FACTOR = REGIONAL GOVERNMENT OR SCHOOL EMPLOYMENT/ REGIONAL POP. 1970 •98

The 1970 Employment Allocation Factor is applied to the regional increases of local government and school employment levels for 19 8 0, 1990, and 2000. These employment figures for each PAZ are adjusted to include announced non-centralized announced facility plans.

Distribution of public employment to traffic zones

The distribution of PAZ public employment controls to traffic zones uses the same procedure as that of the commercial, wholesale, and industrial allocation.

4 99

2 Austin., Texas

Introduction

The methodology used in developing a landuse plan is simple. An analysis of growth and landuse in Austin was done from 1927 onwards and a detailed analysis of growth was examined for the 1955- 70 period. Eighteen subareas of the city are analyzed in terms of 1970 landuse, developments between 1970 and 197*+ and factors which will affect future development in each area. Individual landuse maps, data and commentary are provided for each subarea.

Projections are based entirely upon the continuation of present trends. These projections also assume that there will be no significant changes in municipal development policy. The process of developing the landuse plan is shown in figure 2 2.

Region Divided into 18 Subareas

Detailed Maps Prepared for Each Subarea (1970) Showing Categories of Landuse

Trends Population, Housing and Economic Forecasts for the Region for Land Absorption 1980 and 1990 Coefficients

Allocation to Subareas and Census Tracts

FIGURE 22

LANDUSE PLAN DEVELOPMENT PROCESS, AUSTIN, TEXAS 100

Methodology

Existing landuse for 1970 is presented in a series of maps depicting seven categories of landuse. Each map consists of a group of planning areas identified by a six-digit number. The first four digits refer to 1970 Travis County census tracts and the last two digits refer to planning areas within census tracts. In addition to the symbolic landuse presented on the maps, tabular data for each landuse, in acres, is presented for each planning area. A description of the 1970 landuse, development between 1970 and 197^» and factors which might affect future development are presented with each map.

Landuse data was interpreted from April, 1970 aerial photo­ graphs, from which the area of each type of landuse in each planning area was calculated. In addition, the total area of each planning area was measured and the sum of all. developed uses was subtracted from the total area to derive acreage of undeveloped land.

The following landuse categories are considered:

1. Low Density Residential 2. High Density Residential 3. Commercial U. Industrial 5. Parks and Recreation 6. Public and Quasi-Public 7. Undeveloped 8 . Streets Parks and recreational areas are included under Public and Quasi- Public. Only public streets are included in the eighth category.

Landuse -projections

Landuse projections are based on the 1970 inventory of landuse and projections of employment and housing units developed separately.

Projected changes in employment and housing units are translated into acres through the use of land absorption coefficients. These coefficients are determined locally and simply indicate the number of acres in use per employee or per housing unit in 1970. Coefficients are determined for low and high density residential, commercial and industrial uses. These coefficients have been revised to reflect chan­ ging densities since 1970 in the high density residential and commer­ cial categories, as follows: 101

Use Units/Acre Low Density Residential 3.88 High Density Residential 23.26 Commercial 1^.08 Industrial lU.08

The unit for the residential, uses is a housing unit. For the commer­ cial and industrial uses, the unit is an employee. The commercial and industrial coefficients were calculated independently and hy chance had the same value. The definitions of the above categories are:

Low Density Residential— single family residences, duplexes, mobile homes, farms (residential area only).

High Density Residential— apartments, private dormitories.

Commercial— retail, services, offices.

Industrial and Related— manufacturing, wholesale, private transportation terminals, construction machinery storage yards, wrecking yards, quarries, feed lots, etc.

Public and Quasi-Public— local, state or federal government, (including parks, power plants, schools, public office buildings, utilities, etc.) private recreational clubs, golf courses, churches, hospitals, cemeteries, etc.

Landuse distribution

Acreage projections were distributed geographically once aggre­ gate projections for each landuse category had been made. The distribution of projected landuse changes was conducted by the Planning Department staff after a review of existing development plans of private developers and planned infrastructure construction including utility lines, expressways, major arterials and school sites. No major changes in municipal policy were anticipated. The projections presented here, therefore, represent the probable results of the continuation of existing trends. Table 10 presents data for six categories of landuse for 1970, with projections to 1980 and 1990.

A substantial amount of the 9,^08 acres of low and high density residential growth projected by 1980 will occur contiguous to existing residential development. The pattern of dispersed residential development in certain parts of the area is projected to continue. These scattered developments rely upon septic tanks and small-scale treatment plants for waste disposal where municipal systems are not 1-02 available. As a result, development can occur at considerable dis­ tances from the existing and projected municipal sewer service area. These developments are dependent, however, upon the arterial system for access and will locate near major highways and smaller ranch and farm roads which service these major highways. The development of farm and other acreage around these dispersed residential subdivisions should proceed as the municipal sewer system is extended.

TABLE 10

1970 TRAVIS COUNTY, TEXAS, LANDUSE WITH PROJECTIONS TO 1980 AND 1990 (IN ACRES)

Landuse 1970 1980 Percent 1990 Percent Percent Change Change Change 70-80 80-90 70-90

Low density residential 20,081 28,235 +kl 38,k80 +36 +92 High density residential 1,195 2,kk9 +105 3,206 +31 +168

Commercial 3,070 k,k97 +k7 5,709 +27 +86

Industrial 2 ,11k 3,205 +52 k,357 +36 +106 Public and Quasi-Public 20,268 23,120 +lk 2k,791 +7 +22

Undeveloped 6 17,02k 600,203 -3 582,626 -3 - 6

High density residential development is expected to increase by 1 ,25^ acres, or 105 percent by 1980 and by an additional 757 acres or 168 percent by 1990. Most of the growth in this category should occur in new multi-use complexes developing in the suburbs. Con­ version from undeveloped and low density residential uses to high density residential uses has been commonplace in the last decade. This conversion has occurred mainly near the University and in older neighborhoods. Recent trends indicate that this conversion process is being replaced by apartment locations in more peripheral, undeveloped locations.

Increasing costs of gasoline and liquefied petroleum gas would seemingly act as restraints on residential locations distant from employment centers and natural gas service areas. But such increasing costs are apparently offset by other factors which encourage decen­ tralization. These factors include increasing property tax burdens in 103 the city of Austin and the Austin Independent School District (AISD), fear of social conflict arising from racial integration in the AISD, and an increasing valuation by many people of the natural amenities available in the rural areas surrounding Austin. The slow but steady growth of high-income second-home residences on Lake Travis is expected to continue. An additional. 1,U27 acres of commercial development, or an increase of ^7 percent is expected by. 1980. A slightly smaller increase of 1,212 acres is expected between 1980 and 1990. These projections are based upon the expected continuation of current trends and indicate that the amount of commercial development in Travis County will nearly double between 1970 and 1990.

Most of the projected commercial development will occur on undeveloped land. Some conversion from low density residential to office or light commercial uses is expected to continue in the older residential, areas of the city, especially as these areas deteriorate.

An additional 1,091 acres of industrial development are expected to appear on circumferential highways north, south and southeast of the city by 1980 and an additional. 1 ,1 5 2 acres are expected in these areas by 1990. Most of the new industrial land uses should locate near the existing industrial district along F.M. 1325 and in the industrial district along Ed Bluestein Boulevard, along Ben White Boulevard and in a few isolated locations to the south and southwest of the city.

Nearly all of the industrial plants locating in or near Austin during the past ten years have resembled extensive one- or two-story office facilities. This trend is expected to continue.

Public and quasi-public uses are expected to increase relatively little between 1970 and 1990. In 1970 there was more land used for public and quasi-public purposes in Travis County than for low density residences. This relatively high proportion of public land is due to the fact that Austin is the state capital and the center of many state facilities and programs. Future expansion of state facilities should occur on extensive, undeveloped tracts already owned by the state. Most of the ^,523 acres of additional public and quasi-public land expected by 1990 will occur in the form of parks and schools. Governmental regulations, especially those of the City of Austin, are designed to control the quality and location of landuses. These controls are moderately successful to varying degrees in achieving their goals and include such tools as the comprehensive plan, zoning and subdivision ordinances, the capital improvements program and annexation. Other controls such as natural and environmental features, financial considerations and societal life styles play a significant role as well. 10U

Private development decisions are the major factors that deter­ mine general landuse patterns. All of the controls, incentives and constraints, whether governmental or natural, interact with the private development decisions of business, industry and the individual property owner. The coordination of public and private decisions will allow the city to provide municipal services efficiently and to aid in planning for the optimal use of air, land, and water resources. 105

Canton. Ohio3

Introduction The Stark County Area Transportation Study (SCATS) is an arm of the Stark County Regional Planning Commission located in Canton, Ohio. It is now in the continuing phase of its transportation plan­ ning process. This ongoing process involves re-evaluating previous forecasts. The method of allocating regional totals to small areas like planning areas and traffic zones was done by traditional manual methods. The use of a land capability analysis technique developed by the Ohio Department of Natural Resources was relied on quite extensively while working on landuse potential, capacity and develop­ ment.

Methodology

Framework

Economic and population forecasts for the years 1970-2000 with interim forecasts of 1975, 19 8 0, 1985 and 1990 were prepared on a county-wide basis. County-wide control totals are distributed to traffic zones. This is done for population, dwelling units, income, automobile registration, labor force, employment, and landuses. A general flow chart of this distribution is shown in figure 23.

The basic forecast unit is the traffic zone. There are 6ok of these detailed units within the county. An intermediate forecasting unit is the Traffic Zone District. These are aggregated traffic zones and are combinations of census tracts. Traffic zone districts are used as intermediate steps in forecasting from county-wide control totals to traffic zones. A great amount of emphasis is placed on the dwelling unit and population forecasts, because it is from these two variables that many of the other forecasts are derived. The forecants of population for traffic zone districts are developed ■using political unit forecasts. Base population data is obtained for each traffic zone district from 1970 census information, from which a persons per household rate is derived for each traffic zone district. A vacancy rate and the number of persons in group quarters is also obtained. A population per household trend is developed for each traffic zone district. Dwelling units are allocated among the political units and traffic zone districts after setting the control totals on the basis of population. Population Employment Countywide Countyvide

Population Employment Planning District Traffic Zone Pi strict

Population Population Political Unit Traffic Zone District Employment Employment Traffic Zone Political Unit Dwelling Unit Dwelling Unit Political Unit Traffic Zone Landuse District__ Landuse Floor Space Floor Space Traffic Zone Political Unit Dwelling Unit Traffic Zone

Population, Auto, Labor Force School Enrollment, Traffic Zone

Auto, Labor Force, School Enrollment, Political Unit

FIGURE 23 H.o o\ GENERAL FLOW CHART OF POPULATION AND EMPLOYMENT DISTRIBUTION, CANTON, OHIO 107

The traffic zone dwelling unit forecasts are worked out as follows:

1. 1970-1975 traffic zone dwelling unit increase (or change) is used as a dwelling unit base for each individual zone, using building permits on a traffic zone basis.

2. Factor score ratings of each zone are developed and used.

3. The amount of available land for development as well as areas of known and proposed multifamily development are considered.

U. Future highway construction and urban renewal are considered.

5. Commercial and industrial landuse (i.e., such as industrial parks) is considered. The factor score ratings under Step 2 are analyzed for each traffic zone for the following items:

1. Central water service 2. Central sewer service 3. Proximity to schools b. Distance a. downtown b. to major’shopping centers or malls c. to employment centers 5. Highway systems (only traffic zones within one mile of a freeway interchange) 6. Availability of mass transit service 7. Existing landuse 8. Population change 1960-1980 9. Proximity to major recreational areas 10. Existing population density

The higher the factor score rating for any zone, the greater the potential for development.

Traffic zone forecasts of population

Upon completion of the forecasts for population and dwelling units by traffic zone district and the forecasts for dwelling units by traffic zone, forecasts of population by traffic zone are undertaken. This is the final step, prior to accuracy checks. The accuracy of the projection depends on what preceeds, and the accuracy of further 108 forecast years depends upon the adjustments made to each ensuing forecast year. This process is shown in figure 2b.

Political Unit Population Forecasts

Political Unit Dwelling Unit Forecasts

Traffic Zone District Dwelling Unit _____ Forecasts_____

Traffic Zone Dwelling Unit Forecasts

Traffic Zone Population. Forecasts

FIGURE 2k

POPULATION DISTRIBUTION, CANTON, OHIO

Population forecasts are based on the forecasts of dwelling units by traffic zone, the base 1970 population per household by traffic zone, and population per dwelling unit by traffic zone.

Mean income

The average income per family is forecast for political units using the trends developed from OBERS projections of per capita income. Median income would have been the more desirable figure to obtain, but for revalidating the SCATS (19^5) models it was more convenient to calculate mean income. The 1970 mean income for the cities was available from the 1970 census. Township base figures were obtained 109 by averaging the mean income figures of the census tracts comprising the townships. Once 1970 base figures were determined, the trends observed using the OBERS projection were applied. All. forecasts are based on 1970 census data and are in 19&9 dollars. The mean income by traffic zone was obtained from 1970 FHWA package.

Autos

Census data provides a category on autos available by household. Using this data on a traffic zone basis for the 1970 base, forecasts of autos available were made by traffic zones. Autos registered are always greater than autos available. First, the control totals for Stark County were fixed and then forecasts were made on autos per dwelling unit basis.

Miscellaneous

Labor Force, School Enrollment and Employment, and Landuse Floor Areas are all based on regional accounts of dwelling units and pop­ ulation. 110 k Chicago, Illinois

Introduction

The Cooperative Landuse Transportation Project was a joint venture of the Northeastern Illinois Planning Commission (NIPC) and the Chicago Area Transportation Study (CATS). The project had "been agreed upon between the two agencies by early 1962, effectively set in motion during 196J+, ami brought to a hectic climax in 1966 and 1967. It culminated in the formal adoption of a plan in 1968.

CATS did not develop a sophisticated mathematical model for land-* use forecasting, but sought to identify the significant factors that influenced development. Although the final predictions were based on the collective judgment of the staff, certain devices were used to insure reasonably accurate forecasts. These devices were: sound estimates of total growth expected in the area; and accurate knowledge of the quantity and quality of available land in the study area that was available for development. Figure 25 is a simplified flow chart of the procedure used to estimate landuse. This procedure is often referred to in landuse literature as the density-saturation gradient method for landuse analysis and forecasting.

Methodology

General framework

To accomplish the study, CATS divided the Chicago Study Area into regions, rings, districts and zones, listed in descending order of importance, and used these to control.the consistency of the fore­ casts. The basic input for the estimates included the following:

1. Landuse data as of 1956. This data included population characteristics, employment and significant landuse features.

2. Urban redevelopment plans for residential areas and the Central Business District.

3. A population and economic forecast for 1980, which in­ cluded a breakdown of employment by type of industry.

Data concerning existing zoning ordinances and future community plans that would have an effect on vacant land development. The CATS study group believed that community activities com­ pete for available sites, and those that can afford to pay the highest AVAILABLE INFORMATION PROCEDURE 1980 ESTIMATE

1956 Landuse Data Unusable Unusable Land Vacant Land

Landuse Usable Redevelopment Plans Vacant Land

Used Land 1980 Population and \/ -±_ Economic Forecast Total 1980 Allocate New 7 Land or Additional Requirements Landuses Zoning and Community Plans Vacant Land

FIGURE 25

CATS LANDUSE ESTIMATING PROCEDURE, CHICAGO, ILLINOIS

H t-» H 112 prices for sites will do so. Accessibility increases the value of a site. They found that residences tended to develop on the periphery, and that established neighborhoods remained stable. Based on the stability displayed, the same population density and proportional, landuse was predicted in 1980. for. the study area segments, except those being extensively redeveloped, such as the CBD. Vacant land was analyzed and divided into usable and unusable categories, and the usable land was assigned to the 1980 requirements. Usage was based on past ratios obtained from historical data. A comparison was made between the net residential densities and population by distance from the Loop between 19^0 and 1956. The peak movement out from the CBD was kept at the same rate for 1956- 1980 as it was from 1920-19 5 6. Concentration of population and workers decreased with distance from the CBD. Population holding capacities for each zone were cal­ culated by multiplying the acreages of designated residential land by the estimated residential density for each zone and adding the results to the 1956 populations. All vacant usable land was developed on this basis for residential uses, except that set aside for other uses, primarily industrial and commercial. Industrial development was forecast using the input-output technique.

Some 1136 was made of statistical methods in analyzing historical data to establish trends for use by the study group in forecasting future development patterns.

Procedure The objective of making the landuse forecast was to provide a basis for estimating future travel. A set of observations of existing regularities are used in the forecasting procedure. 1. There is a decline in the intensity of land development in proportion to the distance (or access time) from the CBD.

2. There is a decline in the amount of land in use as a proportion of available land which is related also to distance from the CBD.

3. There is a stability in the proportional amounts of land devoted to each type of landuse.

The model is built around a strong system of landuse accounting for small territorial subdivisions of the study area. For each such 113 district in turn, the future inventory of landuses is extrapolated from the initial inventory according to rules (.modified by judgment) specific to the kind of use. Six landuses are recognized: residential, commercial, manufacturing, transportation, public buildings, public open space, and streets. Vacant land is classified as residential, commercial, or industrial, according to its status under local zoning ordinances. Unusable land is also accounted for.

The initial landuse pattern of each district is modified in six steps:

1. Specific parcels of land in some districts are designated for conversion to public open space and transportation uses (e.g., a new airport). The designations are based primarily on existing plans of public agencies for such development.

2. Commercially zoned vacant land in some districts is desig­ nated for shopping centers and heavy commercial uses. These designations are based on announced private plans and staff judgments.

3. Residentially zoned vacant land is designated for residential use. The amount so designated in each district depends on the location of the district and it3 residential holding capacity at existing or slightly modified net densities. The percentage of a district's holding capacity to be filled by 1980 is defined as a function of distance from the Cen­ tral Business District, with sectoral and local modifica­ tions based on staff judgments. 4. For residentially oriented uses, per capita norms are applied to the estimated I98O population of each district as determined in the third step. Thus space for streets, local commercial facilities, public buildings, and recrea­ tion is set aside in each district.

>. Industrially zoned vacant land is designated for manufac­ turing use. The amount so designated in each district depends on the location of the district and its manufacturing holding capacity. Trends in net employment density in manufacturing establishments, both over time and by distance from the CBD, serve as the basis for 1900 forecasts of such employment density for each district; this projected density, in conjunction with "Che amount of industrially zoned space, determines the district's holding capacity. The percentage of this capacity to be filled by I98O is defined as a function of distance from the CBD, with sector and local modifications based on staff judgments. 11^

6. Since net activity density and acreage in each use have been explicitly predicted for each district, the implied popula­ tion and employment totals for the district can he calcu­ lated. These are summarized for the study area as a whole and compared to independent projections of the area's population and employment. The landuse. forecase (acreage occupied) is then systematically modified so as ter reconcile the implied activity totals with the independent projections.

The inventory of landuses is projected for each district separ­ ately; the forecast is based on that district's initial inventory, its zoning map, and its location. The resulting landuse is indeed modified so that it adds to a control total.

Sample Problem

A sample problem showing the use of the Density-Saturation Gradient Method will help to. illustrate the. procedure used. This illustration is in many respects a gross oversimplification of the actual procedures followed in a "real world" situation.

This sample problem will be limited to the analysis of residential landuse only. Residential land accounts for the greatest portion of all urban land in use and residential patterns are typically most regular in nature and thereby most suited to analysis by the Density- Saturation Gradient procedure. However, this method is not limited to the analysis of residential land only. This procedure is concerned with the search for underlying regu­ larities in the pattern of landuse in the urban area and in developing an understanding of the factors, and the extent of their influence, which appear to significantly affect the direction and intensity of growth.

Figure 26 illustrates a hypothetical study area subdivided into a number of traffic analysis zones. Superimposed upon this are the familiar ring-sector boundaries which, in this case, divide the total study area into four quadrants and three concentric circles emanating from the approximate center of the area. The simultaneous consideration of the sectors and rings structures the study area into the basic unit of analysis for the Density-Saturation Gradient method, the analysis district.

Figure 27 is a magnification of the structure of a typical analysis district. The particular district illustrated is bounded by the. 3 1/2 and U 1 /2 mile radii, and by the sector boundary lines which determine the north-east analysis quadrant. Most landuse information is collected and recorded to the traffic analysis zone rather than 115

Sector 2 Sector 3 3 Ring

Sector 1 2 Ring Sector U

1 Ring-

FIGURE 26

GENERAL CITY STRUCTURE SHOWING RINGS AND SECTORS

20 Zone 29

Sector Line /

FIGURE 27

ZONE STRUCTURE OF A TRAFFIC ANALYSIS DISTRICT 116 to the analysis district. For this reason it is most convenient to approximate the analysis district- as being composed of a number of whole zones. In- the sample problem.we have arbitrarily decided that those zones whose centroids fall-within the boundaries of the district will be considered as part of the analysis district. In figure 27 we see that the-analysis district will actually be composed of analysis zones 20, 2b, and 29. Zones 30,. 31, and 3V are not part of the dis­ trict as their centroids fall beyond the boundaries of the theoretical district. The analysis of the existing pattern of residential development will depend predominantly upon the type of. data illustrated in table 11. The calculations shown in table 11 are for a particular analysis district. All districts would be included in a full-scale residential landuse analysis. Column 2 contains the gross area,, in acres, for each analysis zone in the district. The total resident population at the time of analysis is shown in column 3. Column b contains the area of each zone which is currently in residential use. Column 5 shows the maximum amount of additional land in each zone which could ever conceivably be put into residential use. The determination of the values in column 5 are by no means straightforward or obvious. From the total amount of land available for development in each zone, must be subtracted those proportions which are or are likely to be reserved for other than residential use. This would include areas devoted to industrial parks, land planned for public uses,, such, as park land, streets and highways, public buildings, and other areas, which by virtue of landuse controls, are restricted to other than residential development. In addition to considerations of this nature, the values in column 5 must be sensitive to accepted planning policy decisions. It is at this point that the analyst is called upon to translate the decisions reached with regard to the desired patterns of future resi­ dential location into land availability values. If the accepted residential plan restricts growth in a given area, then this should be reflected in the amount of additional land available for residential development for the appropriate zone.

The initial step in the Density-Saturation Gradient Analysis method is to plot the curves of residential density versus distance from the CBD. Curves can be drawn for the total study area and for individual sectors. The values used in the plotting are net resi­ dential density for each analysis district plotted against the distance of the district from the CBD. Equation 1 in table 11 shows the sample calculation of net district residential density as the quotient obtained by dividing total land in current residential use into total existing resident population. For this problem the residential density of the sample district is 21.7 persons per net residential acre. The plots of residential density can then be studied for underlying 117 regularities and for variances among sectors. It is through this kind of analysis that -we would hope to gain greater understanding of the existing structure of the study area. The calculation of residential capacity is shown in equation. 2 of table. 11.. The holding capacity of a district is equal, to the current resident population plus the product of expected future residential density and the maximum amount of land available for residential use. The resulting figure then is an approximation, of the population of the district under the assumption that all the land available for residential development is consumed.

TABLE 11

ANALYSIS OF EXISTING LANDUSE PATTERNS USING DENSITY-SATURATION GRADIENT METHOD

(1) (2) (3) (U) * • (5) Total Existing Existing Land Maximum Addt'l. Zone Area Resident in Resi­ Amt. of Land (Acres) Population dential Use Available for Residential Use

20 287 1768 76 35

2k 296 2008 103 Uo

29 lM* 1230 52 25 Totals (District) 727 5006 231 100

1. Average District Residential Density _ 2 Col. 3 5006 21.7 (in persons/net residential acre) '£Col. k 231

2. District Residential Capacity = 'SCol. 3 + DEN x '21 Col. 5 = 5006 + 21.7 x 100 = 7176 persons— holding capacity

3. % Residential Capacity (District) _ Z'Col.'3 5006 69 8# Capacity “ 7176" ~ 118

A value of 7,176 persons was calculated as the holding capacity of the sample district under the assumption that all future residential development took place at the same density as currently exists, specifically 21.7 persons per acre. Exactly what value to use for residential density in computing capacity, must result from the analysis of existing density patterns, the judgment of the analyst, and zoning policy.

. The final calculation shown in table 11 is the computation of percent saturation or capacity currently existing in the sample district. It is determined, as shown in equation 3, by dividing resi­ dential capacity (calculated in equation. 2) into total district resi­ dent population (total of column 3). The percent capacity calculated for the sample problem is 6 9.8$. The individual district percent capacity values are then plotted against distance from CBD. The lower plot in figure 33 presents the typical results one might expect for such a plot. The point circled on this curve represents our sample district, falling approximately U miles from the center of analysis and having a percent capacity of 69.8%. Plots similar to that in figure 33 can be prepared on a studywide basis and on a sector basis. The analysis of these curves should provide a great deal of insight into the residential structure of the area and into the differences existing in the historic growth patterns between sectors.

In this sample problem the objective of the forecasting process will be to determine the resident population of each analysis district in the forecast year. One may go a couple of steps further and distribute this population to the individual analysis zones comprising each district and then to convert these resident population estimates into land consumption.

We have previously prepared plots of percent residential capa­ city by sector for the existing period. The next step is to forecast what these curves will look like in the forecast year. The upper curve in figure 28 represents this estimation for the particular sector under analysis. The development of the forecast percent capacity curves is the most significant step in the forecasting pro­ cedure. There are no simple rules which can be followed in its deri­ vation. To be sure a single precise procedure will probably never be found to apply to all areas. It must be resultant from: the analysis of the existing urban structure, the analysis of all trend data which exist, the judgment and analytic findings of the analyst, and finally the analysis of planning goals and policy decisions which are likely to affect future residential location. However, experience has shown that the general shape of the forecast curve remains about the same as the existing curve except for a rotation upward in the outer areas. ■forecast

o Existing

♦ H

•H

Miles From Center of Study Area

FIGURE 28

THE FORECAST OF DISTRICT POPULATION USING DENSITY-SATURATION GRADIENT METHOD

An interesting feature of the curves in figure '28 is that for the areas within approximately 1 1/2 miles of the CBD the forecast curve falls below the existing curve, indicating a decline in popula­ tion. This phenomenon has been observed in a number of cities. The general overcrowding- in the central areas often leads to the gradual flight of resident population to suburban areas.

Once this curve has been agreed upon, the determination of forecast resident population by district is a simple matter. The calculation shown in figure 29 consists of multiplying the residential capacity of the district by the forecast percent capacity (determined from the upper curve in figure 28-). Our sample district is forecast to exist at 85?» residential capacity in the forecast year, which means that it will have a resident population of 6,100 persons (.85 x 7176). 120

The final step in the forecast is to distribute the forecast residential growth for the district to the individual analysis zones. Table 12 shows one simple procedure for making this allocation. The incremental growth (forecast district, population minus existing resident population) is distributed.to the individual zones in pro­ portion to the amount of land available for residential development (column 2 divided by the sum of column 2). There are many ways in which this distribution can be. made. Suffice it to say it will demand the study and evaluation of the growth potential and density patterns of the individual zones.

TABLE 12

ALLOCATING DISTRICT POPULATION TO ZONES USING DENSITY-SATURATION GRADIENT METHOD

(1) (2) .(3) (U.) (5) (6) Additional Land Factor Incremental Exist- Forecast Zone Available for ''Col. 2 Growth. ing Pop. Residential Use TECol.2 in Zone Pop. t ...... Col. 5)

20 35 .35 .35 (6100-5006) = 383 1768 2151

2k 1*0 • Uo M (6100-5006) = U3T 2008 2hh5

29 25 .25 .25 (6100-5006) = 2jk 1230 150U

Total 100 109U 5006 6100

This completes the total process of forecasting resident popula­ tion on a zone basis. It remains to convert these population figures into residential land consumption. This will not be attempted for the sample problem. However, it is clear that this would mean esti­ mating, among other things, density of future development.

■While this procedure is complete in itself, many studies may wish to further modify the results. The Density-Saturation Gradient procedure allows only for a cursory and limited consideration of policy and other planning decisions. The analyst is encouraged to expose the results of the forecast, which must be remembered to be essentially a trend forecast, to the further consideration of factors necessary for a realistic and comprehensive end result. 121

Throughout this entire procedure for forecasting future resi­ dential population no control or limitation has been imposed on the total population forecast. That is,, the total resident population in the forecast year is unbounded and depends only upon the future percent capacity curve, the estimated densities, and the amount of additional land available for development. In an actual land use forecast one would have a constraint upon the total resident popula­ tion, namely that indicated by the population forecast. It would therefore be necessary to apply the Density-Saturation Gradient method in a number of iterative steps until a balance is reached with the population forecast. 122

' Columbus, ' Ohio ^

Introduction . The Franklin County Regional Planning Commission (.RPC)-, fore­ runner of the Mid-Ohio Regional Planning Commission (MORPC), was responsible for comprehensive planning in Franklin County, in cooper­ ation with the state and local governments. As part of this planning effort county-wide forecasts of population, employment, personal income, dwelling units, auto-ownership, retail sales, and labor force were finalized for 1985. A landuse allocation model was developed by the study to allocate these county totals to small areas. The model, consisting of a series of regression equations, was a partial failure. The distribution of coun t y totals had to be eventually performed by manual methods dependent largely on the "holding capacity" concept. The outcome of this planning effort was the transportation and landuse plan for 1985 for the region. In 1972, the Columbus and Franklin County Transportation Plan was accepted by the Federal Highway Administration and the Ohio Department of Transportation. At that time concern was expressed that the planning period only extended to the year 1985 and that a twenty- year demand estimate was needed to be able to adequately plan for transportation improvements. In addition, major changes in the social and economic conditions of the study area indicated a re-evaluation of goals and objectives. Early in 1973, an extensive study of expan­ sion of the region was conducted in conjunction with the Bureau of the Census. The study indicated that eight peripheral townships would be urbanized by the year 2000 and future work programs of MORPC should consider this expansion. Thus, in 1973, a major review and update of the Transportation Plan to the year 2000 was begun.

Methodology Process

The work program for the development of the Year 2000 Plan includes the following major tasks:

1. Determine the needs and desires of the people of the region and formulate goals designed to satisfy those needs.

2. Develop year 2000 growth indicators and trip generation variables based on the projected growth pattern.

3. Develop preliminary year 2000 travel demand, identify transportation corridors, and stratify corridor travel demand. 123

U. Develop alternative transportation systems to accommodate travel demand.

5. Evaluate how alternative transportation systems interface with social, land use,.and environmental planning.

6. Synthesize social, landuse, environmental, and transportation plans into a reasonable comprehensive plan.

Task 1 has been completed. In 197*1 and the early, part, of 1975, an extensive goals-setting process was conducted to determine the desires of the people of the region. MORPC held a series of public meetings, utilized public media, and developed citizen committees and Commission task forces. A goals document was then prepared.

Forecasts for the various variables required for the transpor­ tation models were next taken up. Control totals for the following basic variables were established for distribution to small areas.

Population (POP) Total Commercial Employment (COME) Dwelling Unit (DU) Total Employment (TOTE) Median Household Income (MC) Commercial Retail Goods Floor Autos Garaged at Home (CARS) Area (sq. ft.) (CRGFA) Labor Force (LABF) Commercial Retail Service Floor Commercial Shopping Em­ Area (sq. ft.) (CRSFA) ployment (COMSH) Total Industrial Floor Area (sq. Non-Manufacturing Em­ ft.) (TOIFA) ployment (NMFGE) Total Commercial Floor Area (sq. Communication, Transpor­ ft.) (TOCFA) tation 8s Utilities Commercial Land Use Acres Employment (CTUE) (COMAC) Miscellaneous Commercial Communication, Transportation, Employment (Total Utilities Acres (CTUAC) Shopping) (COMIS)

A two-stage allocation process was used. One for the period 197*+ to 1985 considered the realities of the current and past planning efforts. The other, for the period 1985 to 2000, reflected regional growth through the newly developed goals and objectives. An outline of these allocation processes follows. 197*+-1985 development pattern

The 197*+ to 1985 development pattern was based on extending the present growth trend. This approach was selected based on the assumption that it would not be realistic to expect drastic change in growth for a short to mid-range period. Thus, the 197*1-1985 development pattern is based on existing information such as proposed 12U capital improvements for sewer and water supply systems and trans­ portation projects as well as other known land development projects. This approach is supported.by the vast amount of vacant land in existing growth areas plotted for residential development, industrial parks, shopping centers, or strip commercial development.

The population control total was based on the 19T5 migration statistics published by the U.S. Bureau of Census.

The dwelling unit control total was based- on projected popu­ lation and household size. Again, the most recent census data was used to update previous forecasts of household size. The dwelling unit forecast, was used as the basis for -forecasting control totals for automobiles, labor force and income. These control totals were determined by past trend relationships of each variable to dwelling units.

Total employment was determined by projected population and labor force participation rates. Total employment was divided into basic categories (commercial office, commercial shopping, industrial, miscellaneous and others) dependent upon past trends, proposed pro­ jects and forecasted economic activities.

1985-2000 development pattern The development pattern for the period between 1985 to 2000 was tailored to be responsive to guidelines contained in the report, Year 2000 Goals & Objectives, 1975. prepared by MORPC. This was accom­ plished in part by preparing a series of graphic overlays for varying development conditions in relation to the adopted goals and objec­ tives. A composite graphic of the set of overlays indicated those areas suitable for different types of development.

The overlay system of graphics addressed the following issues: 1. Preservation of natural areas, flood plains, and prime agricultural land.

2. Development of water supply and distribution systems.

3. Development of waste water collection and treatment systems.

h. Use of existing transportation facilities and increased transit usage.

5. Delivery of private and public human services and other urban- related services and facilities. 125

Based on the varying degrees of suitability for land development, an estimate of the amount of land.needed to accommodate growth was made.

Landuse allocation

Industrial’landuse. The allocation of industrial landuse. is asso­ ciated -with the distribution of three trip generation variables (TOIFA, NMFGE and TOTE) to traffic zones. Between. 197^ and Year 2000, U655 acres of land will be needed for industrial, development. Industrial land was allocated first because it requires more specialized conditions than other landuses.

Industrial growth was allocated using the following criteria and procedures:

1. Industrial land is expected to retain its present character in the central city area. These older industrial, areas are spawning grounds for new industries. As these new. industries improve themselves and need more room and better transportation, they are expected to move to larger sites in suburban locations.

2. Most industrial development will occur in suburban areas near major transportation facilities. Industrial locations near Interstate interchanges are preferred.

3. Based upon current trends, little industrial development is expected in the ex-urban parts of Franklin County and the townships in adjacent counties.

U. Industrial land development of a suburban area was further prioritized based on accessibility and type of industrial development expected. Industrial parks near major trans­ portation facilities were given the highest potential for development.

Residential landuse and associated variables. The allocation of residential landuse is associated with the distribution of the fol­ lowing five trip generation variables to traffic zones: POP, DU, INC, CARS, and LABF. After allocating industrial land, 230,608 additional dwellings were allocated between 197^ and Year 2000. Residential development was allocated to five separate areas (urban core, urban area-inner fringe, suburban, Franklin County-outer fringe, and townships in surrounding counties) for two types of dwellings (multi-family and single-family). 126

A. Multi-Family Development Multi-family dwellings were allocated in two stages— high-rise dwellings and attached low-rise dwellings. Many of the high-rise dwellings were allocated in coordination with the City of Colum­ bus. Most of the new.high-rise’apartments were included in the downtown area's anticipated renaissance. Other Units, including dwellings for the elderly, were located primarily along the major transportation facilities radiating from the regional center.

The remainder of the multi-family units included low-rise apart­ ments, condominiums, townhouses, duplexes, etc. The average density of these units is expected to be about 16 units per net acre. Most of these units (87.0$) were located in suburban areas with the remainder being located in the small settlements in adjacent townships (0.5$), in the part, of Franklin County outside the suburban area (3.0$), and in the residential areas near the regional center (German Village, Victorian Village and Italian Village, 9.5$).

The multi-family distribution process considered the following:

1. Sixty-two percent of the site will be used for structures with the remaining 38$ being used for streets, neighbor­ hood commercial centers, open space and other public and semi-public uses.

2. Areas designated by local jurisdictions for multi-family use will have top priority for development.

3. Additional land needed (not designated by local juris­ dictions) will be selected in conformance with the following ranking system:

First— residential and commercial land adjacent to freeways, shopping centers, and multi-family developments; Second— residential and commercial land adjacent to shopping centers, and multi-family developments; Third— residential and commercial land adjacent to freeways and multi-family developments; Fourth— residential, commercial land adjacent to freeways and shopping centers. Fifth— residential and commercial land adjacent to multi-family developments; Sixth-residential and commercial land adjacent to freeways; and Seventh— residential and commercial land adjacent to shopping centers. 127

B. Single-Family Development Most of the new single-family units are expected to he huilt in the suburban areas (80.Q?5), with fewer being built in the rural portions of Franklin County.(10.0%) and the townships in the adjacent counties (10.0%). Allocation of single-family units to suburban, areas was accom­ plished with considerations being given to the following:

1. Density at which the dwelling units are allocated, will be on 1970 existing and 1970 through 1976 proposed development densities. In this case an average between 1970 densities and 1970 through 1976 proposed development densities was assumed, with the further stipulation that in no case would densities exceed six units per net acre.

2. Single-family residences will be allocated to areas appro­ priately designated by local jurisdictions.

3. Sixty percent of the single-family sites will be used for dwellings with k0% being used for streets, neighborhood commercial facilities, open space and other public and semi-public uses.

h. Traffic zones with residential land available for develop­ ment will be ranked according to the following criteria: a. Sewer service: First— service available with reserve capacity; Second— service available without reserve capacity and in areas scheduled for service through capital improvements programs; Third— areas not now being served and not included in the capital improvements programs but included in long-range programs; and Fourth— areas not now served nor expected to be served in the foreseeable future. b. Accessibility to employment centers: First— areas within seven or eight minutes of expected 1985 employment centers; Second— areas between eight and sixteen minutes of expected 1985 employment centers; and Third— areas more than sixteen minutes from expected 1985 employment centers. 128

c. Proximity to existing urban services (schools, parks, libraries, etc.): First— within existing, urban areas; Second— within one mile (the approximate width of an average neighborhood) of existing urbanized areas; and Third— areas more than one mile from existing urbani­ zation.

In developing the ranking, each of the first priority conditions was given the value of one. Second priorities were given two, and so forth.' Therefore, traffic zones with the lowest cumulative values.had the highest ranking and those with the highest cumulative values had the lowest ranking.

Allocation of single-family units to the rural portions of Franklin County and the townships in the surrounding counties was based on:

1. The existing development patterns of trends.

2. Anticipated sanitarysewer service areas.

3. Development plans ofdevelopers.

C. Distribution of Labor Force, Income, and Automobile For transportation planning purposes, labor force is defined as employment by place of residence, income is defined as the median household income, and automobiles are listed as cars garaged at home. Thus, distribution of these variables will be based upon distribution of dwelling units. In existing developed areas, trends can be used in projecting future variables. In newly developing areas an adequate base does not exist. Therefore, variables were assigned to new areas at a rate equal to the average of the study area for the year (1985). The change in variables per dwelling between 1985 and 2000 was then proportionally dis­ tributed.

The average number of persons working per dwelling is ejected to increase through the year 2000. The averagenumber of workers perdwelling in the year 1985 is expected to be 1.03,and 1.07 in the year 2000.

The median household income is also projected to increase between 1985 and the year 2000. In 1985 the median household income is expected to be $16,571. In 2000 the. median household income is expected to be $2U,900. In both cases, the median income is expressed in 197^ dollars to compensate for inflationary trends. 129

Automobiles per dwelling unit is expected to increase from 1.19 autos per dwelling in 1970 to.1.35 automobiles in 2000.. This rate of increase is somewhat less thaii experienced in the past and is in response to recent changes in the economy and the world energy situation.

Commercial landuse and associated variables. The allocation of commercial landuse is associated with the distribution of the following ten trip generating variables to traffic zones: COMSH, CTUE, COMIS, COME, TOTE, CRGFA, CRSFA, TOCFA, COMAC, CTUAC.

Commercial shopping and service floor area control total was projected by assuming an employee per floor space ratio. In 196^ shopping and service activities averaged 725 square feet of floor area per employee. It was assumed that the square feet, per employee ratio would remain constant through the year 2000. After obtaining total floor area, it was then allocated to three types of shopping facilities (regional, community, or local).

A. Regional Commercial Regional shopping center sites were developed according to the following criteria:

1. Centers should be accessible (within 1.5 miles of a freeway- type facility).

2. Anticipated market areas should be sufficient to suppdrt a shopping center. Considerations include:

a. households in the service area; b. overlap with other service areas; and c. potential for full development of service areas in the future.

3. Known plans of developers and entrepreneurs must be con­ sidered.

k. Sites should be "preferred" as commercial centers by local jurisdictions. B. Community Commercial Community shopping centers were allocated on the following basis:

1. Community centers may be less accessible than regional centers but should be located on major arterial streets (usually at the juncture of two arterials). 130

2. Anticipated market areas should be sufficient to support the center. Considerations include: a. households in.the service areas; b. overlap with.other service areas of community and regional centers; and c. potential for-. full development of service areas in the future.

3. Known plans of developers and entrepreneurs must be considered.

U. Sites should be "preferred" as commercial centers by local jurisdictions.

C. Local Commercial Local shopping centers (being closely associated with residential areas) were distributed considering the following:

1. Traffic zones containing residential development are nor­ mally bounded (at. least partially) by arterial streets suitable for either neighborhood centers and/or linear commercial development.

2. Neighborhood shopping facilities may be linear in form or in planned centers (it is difficult to anticipate which form may be developed).

3. Local shopping facilities are most closely associated with residential development. Growth in these facilities, between now and the year 2000, was based upon a direct relationship between the dwelling units and commercial development.

1+. In rural Franklin County, and townships in adjacent counties, commercial development was allocated according to established patterns.

D. Commercial Employment Distribution Commercial and service employment were distributed based on pre­ viously established employment floor area ratios.

Office employment,'floor space and'acreage. The initial office development projection was based on ratios of square feet of floor space per employee. 131 fllT nprt-hjoa of office space. The majority, of the new offices were located in the regional center (.60$. of the. added floor space) with the remainder being located in.suburban areas, in office parks and in close proximity to shopping centers.

The regional center office projections were coordinated, with, the City of Columbus, Department of Development. Suburban office development is based on identified office parks and a telephone survey to identify levels- of activity. Office parks offer-many of the. same advantages that were found in industrial parks and, thus, are expected to be heavily favored for new office locations (37$).

CTUE employment distribution. Communications, transportation and utilities employment (CTUE) was allocated in two phases. The first phase involved allocation to known or anticipated sites such as sewer plants, airports, truck terminals,. the COTA maintenance facility, the Ohio Center, etc. The second phase considered all other projected locations. In the latter case, the increase in CTUE employment was proportionally allocated to traffic zones containing existing CTUE employment.

Miscellaneous employment distribution. Miscellaneous employment in­ cludes a number of employment categories. Where possible, miscel­ laneous employment was allocated to known or anticipated locations, i.e., hospitals, universities, institutions, schools, etc. The remainder of the miscellaneous category, which includes persons working in private households, personal services, entertainment, mining, agriculture, were distributed in direct proportion to additional residential dwellings. 132

Kansas City, Missouri^

Introduction

Metroplan, the predecessor to the Mid-America Regional Council (MARC), prepared a regional plan in 1969- Since that time, the con­ struction of several major land developments and a change in the expected socio-economic characteristics of the area have necessitated the complete re-evaluation of transportation planning and future landuse needs.

Since transportation facilities are one of the major influences determining regional growth, the MARC staff was directed to prepare alternative landuse plans. These alternative plans are "based on the various landuse configurations resulting from population and employment distributions as influenced by various transportation systems.

After extensive analysis, the population control figures of 1,500,000 people by the year 1980 and 2,000,000 by the year 2000 were agreed upon by the constituent local governments.

A flow chart showing the overall procedure of distributing population and employment region control totals to zones is given in figure 29.

Methodology

General

Socio-economic distributions are based on two distinct landuse alternatives for the year 2000, Plan A and Plan B. Plan A assumes that the region will continue to grow following the existing trends, utilizing the highway and interstate facilities as the major trans­ portation system. Plan B, on the other hand, assumes that a sophisti­ cated rapid transit system will be initiated in the 1980's in addi­ tion to the conventional highway system.

To facilitate the distribution of projected population and employment for the years 1980 and 2000, the eight-county study area is divided into 8UU zones. One of the basic tools used is the Land Factors Atlas, consisting of base maps of the streets and existing and proposed transport system and a set of thirteen transparent overlays. These overlays illustrate existing landuse, community facilities, zoning, conditions of structures, proposed landuse, existing water service areas and sewer service areas, slopes and flood plains, air­ ports, city limit boundaries, origin-destination zone boundaries, census tract and block boundaries and soil types. 133 Regional Totals' POP, EGON 1970-2000.

/: V. Land Factor Atlas Vf Landuse Alternatives Plans A and B

\/ DU Aneilysis (195C>-70) S'. r vl/ Population Distribution 1980 & 2000 Analytical Procedures: DU Forecast 1. The 100% Zone Analysis 2. The D.I.S. Zone Analysis 3. Holding Capacity of Zones

Population Distribution to Zones

\s E Distribution of M Agricultural P Mining L Government 0 Retail & Service ‘ Y Industry M Unclassified E N' T Distribution of Employment to Zones

Employment Density Check

FIGURE 29

GENERAL FLOW CHART OF-LANDUSE DISTRIBUTION PROCEDURE, KANSAS CITY, MISSOURI .131*

A dwelling unit analysis for the years 1950-1970 establishes trends for projecting the housing density types required by the future population in the years 1980 and 2000. Standards of low density at 1-6 units/acre, medium density at 7-2^ units/acre, and high density housing at 25+ units/acre are used for determining how much develop­ able land each density type would demand.

Analytical procedures

Two major analytical procedures are used for the purpose of distributing population and employment for the years 1980 and 2000: the 100$ Zone Analysis, and the Determinate Index System (DIS) Analysis. In addition, holding capacities of zones are also estab­ lished.

The 100$ zone analysis

The 100$ Zones are defined as those zones in which 10$ or less of the total area is available for development in 1970. If more than 10$ is undeveloped, the zone is handled by the DIS Analysis. These 100$ zones represent the inner areas of Kansas City. The basic task is to determine whether these zones will remain stable, gain population, or lose population.

In the first place the determination is made whether or not the vacant land has a potential for residential development. Predomi­ nantly industrial and commercial zones are excluded. So also vacant land having excessive slope or located in a flood plain is eliminated. Also, in the case of fully developed zones, some thought is given to redevelopment issues, which may involve changes from residential to non-residential use or vice versa. The determination of new densities to such zones is also considered.

The Determinate Index System (DIS) Analysis

This technique distributes future population growth to zones which have vacant land suitable for residential use. It is based on the theory that certain determinates or development factors, such as existing or proposed water and sewer lines, existing planned and com­ mitted freeways and expressways, and proximity to major activity centers, induce residential development. Specifically, three factors are considered:

1. Existing and proposed water lines with mains of eight inches or more in diameter.

2. Existing and proposed sewer lines with mains of eight inches or more in diameter. 135

3. Existing or committed freeways and expressways.

A fourth factor called the "K factor" is used to allow the planner some opportunity to raise the probability of development based on proximity to major activity centers such as committed residential development, new town sites, and intersections of radial and circum­ ferential freeways and expressways. Based on the above-mentioned factors, a weighted index factor is established for each zone for use in distributing population control totals.

Regional population distribution

The next step was to establish the holding capacity by density type. Technically, the holding capacity is the saturation point or the total number of dwelling units an area can absorb under assumed densities of residential development.

Four basic steps are taken in calculating the desirable holding capacity:

1. The dwelling unit analysis for forecast years 1980 and 2000 is used to determine the percentage of population increment that would demand each density type of housing and also the percentage of developable residential land that each density would require.

2. By utilizing the land factor atlas, the amount of vacant and buildable land, defined as developable, by zone for the forecast years 1980 and 2000 is determined.

3. The developable residential land for each zone is divided into three density categories for the forecast years by vising the results of the analysis. Naturally the 1980 distribution precedes the 2000 distribution, so that fresh holding capacities have to be calculated for the year 2000. U. Existing zoning regulations are considered to determine whether vacant land can be used for possible residential development.

Some of the factors involved in this decision are public preference of one area over another, proposed major residential development, proposed new town sites, and intersections of radial and circumfer­ ential freeways and expressways. The gravity allocation model is then applied to distribute the total population to traffic zones, using the following expression: 136

Y, = (x)_5A--- % ----

£

Where X = Total population to he distributed Y = Population to be distributed in Zone A Z = Weighted index number for unit of developable land Q = Holding capacity of developable units n = Total number of zones to which population is to be distributed

The final step in the population distribution process is to establish the residential net density, based on the incremental popu­ lation gains or losses, for each individual zone. The residential net densities are calculated by means of a simple computer program having the following format:

1. The 1970 population is distributed to each zone. 2. The number of residential acres is calculated from the 1970 landuse map (the land factors atlas).

3. The 1970 population is then divided by the 1970 residential acreage to establish the net density by zone. U. The 1980 increment population is distributed to each zone by high, medium and low density categories, indicating either losses or gains resulting from the projection analysis.

5. The increment population by high, medium and low density is aggregated and then added to or subtracted from the exis­ ting population to obtain the future population by category for each zone.

6. The residential acreage by high, medium and low density categories is aggregated and then added to or subtracted from the existing acreage to obtain the future residential acreage for each zone.

7. The acreage increments by zone are then added to or sub­ tracted from existing acreage to obtain the total future residential acreage. 13T

8. The 1980 population is then divided "by the future residential acreage to obtain the 1980 net density for each zone.

The above eight-step procedure is then repeated for the year 2000. A final check is then made on each distribution prior to transferral of the data to graphic form. Employment pro,lections and distribution

The employment projections and their distribution are carried out in the following steps: (l) The employment base for 1970 is determined for eleven industrial categories; (2) Regional employ­ ment projections by industrial category are made for 1980 and 2000 based on an analysis of labor force participation rates by specific age and sex categories; (3) The projected employment levels for these years are distributed to seven industry categories; (U) The industry employment for 1980 and 2000 is distributed to zones; (5) Employment density is calculated for each zone; and (6) The zone distributions are aggregated to three industrial categories for submission to the State Highway Departments.

The following paragraphs describe the process by which the 1980 and 2000 regional employment is distributed to the zone level. The regional employment projections are displayed in table 13. Planned and committed private developments are assumed to be developed according to the schedule of the developers. However, the great majority of the 1970 employment is expected to remain in the same location during the 1970-1980 decade. Employment is only removed from a zone if there is very definite information on plans for change.

For the 2000 distribution, greater consideration is given to the possibility of employment decreases.

Agricultural employment

The mechanism of distributing the 1980 agricultural employment includes three steps: 1. Unless a major landuse conversion is planned for a zone by 19 8 0, most of the agricultural workers employed there in 1970 will be there in 1980. A major landuse change such as a large park or reservoir will displace workers at approxi­ mately the same fraction of total employment as the fa­ cility takes up the land in the zone. 138

2. The remainder of the regional agricultural decrease is apportioned on a county "basis in the following way:

Each County's Remaining Regional Each County's Percentage of Decrease in Total Portion of the 1970 x Agricultural = the Total Agricultural Employment Decrease Employment

This figure is then subtracted from the 1970 county agri­ cultural employment to determine the county1s 1980 total.

3. To distribute the county decrease to the zones in the county (all zones being decreased) the following ratio is used:

That Proportion of 1980 Agricultural Employment in County = 1970 Employment for 1970 Agricultural Employment in County Each Zone that will be there in 1980

TABLE 13

EMPLOYMENT BY INDUSTRIAL CATEGORY, KANSAS CITY METROPOLITAN REGION, 1970-1980-2000

1970 1980 .... 2000 Agriculture 9,992 8,050 7,100 Mining 730 500 kjk Industry 237,000 279,1+89 366,928 Retail Trade 87,158 109,855 1^9,259 Services 106,768 11+1,963 202,038 Government 68,582 89,189 13U,338 Unclassified 56,598 66,277 78,69*1 Total Employment 566,828 695,323 938,831 Unemployed 29,833 36,595 1+9,1+12

Total Labor Force 596,661 731,918 988,21*3 Labor Force Participation Rate 1+5.0$ 1+8.8$ 1+9.1*$ 139

The methodology for distributing the 2000. agriculture employment is slightly different. Counties with the greatest population gains lose the most agricultural, employment. First, employment in zones with proposed, major land use changes is deleted, as in 1980. The remaining decrease is taken out in two ways: (l) agricultural employment is removed from zones which show substantial population increases in 2000; and (2) in the rural areas, the decrease is done proportionally as in Step 3 for 1980.

Mining employment

Since mining is restricted to so few zones, the distribution for both 1980 and 2000 is done at the regional level. This involves three steps:

1. Examine locations with mining employment.

2. Deduct employment from these locations where urbanization (indicated by large amounts of population growth) is pro­ jected to take place. 3. Reduce others proportionally to the total remaining.

Government employment

Government offices in the KCMR are contacted to see what plans for relocation or expansion are available. Schools are the major source of new locations information. The net employment increase resulting from these surveys is only about 30 to k0% of the total needed to be distributed for 1980. Only a few agencies had any plans after 1980.

To distribute the remainder of the 1980 government employment, it is necessary to assume that small increases would occur at all 1970 locations. A percentage to be applied to all 1970 government employment by zone was established by the following formula: Total 1970 Government Employment Remaining 1970-1980 Increase in = ^23. Government Employment For 2000, the task is more difficult because of the assumption that government employment would be tending toward dispersal to areas of population concentration. The methodology involves three steps:

1. The incremental growth between 1980 and 2000 in government employment by county is proportioned by each county's share of total population growth. This step gives the 2000 government employment distribution by county. 2. Within each county, government employment is assigned to metro centers and activity centers. The amount is hased on the amount of population distributed to the area.

3. Remaining employment is distributed to other zones in the county on a proportional basis to the 1980 government employment.

Retail and service employment

All planned new and commercial developments are located from sources including newspapers, private developers and local government planning departments. If employment estimates for planned develop­ ments were not available, ratios for employees per square foot for the retail selling space, general office and financial type offices are applied. Employment is then distributed to these areas assuming 75% capacity for the more definite projects and about 50# capacity for the more speculative projects.

Industrial employment

Four reports published by MARC were used. These reports are:

1. Supply of Industrial Land, 1970

2. Demand for Manufacturing and Non-Manufacturing Industrial Land (1968-1990)

3. A comparison of the existing supply and Projected Demand for Industrial Land U. Concepts for Industrial Land Development Unclassified employment

The unclassified employment in 1980 and 2000 was distributed to acres of the greatest population growth. The five steps involved were the same for both distributions:

1. Much of the unclassified employment was distributed to retail and service locations such as shopping centers and metro centers. The remainder was distributed to zones with population increases of 500 or more.

2. Each county's increase in population was converted to a percent of the total increase. This percent was applied to the unclassified total employment to determine each county's share of the increment. lUi

3. Each, county's share was proportioned, inside and outside the cordon line on the basis of incremental growth inside and outside the cordon line.

Unclassified employment outside the cordon line was dis­ tributed to incorporated cities.

5. Employment inside the cordon line was distributed by taking those zones that increased more than 500 people and finding each zone's percent of the total gain for all zones increasing substantially. Each zone's percent was applied to the county's increment, and added to the previous unclassified total for that zone.

Using these sources, the 1980 distribution involved three steps: 1. Employment increases in industry, 1970-1980, are converted to acres. Using the "concepts for development1' study, ratios of average employees per gross acre are established. These were averaged for the four industrial categories as follows: Industry Employees/Gross Acre

Construction 15.8 Manufacturing 9•0 T.C.P.U. 8.8 Wholesale Trade 6.3

The resulting demand for acres by category is:

Industry Acres

Construction ij-39 T.C.P.U. 1,173 Manufacturing 1,872 Wholesale Trade 1,371

2. Acres were distributed by category to analysis areas on the basis of most attractive area to least attractive (being sure to include all analysis areas which filled the location qualifications on the industry category). 3. Acres were converted to employment by the ratios estab­ lished in Step 1. Employment was then distributed to zones. Ik2

For the 2000 distribution, the mechanics were slightly different. The Industrial Plan, an unpublished MARC report, was the main source document. 7 Oklahoma'City. Oklahoma

Introduction The Association of Central Oklahoma Governments (ACOG) has thirty-four jurisdictions within its region, and all forecasts and/or distributions were made in full cooperation with these minor civil divisions. The distribution of regional totals to small areas was performed by manual methods. Prescriptive and planning techniques were applied. This study was performed to monitor the 19^5 fore­ casts. A simplified flow chart of the methodology is given in figure 30.

Methodology

Different approaches were used to fix the population, employ­ ment and other variables on a region-wide basis. These figures had to be accepted by constituent cities within the study area. Under individual work elements the task of allocation of areawide figures to traffic zones was then taken up. Preliminaries

Initially a determination of the amount of developable land in each traffic zone was made. Since several traffic zone boundaries had been moved and other traffic zones had been split into smaller ones, the first task was to determine the total acreage of the altered zones.

After total acreage was determined for each zone, restrictions to development were considered. Major flood plains of the North Canadian, Canadian and Deep Fork Rivers were determined from records of flooding, the Corps of Engineers Preliminary flood plain data and soil conservation data delineating alluvial soil types. Acreages of flood plains were subtracted from zone totals. Other factors such as lakes, freeways and other barriers to development were also con­ sidered. Aerial photo interpretations were also used to help determine the total developable acreages in each zone. Noise zones associated with airports were also considered in the determination of developable land.

Existing landuse acreages were then applied to each zone. These were derived from the landuse survey completed in 1970. These were used as control totals in all areas except the urban renewal areas of Oklahoma City. In most cases, the 1995 landuse totals were not al­ lowed to appear less than the existing landuse totals. Landuse Plans of Individ, Determination of Amount Cities Showing of Developable Land Future Development in Each TZ

Regional Landuse Plan

Map Showing TZ’s in Year 2000. Dev. Area TZ'a Outside Lev. Area

Landuse Models: Residential Model TTigfr Tnt.emgi-hy M o H p I

■•Projecting Pop. to 1995 "by TZ ‘Projecting DU's to 1995 “by TZ Projecting Mean Family In­ come by TZ Projecting Employment to 1995 Preliminary Landuse Projections by TZ Mailed to Cities Projecting School Attendance and Planning Commissions by TZ

Time Staging Process (Near, Intermediate, Distant Future)

1995 Socio-Economic Variables

FIGURE 30

GENERAL FLOW CHART OF LANDUSE DISTRIBUTION PROCEDURE, OKLAHOMA CITY, OKLAHOMA .1>5

Where available, landuse plans of the cities in the area were then utilized to determine future development. Projected acreages of landuse for areas covered by landuse plans were taken directly from the landuse plan maps. The development of acreages in other areas was accomplished utilizing the Regional Land Use Plan.

The Regional Land Use Plan delineates an area most logical for future development to the year 2000. Therefore, it was. assumed that the majority of development would take place in this area. A map was prepared to determine which traffic zones were inside the year 2000 development area and which were outside. In those, zones outside the boundary, little or no development above what is presently existing was projected. In the areas that were-inside the boundary, but not covered by existing city landuse plans, landuse models from the Regional Land Use Plan were used to project landuse.

Two models were utilized to project future landuse. These were the residential model and the high intensity model. The following percentages of landuse were utilized in applying these models:

Industrial Commercial Residential Other

Residential Model 0# U# 60# 36#

High Intensity Model 10# h% 50# 36#

In certain areas the landuse projections were based on special studies or other indicators of future landuse. Much of the future industrial landuse was obtained from the Oklahoma City Area Chamber of Commerce. The Crossroads Shopping Center, which will be the largest shopping center in the area, was also included in the pro­ jections for commercial landuse. The North East Study, prepared by the Oklahoma City Planning Department, was used for landuse projec­ tions in the Lincoln Park-Cowboy Hall of Fame area. The Oklahoma City Urban Renewal Plans were also -used in their area of jurisdiction.

The preliminary landuse projections were then mailed to cities and planning commissions in the area. An enclosed letter and blank form were used to invite participation in the development of future landuse by traffic zones. An attempt was made to obtain the cities' feeling for the order in which they expected traffic zones to develop in their cities. To do this, each city was asked to indicate, on the forms that were provided, whether the zone would develop in the near, intermediate or distant future. The cities' replies, along with the landuse plans mentioned above, formed the basis for projecting several socio-economic factors to 1 9 9 5*

Target population figures for each entity approved by that local government were allocated to the traffic zones on the basis of 1U6 landuse acreages developed before. The 1995 target population figures are shown in table lU.

TABLE lU

1995 PROJECTED POPULATION BY CITY, ASSOCIATION OF CENTRAL OKLAHOMA GOVERNMENTS

Projected Entity 1995 Population

Bethany 33,171 Choctaw 10,791+ Del City 1+0 ,1+71+ Edmond 39,1+75 Forest Park 1,971 Hall Park 1,278 Harrah 3,655 Jones 2,359 Luther 3,01*9 Midwest City 80,086 Moore 52,000 Mustang 18,359 Nichols Hills 5,329 Nicoma Park 5,81+1 Norman 123,680 Oklahoma City 1+98,135 Piedmont 1,713 Spencer 12 ,2 8 6 Valley Brook 2,053 Village 18,950 Warr Acres 13,310 Yukon 30,573 Cleveland County 99 Oklahoma County 5.812

Total 1,00*+,1+52 After reviewing the landuse information developed pre­ viously, it was determined that the total study area was planned to hold more people than were projected to be in the area in 1995. Therefore, the residential landuse acres were altered to accurately reflect proportional relationships desired by the local governments, where provided, and where local input was not provided, to reflect staff knowledge of existing and proposed development. A tentative number of dwelling units per traffic zone was developed utilizing the landuse information. Using a mean household size of 2.65, pop­ ulation was assigned to each traffic zone. Traffic zones were then 1U7 accumulated by local government and appropriate adjustments were made to reach the target population figures.

Projecting dwelling'units t o '1995 bytraffic zone

After population figures were developed for each traffic zone using tentative dwelling unit counts, the counts were altered to reflect average household, sizes in each, individual zone. To accom­ plish this a factor was derived to reflect the change in the mean household size in. the study area, between 1970 and 1995. This factor was then multiplied by mean household sizes in 1970 in each traffic zone to give the same figure for 1995. The individual zones were then personally inspected to determine where situations might occur that would not be accurately reflected by the above method. In these cases special alterations were made.

Pro j ecting 1995 mean' family income by traffic zone

A base for making projections of income for 1995 was developed using historical and projected data prepared by the U.S. Bureau of Economic Analysis and contained in the OBERS projections. The OBERS projections are a combined effort of the Office of Business Economics (OBE) and the Economic Research Services (ERS) of the U.S. Department of Commerce. The historical data ran from 1969 to 2020. The areas reviewed in the OBERS reports included the Oklahoma City economic area, the lower Canadian River basin, and the state of Oklahoma. Historically the annual percent change of per capita income ranged from -1.2 to 6.9. In the projections the percentage of annual change ranged from 3.1 to 3.9* Therefore income for 1995 was projected to increase at a rate of 3.5$ per year. All income figures were adjusted to reflect the 1965 dollar as the base.

A mean family income was assigned to each traffic zone based on the projections of the percentage of annual change in income prepared by the U.S. Bureau of Economic Analysis. Using 1965 dollars as the base, the 1995 mean family income of $18,180 showed a 137.9$ increase over the 1971 mean family income of $7,6^1. Projecting employment to 1995 by traffic zone

In projecting the distribution of 1995 employment, two objectives were attempted. They were:

1. Distribute the overall study area employment according to projected landuse changes as expressed in acreage figures.

2. Maintain a consistent and reasonable distribution in view of the available 1971 data. Basically, the procedure used to meet these objectives vas to: (l) Set target employment figures for the entire study area and then /to divide this target figure between the central business district, the central area, and the remaining study area. These areas were originally delineated by Wilbur Smith and Associates. (2) Develop appropriate factors (based on the amount of employment in each zone in 1971) to distribute the 1995 employment. (3) Modify the resulting distribution to reflect accurately data provided by the Oklahoma City Urban Renewal Authority and to reflect changes suggested by local planning Departments and other sources.

The CBD projected employment of l+l+,2l+5 represents 8.97# of the total regional employment. The factors used to distribute CBD employment ranged from a low of 37.9 employees/acre for a normal retail zone to 1552.1 employees/acre for a major other employment zone.

The employment for the central area was projected to reach 127.032 or 25.2# of the total regional employment. The factors used to distribute central area employment ranged from 1 9 .8 employees/acre fora normal retail zone to 381.3 employees/acre for a major other employment zone.

The remaining 65.83# of the region's employment was distributed by factors of 6.6 employees/acre for retail employment and 7.8 employees/acre for other employment.

It should be noted that the basic landuse to employment rela­ tionship mentioned does not utilize public landuse; therefore public employees (including school employees) could not be represented using this procedure. To correct this situation, all public facilities (federal, state, local, and schools) were listed and individually projected. These projections were listed by zone and added to the factored data. Zones located outside of the central urban boundary were pro­ jected at a level slightly above their existing employment.

In 1995 there are pro j ected to be 1+99,176 employees in the OCARTS study area. This figure includes 88,298 retail only employees and 1+10,878 other employees. When the population and employment are compared, a ratio of .I+969 employees per population results. Projecting 1995 school attendance by traffic zone

Each school district within the transportation planning area was contacted for information relating to expected or planned school closings and for estimates of optimum sizes for elementary schools, junior high schools, and senior high schools. The school-age population was developed in the ACOG Economic and Population Study, and each age was assigned to a specific school grade.

The optimum school sizes (625 for an elementary school, 1200 for a junior high, and 2000 for a senior high school) were then applied to the population not served by existing schools and were located in those areas where further or new development is expected to occur.

College enrollment was projected independently by each college and simply compiled by traffic zone.

In 1995 school attendance was estimated to total. 271,^70. This includes 63,600 projected college students, 9^SM*2 students enrolled in grades 7-12, and 113,^28 students enrolled in grades K-6. •150

Pittsburgh, Pennsylvania

Introduction

The Southwestern Pennsylvania Regional Planning Commission (SPRPC) was authorized by federal and state legislation to prepare a year 2000 comprehensive landuse plan for the City of Pittsburgh and the six surrounding counties of Allegheny, Armstrong, Beaver,. Butler, Washington, and Westmoreland. The plan is a composite of several functional plans, including transportation, environmental engineering, open space, housing, urban development and special facilities. These plans establish policies and programs for approaching regional development within a long-range context.

Forecasts of year 2000 landuse activity were prepared by utilizing computerized models. These models permit an accurate, detailed, and efficient analysis of future development. They consist of statistically valid, mathematical quantifications of the complex interrelationships among jobs, housing, population, income, land and other locational factors.

However, the most sophisticated and accurate techniques for forecasting future activity reveal only two things: what is most likely to occur if past trends and interrelationships continue, and what the probable impact on development will be if specified actions are taken. Nothing in the forecasting procedure can be sensitive to environmental quality, social objectives, or fortuitous events. The determination of what sort of region this should be in the future is absent. This task is left to human judgment. It must be predi­ cated on some consensus of what "good" development is.

Working sessions for the review and adjustment of the forecast were held specifically for this porpose. The City of Pittsburgh and the six county planning departments worked with SPRPC in reviewing the data to be certain that it reflected present development activity and a reasoned prognosis of the future. Where necessary, adjustments ■were proposed, evaluated in both local and regional perspectives, and incorporated if agreeable to all participants. These adjustments thus enhanced the credibility of the statistical forecast, promoted consistency between local and regional development plans, and estab­ lished consensus on the dimensions of regional development.

The general relationship between the technical process and policy considerations is given in figure 31. Collect 151 Base Data Develop Mathematical Models

Prepare Regional Working Sessions, to Forecasts of & Review and Make People, Jobs, Income Adjustments and Interventions

------

Prepare Small Working Sessions, to Area Forecasts of C—7 Review and Make People, Jobs, Adjustments and Income and Land Interventions

J&L

Prepare Small Area / Propose and Evaluate Plans c*r Demand Forecasts For and Make Adjustments Transportation and Interventions Sewer and Water Recreation Housing \✓ Conduct Delegation Meetings

V Prepare Regional Development Plan

Technical Policy

FIGURE 31

THE COOPERATIVE PLANNING PROCESS, PITTSBURGH, PENNSYLVANIA 152

Methodology

The SPRPC models operate at two scales:. regional and small area. The regional forecasting model estimates the level of region- wide employment, population, housing and income. The small area model allocates these measures to 968 small areas known as Traffic Analysis Zones. These zones vary in size from a city "block in down­ town Pittsburgh to a township in a rural area.

Two mathematical models, SPIM and SPAM, are described here. The Southwestern Pennsylvania Inter-regional Model (SPIM) is one which forecasts employment, population, migration, income and housing on a regionwide basis. The Southwestern Pennsylvania Allocation Model (SPAM) was developed to help allocate regional totals to traffic zones. The model is based on historical data which have been correlated with the locational characteristics of employment centers and residential areas. In essence, locational, forecasts are derived indicating where private interests are most likely to develop resi­ dential or employment centers given certain alternative public invest­ ment policies. Consequently, the models are designed to evaluate the impact of sewer, open space, transportation and other such plans.

SPAM can be summarized in its stepwise progression of dis­ aggregation. Manufacturing employment is allocated first; then non­ manufacturing employment with the exception of retail centers' employ­ ment is determined and assigned to appropriate traffic analysis zones. The land required for nonresidential use is absorbed or accounted for as developed land; housing units are removed for both natural removal and policy decision removal and then new housing is allocated. New housing unit allocation is based upon new employment allocation. Land required for such housing development is absorbed. The last step reallocates people, income and shopping employment to the sub-regional level. The shopping center allocation model is employed last, since this forecast is based upon both forecasted regional household income and new housing unit forecasts. (See figure 32 .)

The technical process which forecasts housing units and employ­ ment in each traffic zone is termed allocation because the summation of these zonal forecasts must equal the respective regional forecasts. Once an aggregate forecast has been established for the region, it must be allocated to traffic zones, as indicated earlier.

The aggregate forecast could be done by a "level of activity model" for the forecast year, or by a "change model" which takes into account the change since the preceding forecast or base year. In the allocation models which have been documented in this paper, the "change model" has been used. 153

Forecast Aggregate Control Totals (SPIM)

Allocate Employment Except Shopping Centers

Absorb Land For Non-Residential Uses

Allocate Change in Housing Units

s /

Absorb Land for Residential Uses

Yes iwered Land Adjustment Required!

No

Forecast Income and Population

Allocate Shopping Center Employment and Land Used

Employment and Housing Units Allocated

FIGURE 32

SOUTHWESTERN PENNSYLVANIA ALLOCATION MODEL, PITTSBURGH, PENNSYLVANIA '15b

Two basic statistical models have been used, in SPRPC's alloca­ tion process. They are the trend model and the regression model. The time series or trend model is one of the..most common approaches used in forecasting. In a strict sense, it is not a prescriptive model but a forecasting technique.

The regression model, unlike the trend model, builds into itself interrelationships. On the other hand, the trend model, to compensate for the lack of adequate prediction ability, would have to be exter­ nally manipulated to reflect other relationships, not inherent in the trend to make it prescriptive. The general procedure for developing the regression equations is given in figure 33.

The normal procedure is to prepare a SPAM forecast without any zonal overriding interventions. This no-intervention forecast is examined with any adjustments being made by a "handset." These "handsets" are unaffected by subsequent model runs, while the remaining zones are affected to the extent that the sum of the "handset" and the "non-handset" zone changes must equal the regional control total. Figures 3^+, 35, and 36 indicate the procedure of allocating employ­ ment, housing, income and population to small areas. 155

1 Formulate Hypothesis

2 Select Variables

3 Plot Dependent Variables vs. Independent Variables

>1 k Run Correlation Analysis and Multiple Regression

5 Modification Needs

___ Evaluate Regressi^n^Results^~~ Z ^ =>m

6 Map Regional Equation

vV 7 _ Evaluate Equation's Per-fni-mann^ -- ^

8 Equation Complete

FIGURE 33

EQUATION DEVELOPMENT METHODOLOGY, PITTSBURGH, PENNSYLVANIA 156

Input From SPIM

Step 1

Allocate Three Manufacturing Categories (Trend Model) .Step 2

Allocate One Manufacturing Category and Non-Manufacturing Category (Regression Analysis)

Step 3

Account for Impact of Transit on Specified Non-Manufacturing Categories

Step U

Absorb Non-Residential Land Required

Operate Housing Unit Allocation Model

Step 5 Allocate Special Non-Manufacturing Employment Such As Shopping Centers

Employment Allocated

FIGURE 3b

EMPLOYMENT ALLOCATION MODEL, PITTSBURGH, PENNSYLVANIA Input From Employment Input From SPIM Allocation Model (Except Shopping Ctrs.) I------Ster> 1 Housing Units Removed Policy and Natural

Step 2 N/ Allocate New Housing Units

Step 3

Determine Structure Type

Step U f - Account for Transit Impact

Step 5 •V

Absorb Non-Residential Land

Step 6

Adjust

Housing Units Allocated

FIGURE 35 HOUSING UNIT ALLOCATION PROCESS, PITTSBURGH, PENNSYLVANIA Input From Employment Allocation Model (Except Shopping Centers)

Allocate Population

Allocate Regional Income

* L Income and Population Allocation

FIGURE 36

INCOME AND POPULATION ALLOCATION PROCESS, PITTSBURGH, PENNSYLVANIA 159

Rockford, Illinois^

Introduction A Dwelling Unit (DU) Allocation Technique for Winnebago County was developed by the Rockford Area Transportation Study (RATS) in December 1971 for allocating housing growth to various parts of the City of Rockford and Winnebago County. The total population for the study area for 1975 through. 2000 was forecast, from which occupied housing and total housing was derived. The allocation to sectors, districts and traffic zones was performed manually through this DU Allocation Technique. The following housing forecasts were used:

Year Total Population Occupied Housing Total Housing

1975 262,900 8 3 ,^6 0 87,668 1980 283,U67 91, MU 95,78U 1985 307,700 102,567 1 0 7,061* 1990 33^,125 111,375 116,258 1995 360,U78 120,159 125,^27 2000 387,253 1 2 9,08U 13U,7^3 The 1970 gross vacancy rate was 5.2$ The occupancy rate was 3.21. It is assumed that the vacancy rate will decline and return to the i960 level as the economy improves and that the occupancy rate will also decline as urbanization continues. Therefore, the fol­ lowing vacancy and household rates were used:

Year Vacancy Rate Persons/Occupied

1975 h.8% 3.15 1980 b.5% 3.10 1985 k.2% 3.00 1990 k.2% 3.00 1995 k.2% 3 .0 0 2000 k.2% 3.00

Methodology This model was designed using a gravity-allocation technique and it was presented in the Rockford Area Transportation Study report entitled "A Dwelling Unit Allocation Technique for Winnebago County" in 1971. The purpose of this model was to distribute or allocate the future dwelling unit increases in Winnebago County, first to the sectors in the county and then to the individual districts, within each sector. In this way, the model was not intended to make actual l6o housing projections into the various segments of the county. It is, then, independent from the projections themselves, but it does allo­ cate the projected growth during a given time period.

In simple terms, a gravity-allocation model can be expressed as follows:

Strength of growth-related factors in the Percentage, of new growth to ...... sector...... be allocated to the sector ” Strength of growth-related factors in all sectors

In the allocation process for Winnebago County, the Central Business District has been omitted because residential growth is expected to be small or almost negligible. Eliminating the CBD thus eliminates sectors 0 and 1 of the RATS system from the allocation procedure. District 1 from each of sectors 2 through 7 have also been omitted because they are also located in the CBD area. Finally, sector 8 has also been removed from the allocation process because it is almost completely rural with only minor growth expected. The sectors and districts that have been omitted from the regular alloca­ tion process, however, have been given a very general allocation amounting to 3.k% of the total growth for Winnebago County.

The remaining area, sectors 2 through 7 excluding district 1 of each sector, was analyzed in terms of growth-related factors for the allocation process. The following factors were selected for the allocation of residential growth to the sectors: (l) transportation capacity within the sector; (2) transportation linkages among the sectors; (3) interaction within the sector; (H) employment; (5) major shopping centers; (6) special attractions (e.g., colleges, golf courses); and (7) major parks.

After these factors were expressed in quantitative terms, several sector matrices were developed in order to determine the relative weight of each of the seven variables. The final results are shown in table 1 5. In the second phase of the study the same gravity allocation technique was used with a new set of variables (distance from CBD; trunk services; four-lane roads; and elementary schools) to distribute the sector allocations down to the district levels. Once again, the growth indicative variables or factors were selected on the basis of planning judgments and applied to sectors 2-7 (excluding district 1 of these sectors).

At the end of this phase of the study, detailed consideration was given to making the remaining minor but necessary allocations to the rural sector 8, and to the Rockford CBD (district 1 of sector 2-7 and i6i

sectors 0 and l). The sector and district housing allocation in percentage form were then used in conjunction with the most recent county housing forecasts to arrive at an actual number of housing units to be absorbed into the various sectors and districts over several projected time periods.

TABLE 15

PERCENT OF NEW GROWTH FOR 1970-1995 ACCORDING TO ALLOCATION MODEL BY SECTORS, ROCKFORD, ILLINOIS

Percent of New Sector Dwelling Units

2 2k.2 3 35.7 k 11.2 5 U.9 6 8.3 7 15.7 Total 100.0

The final phase of the study dealt with considerations to be weighed in breaking down the district allocations into the smallest aerial units in the zone. It was decided that the most important factors to be considered at this level would be existing dwelling unit densities, current zoning, and existing landuse plans. These three factors determine the potential for residential development in the individual zones. 162

'Toledo; Ohio10

Introduction

The Toledo Regional Area Plan for Action (TRAPA) was formed as an ad hoc committee to comply with the requirements of the 19^2 Federal Highway Act. Originally a separate entity that functioned under the Toledo-Lucas County Plan Commissions, TRAPA is now an integral part of a larger regional agency, the TMACOG. The original TRAPA was organized to conduct regional landuse and transportation planning, A landuse inventory of the TRAPA area (Lucas County plus the northern part of Wood County of Ohio and three townships of Monroe County in Michigan) was completed in 1965- The landuse analysis and the 1985 forecasts were completed in August, 1973, and the results documented in a technical report. These data were used to arrive at a comprehensive regional land­ use plan and served as the starting point from which all other pro­ jections, including those related to transportation, were made. The dwelling unit projections within each planning area (33 planning areas in all), with their related commercial and industrial service facilities, provided the data "base used in assembling travel data and trip making for the proposed transportation system.

Methodology

A simplified flow chart of the methodolody used by TRAPA is shown in figure 37.. The sequence of operations was as follows:

Surveys, regional totals and patterns of land consumption

A survey of all landuses in the region was conducted concur­ rently with the Origin and Destination Survey. Additional studies were made of the physical features and geology of the region, existing utility systems, storm drainage patterns and existing public utilities such as parks, schools, fire stations, etc. The potential landuses of the soils in the region were mapped. An inventory and analysis of sanitary and water supply systems in use in 1966 was conducted. This inventory of public facilities provided the location, size and general service area of schools, parks and recreation areas (open space), and public service and cultural and institutional facilities (see Box l).

Through public participation a survey of community attitudes and preferences was made with regard to housing, shopping, public trans­ portation, neighborhood characteristics, mobility, recreational activities, regional identification, urban services, regional environ­ ment and public expenditures (see Box 2). 1. 1965 SURVEYS 163 0-D Surveys Physical Utility Storm Public Features 8 Systems Drainage Utilities Landuse Geology Patterns Parks, Schools

2. Community Master Plan Attitudes Study

r y r Population & S7 Economic Studies Patterns of Land Consumption for lQ8q ft P01S for 1950 & 1965 JuL i&r Patterns of Land Consumption for 1985 & 2015 tr"Goals for the Region

Urban Form Study ~j—

XL. 9. Alternate Landuse Plans for Year 2015 & 1985, Concept Plans A & B

1 0 . N11. Policies s Final Selection of 1985 Landuse Plan (Concept Form)

Distribution to Small Areas of Landuse & Socio-Economic Data

FIGURE 37 GENERAL FLOW CHART OF LANDUSE PLANNING PROCESS, TOLEDO, OHIO 16U

Using the 1952 Master Plan study for the Toledo urhan area, a frame of reference was available from which to view changes in the pattern of land consumption in the region (see Box. 3). The 1950 and 1965 land- consumption rates and-landuse data were tabulated, under the categories of Residential, Commercial, Industrial-Utilities, Recreation-Institutional, Roadways and Agriculture/Vacant (see Box U).

With the help of regional totals forecast for 1985 and 2015 for population and employment (see Box 5)> landuse consumption for 1985 and 2015 was estimated for different categories: Residential, Com­ mercial, Industrial and Utility, Recreation and Institutional, Roadway and Vacant. Projecting landuse trends for the future depended for the most part on the size,, density and locational preferences of the population. Thus land development totals for 2015 were fixed in six categories as shown above (see Box 6). Development of the landuse plan concepts

After performing the analysis of inventoried items, there remained the task of synthesizing the results into a meaningful and. consistent landuse plan. First, goals for regional development were determined and adopted by the Coordinating Committees (see Box 7). Goals, however, can each be accomplished in a number of different ways. Therefore, a. number of urban forms were examined to judge which form best suited the goals of the community. The analysis concluded that the "Regional Centers" form appeared most appropriate for the region (see Box 8). Two alternative landuse plans were determined to be potentially feasible and sufficiently different from each other to offer a meaningful choice. The alternatives, designated as Plans A and B, were based on differing policy emphases, but with identical estimates on the total scope of development,e.g., regional population total, area urbanized.

Plans A and B were first designed for the year 2015 or 50 years beyond 1965, the study base year. Looking into the long-range future was considered useful because it put the 1985 plan into perspective as a point along the way. The 1985 versions of Plans A and B were designated with the same emphasis as the 2015 plans (see Box 9).

Landuse and socio-economic distribution to planning areas

The distribution of residential capacity to planning areas was undertaken first. The concept of full development or "holding capa­ city" was utilized in performing this distribution, for which the maximum amount of developable land within each planning area was established. In this way, present development could be viewed as a percentage of ultimate capacity and a determination made of the current state of the development cycle. The region was divided into 33 planning areas and the time required for each to move from its 165 present state to full development varied by the size of. the area, its distance from existing urbanization and the relative attractive­ ness of. the area for development. Four distinct stages of growth were identified:

1. A very slow period of development when areas move from totally rural to rural plus non-farm. This is when large lot homes are scattered on main roads and generally there are no public utilities or subdivisions.

2. A slow growth of the early subdivisions, often without total, public utilities. At the end of this stage the area may be 305? developed. 3. A rapid growth "boom" period, often heralded by installation of public sewers, in which land urbanizes rapidly, land prices rise, and the area is 10% developed in 10-20 years.

1+. A slow growth period as total capacity is reached. The slowing of growth is due to high land prices and difficulty in assembling the last pieces of land for development.

A set of Gompertz curves was drawn of typical patterns of growth. This development process was simulated for each planning area by applying rates of growth commensurate with the area’s size and location. Then settlement patterns under the alternate plans’ assumptions were projected. Future dwelling units for each planning area were calculated based on projected average densities for both single and multiple family units. Future population was calculated by multiplying persons per household by the projected number of dwelling units.

A summary of the stages of growth and development (as a per­ cent of ultimate capacity) and time required to achieve a given developmental stage under different growth rates was finalized for 2015. Similarly, a staged development summary by planning area was prepared for 1985. In 1970 the committees concurred with the staff that a single landuse plan for 1985 should be finalized and Plan A was modified accordingly. This proposed landuse concept plan for 1985 was based upon observed and projected rates of land consumption and relation­ ships between landuses (see Box 11). Policies were also adopted for the 1985 regional plan (see Box 10) which represented the means by which these goals would be translated into specific courses of action. Population and dwelling units were thus distributed to planning areas. These planning area totals were then further distributed 166 to census tracts/traffic zones. Other variables for which study area totals were established, such as EmploymentLabor Force, Auto Ownership, Income and School Enrollment, were systematically distri­ buted to planning areas and then to traffic zones or census tracts as required.

Sample Problem

A four-celled region will help illustrate how the allocation to small areas is accomplished.

A region has a 19^5 population of 67,200. The acreage of the plan­ ning areas is known. A compre­ hensive capacity . • study has been done. An economic study has also been accomplished.

Table 16 provides the basic information.

TABLE 16

INVENTORY OF LANDUSE DATA, TOLEDO, OHIO

Plan- 1965 Total 19^5 Acreage Total Acreage ning 1965 Persons/ Acres Resi- Suitable For Residential Area Pop. Occ. DU in PA dential for Future Use at Capacity ______Acreage Residential Use______

1 31,000 2 .8 3578.1 102U.8 6 9 .2 109^.0 2 25,100 3.6 U66U.2 1 656.9 897.9 255^.8 3 5,Uoo 3.7 6966.0 555.3 160U.U 2159.7 It 5,700 3.3 7U67.9 67 8 .2 U65.3 11U3.5

Calculating the added residential land is done as follows:

1. A deduction is made reflecting the fact that only 95% of the total land can be developed due to factors of parcel shape, size, and ownership. Five percent is considered undevelopable parcels or "waste." l6T

2. Physically constrained land is removed from the potential residential land. This consists of flood plains, park­ land, conservation areas, and land in which bedrock is less than five feet from the surface.: This assumes- no residential development can take place on this land. Like­ wise, a constraint is placed on land lying in the Deep Sand Belt.

3. Vacant land which was "committed" for non-residential uses in 19 6 5* such as industry, commerce, or major institutions, is subtracted. The committed uses were determined from existing zoning, local master plans, and announced major developments.

H. In addition to subtracting physically constrained land and committed major non-residential activities, a final reduc­ tion represents a 23$ roadway component and a 12-17$ figure for residential services such as schools, parks, and commercial areas. This range is consistent with the figures found in the 1965 landuse inventory.

5. And, finally, most important for planning purposes, the net vacant residential land is then translated into the ultimate number of dwelling units of the housing type and density that can be built for the area on the remaining land. Population may then be calculated by multiplying occupied dwelling units by average family sizes for single and multiple family units. As noted previously, average net residential densities of from 3.0 dwelling units per net residential acre (du/nra) to 6.0 du/nra are projected for single family units, with U .5 du/nra being the most typical.

For planning areas which are almost all developed, the calcu­ lation of future growth simply reflects the potential development under zoning on the remaining vacant parcels. The typical development density and not the maximum permitted is used. For areas which are one-third to one-half developed, there is usually enough momentum in development trends to give a good idea of ultimate density. In presently rural areas, however, some judgment has to be used to project the ultimate urbanized density of these areas at some point in the distant future when all landuse in the region is urbanized. For this calculation, the minimum density used for a single family home is three dwelling units per net residential acre, with a minimum of 10$ of the dwelling units in multiple family development at a density of 15-20 dwelling units per net residential acre.

Determination of what density to use is subjective based on current densities, zoning on undeveloped land, and the socio-economic 168 character of the area. For multiple family units, an average density of 18 du/nra is assigned. For planning areas outside the inner city communities, a range of multiple family proportions of total housing stock is between 10# and 30%, with 20#'. being an average proportion. Determination of the exact proportion of dwelling units in the mul­ tiple family category is made on the basis of existing proportions, the degree of the development in the area (assuming that apartments become the only economical use of small parcels of residential land remaining in an area), and the particular character of the area.

For fringe planning areas which are now largely rural, the 10# minimum is assumed. For inner city planning areas where redevelop­ ment will take place, the proportion of new units is assumed to be roughly that of the existing distribution because there are no major indications of any major shift in taxation and assessment policy which increase the proportion of single family dwellings.

Once capacity residential acreage is estimated, the next step in projecting future development is to assign total estimated dwel­ ling units to the land. This is termed "holding capacity." The population capacity is then determined by multiplying total dwelling units by average family size. Dwelling units are projected by a-model which relates the past history of dwelling unit construction in a planning area with the ultimate capacity of that planning area in terms of dwelling units. The ultimate capacity varies based on the estimated density of dwelling units permitted to be built. For example, if 1000 acres of vacant land are developed with one-acre lots, no • - more than 1000 houses can ever be built. However, if developed totally with garden apartments, the area can hold 15,000 units.

Thus, certain general assumptions are made regarding the capacity conditions of density and housing mix, as follows:

1. The closer an area approaches full development, the greater the likelihood of higher density multiple family development occurring. The percentage of total dwelling units which would be single or multiple is determined by analyzing the 1965 dwelling unit mix and subsequent trends in each plan­ ning area.

2. No planning area at full development would fall below a 10# multiple family component and densities of three dwelling units per net residential acre for single family housing and twelve to eighteen dwelling units per net residential acre for multiple family housing. These densi­ ties are considered mi ni.mal for maintenance of full urban services and apply generally to development in urban and presently rural areas of the TRA. 169

The four stages of growth indicate that once certain critical stages are achieved in the development cycle of a planning area, further growth is an inevitable, process. To illustrate, the history of a rural township of 20 square miles and a capacity of 2U ,000 dwelling units might proceed as follows:

1. A very slow growth (l.k% of ultimate capacity per year) for about 1*0 years, going from 0-10%. of ultimate capacity (period 1890-193 0) up.to 2^00. dwelling units or 8-9,000 persons in that period of time.

2. A slow urban growth (average 1$ of ultimate development per year) for about 20 years, going from 10$ to 30$ of ultimate capacity (period 1930-1950) up to 7200 dwelling units or 22.000 persons.

3. Rapid development or '’boom" period (average 2.5$ annually) for about 16 years, going from 30$ to 70$ of ultimate capacity (period 1950-1966) up to 1 6 ,8 0 0 dwelling units or 50.000 people.

k. Slow growth (averaging 1$ per year) for remaining 30 years (period 1966-1996) as remaining land, is urbanized to 100$ of ultimate capacity up to 19 ,0 0 0 dwelling units or 75*000 people.

Total Period of Development = 106 Years

This model illustrates the general pattern of the development cycle. The rate of growth per year is dependent upon its development, which is a function of its image or marketability, as well as of its location. Only the timing of the "1*0$ boom" period is subject to major variation. A "boom" averaging 1.5$ per year will take 27 years, one averaging 2$ will take 20 years, and one averaging 3$ will take 13 years. In any case, the filling up of land generally sends prices up and because density demands utilities, these generate additional growth.

The development cycle model makes it possible to consider alternative landuse plans or settlement patterns. This is because patterns can be simulated by ranking the projected development of areas relative to one another. Table 17 is a summary of the stages of development (as a percentage of ultimate capacity) and the years required to achieve the given state under both fast and slow growth rates.

The larger geographical areas would require a longer time span to achieve full development, but would respond to the same cycle 170

TABLE 17

ESTIMATED YEARS REQUIRED TO ACHIEVE GIVEN STATE OF DEVELOPMENT, TOLEDO, OHIO

Planning Areas With Less than 20,000 Dwelling Units at Capacity

Percent of Fast Slow Capacity . Growth Growth...... No. of Years ' Developed... Rate Rate..... to ' Achieve (fast to slow) Very Slow Growth Period 0-10# 0.5#/yr. 0.3#/yr. 20-35 yrs.

Slow Growth 11-20 1.0 0.5 10-20 21 - 1*0 2.0 1.0 10-20

Boom Period Ul-60 1*.0 2.0 5-10 61-80 2.0 1.0 10-20

Leveling Off 81-90 1.0 0.5 91-100 0.5 0.3 10-20 20-35 Total 85-160 yrs.

Planning Areas With More than 20,000 Dwelling Units At Capacity

Very Slow Growth (fast to slow) Period 0-10# 0.3#/yr. 0.2#/yr. 35-50 yrs. Slow Growth 11-20 0.5 0.3 20-30

21 - 1*0 1.0 0.7 20-30

Boom Period Ul-60 2.0 1 .3 10-15 61-80 1 .0 0 .7 20-30 Leveling Off 81-90 0.5 0.3 20-30 91-100 0.3 0 .2 35-50 Total i 6o-235 yrs. 171 formula. It should he pointed out that the only defect in the holding capacity model is that it may reflect too low a density of development. Lacking many indicators of non-residential development potential in rural fringe planning areas, too high a. percentage of vacant land may have been judged to be usable for residential purposes. Thus holding capacity simply reflected the potential development under typical density, not the maximum permitted.

In summary, table 16 estimates capacity residential acreage, while table 17 estimates the number of years required to achieve a given stage of development, derived from a Gompertz curve. The assignment of total estimated dwelling units to the available land is shown in table 18 The results of the development cycle analysis in each planning area are listed in table 19 The estimated average household size and the 2015 population for each planning area are shown in table 2 0.

The total population compares well with the control figure of 102,000. In this sample problem a one-shot estimate and distribution has been demonstrated. In the real-world situation, several trial cycles would have to be gone through to match the regional control total of population derived exogenously. TABLE 18

DISTRIBUTION WORKSHEET, TOLEDO, OHIO

Dwelling Units Proj. % of Unit Actual Plan­ Existing % DU's/ Existing Acres/ Distrib. at Capacity Ratios ning No. of DU’s Density- Proj. DU/Acre (Total) Area Multiple ' Single Multiple Single Multiple Single Multiple Single Assumptions 1. Use existing 1 U2.6% 51M Lensities 37.5 62.5 193.9 830.9 1*5 55 51*38 DU's 7335 DU's 2. 8 DU/ac for sing. 1*321 7210 22.3 6.7 22 8 2l*7 .2 ac 81*6.8 ac 22 DU/ ac for mult. 22 DU/ac 8.7 DU/ac 38.2 DU/ac for dorms

6 35.9 1621.0 15 85 15% 8ojg 1 . Area is over 1/2 2 T O 6703 1 2.lt lt.1 11.9 3.7 1702 DU's 9652 DU's developed. Use ex­ lUl.8 ac 21*13 ac isting ratios. 12 DU/ac 3.8 DU/ac 2. Assume 1* DU/ac for sing.; 12 DU/ac for mult.

305S 70# 1. Area is 25% dev. 33l*l DU's 7801* DU's Assume same densities 3 5 95 1 0 .8 5Mt.5 30 70 208.8 ac 1950.9ac 2. Assume 1* DU/ac 72 1552 6.7 2.9 16.5 it.5 L6 DU/ac H.O DU/ac for sing.; 1*.5 DU/ac for future. Assume 1.6 DU/ac for mult.; 1 6 .5 DU/ac for future

15? 85# r68 DU's 1*382 DU's It 7 93 1 1 .6 666.6 15 85 1*8 ac 1095.5 ac 150 2153 13 3.2 17 5.2 L6 DU/ac U.O DU/ac -* ro TABLE 19

PERCENTAGE DISTRIBUTION OF REMAINING GROWTH, TOLEDO, OHIO

Rate of Growth______Years Required for Development at that Rate

$ Capa­ Plan­ Sector city Year Dev. DU's Dev. 2015 ning Growth Devel­ Com­ as of $ of Area Rate oped pleted 2015 Capacity 0-10 11-20 21-1*0 1*1-60 61-80 81-90 91-100$

1 Fast 90.3 1.0/1 0.5/18 1981* 12,773 100$

2 Fast 63.3 2.0/8 1 .0 /9 0.5/18 2000 11,35^ 100$

3 Slow lk.6 0.5/10 1.0/19 2.0/10 1.0/19 0 .5/18 0.3/30 2071 7,913 71$

k Slow hk.j 2.0/7.5 1.0/19 0 .5/18 0.3/30 20ltl 1*,738 92$ 173 ifi*

TABLE 20

DISTRIBUTION WORKSHEET OF RESIDENTIAL DEVELOPMENT, TOLEDO, OHIO

Plan­ 1965 2015 2015 Vacancy Est. # 1 ning 1965 Persons/ Persons/ Est. Rate of 2015 Area Pop. Occ. DU Occ. DU DU's % Occ. DU's Est. Pop.

1 31,000 2 .8 2 .6 12,773 U.l 12,21*9 31,81*7

2 25,100 3.6 3.1 11,35U 2 .6 11,059 3U,283

3 5,1*00 3.7 3.0 7,913 3.6 7 ,6 2 8 22,881*

k •_5*.TQ0 3.3 2.9 1*,738 2.7 l*,6l0 13,369

67,200 102,383 175

NOTES

The following reports and studies have been excerpted and summarized in compiling the ten landuse methodologies presented here.

^•Akron Metropolitan Area-Transportation Study (AMATS), Landuse pnd Development Forecast; 1970-2000. Technical'Report (Akron, Ohio: AMATS, Jan., 1976).

^City of Austin, Austin'Tomorrow: LanduSe. An Analysis Of Urban Development in Austin (Austin, Texas:City of Austin, June, 1975).

3stark County Area Transportation Study (SCATS), Population'and Economic Study, Smftll Unit Forecast (Canton,- Ohio: SCATS* 1977); Ohio Department of Natural Resources (ODNR), Land Capability Analysis. County Report No. 3, Stark County (Columbus, Ohio: ODNR, June, 1975)*

^Chicago Area Transportation Study (CATS),' Data PPQ.jactions, Vol. II (Chicago. 111.: CATS, I960); John R. Hamburg and Robert H. Sharkey, Landuse Forecast (Chicago, 111.: Chicago Area Transportation Study, 193TT; ^Mid-Ohio Regional Planning Commission (MORPC).'Franklin'County Regional Transportation Plan (Columbus, Ohio: MORPC, 1972); Mid-Ohio Regional Planning Commission (MORPC), A Technical Report on the Year 2000 Landuse and Trip GOneration Variables (Columbus. Ohio: MORPC, 1977). ^Mid America Regional Council (MARC), Methodology for'Landuse. Population and Employment Distribution. Kansas City Metropolitan Region (Kansas City. Mo.: MARC, 1973).

‘Association of Central Oklahoma Governments (ACOG), Transporta­ tion Technical Reports. 1971-7^ (Oklahoma City, 0k.: ACOG, 197k). O Southwestern Pennsylvania Regional Planning Commission (SPRPC), Forecasting Framework: Jobs. People and Land. (Pittsburgh, Pa.: SPRPC, 197^).

^Rockford Area Transportation Study (RATS), A Dwelling Unit Allocation Technique for'Winnebago County (Rockford, 111.: RATS, 1971).

10Toledo Regional Area Plan for Action (TRAPA), 1965 Regional Landuse inventory/Analysis and'1985'Forecasts (Toledo, Ohio: TRAPA, 1973). APPENDIX B

DATA AND CALCULATIONS FOR THE DSGM

TABLE 21

GROUPINGS OF CENSUS TRACTS ARRANGED BY DISTANCE FROM HIGH VALUE CORNER (DSGM)

Distance Census Tract Distance Census Tract in . Miles Numbers . (i960) , in Miles ..... Numbers ' (i960)

0 28 4.25 2 , 3, 63 .75 29, 34 4.50 13, 43B, 64, 66 1 .0 0 23, 27, 30, 37 4.75 6 2, 65 1 .2 5 38 5.00 5 6, 5 7, 6 1, 68 1 .5 0 17, 1 8 , 2 2, 2 6, 33, 4l 5.25 1.75 19, 36 5.50 58, 6 0, 69, 77 2 .0 0 1 6, 2 0 , 2 1, 42 5.75 2.25 1 0 , 25, 54 6 .0 0 45B, 55, 74, 75 2 .5 0 8 , 9, 11, 32, 35, 40 6.25 59, 78 2.75 15 6 .5 0 3 .0 0 1 2, 14, 24, 31, 44 6.75 3.25 5 7 .0 0 79, 85, 86 3.50 6,39 7.25 3.75 4,67 7.50 72, 73 4.00 7, 43A, 45A

176 TABLE 22

1965 POPULATION, DENSITY AND HOLDING CAPACITY BY CENSUS TRACTS (DSGM) 1 2 3 1* 5 6 7 8 10 Total Addn. Max. Add. Land Acid. Max. Res. Dlit. Area 1965 1965 1965 Land Land for Res. Pop. Pop. Pop. from in Pop. Land in Res. Avail. Avail. x Density on (Holding Max. HVC CT Acres Res. Use Density- for Res. for Resi.(see fig. Add. Capacity) Pop. (miles! LU LU 1 6) Land

2 336.1 5700 179.1 31.83 3.5 182 .6 3.5 x 24 84 5784 98.5? 4.25 3 365.9 7179 222.8 32.22 1.7 224.5 1 .7 x 24 4l 7220 99 M 4.25 4 1*33 ."3 5162 163.7 31.53 15.0 178.7 1 5 .0 x 26 390 5552 93.0% 3.75 5 551.5 821 27.0 30.40 —— 27.0 ------821 100 % 3.25 6 6l4.8 5630 196.9 28.59 2 5 .8 222.7 2 5 .8 x 27 695 6325 89.0# 3.50 7 1*89.0 7069 267.8 26.40 — 267.8 — --- 70 69 100 % 4.00 8 330.1 3381 103.9 32.54 — 103.9 — — 3381 100 % 2.50 9 2l*l*.l 3513 107.7 32.61 1 .0 108.7 1 .0 X 32 32 3545 99.1% 2.50 10 373.2 1*91*1 181.1 27.28 10.0 191.1 1 0 .0 x 33 330 5271 93.7% 2.25 11 1252.0 1*1*90 165.9 27.06 15.0 180.9 15 x 32 480 4970 90.3% 2.50 12 1055,1 3366 121.9 27.61 --- 121.9 3366 100 % 3.00 13 11*77.1 11395 1*64.2 24.55 22.0 486.2 22 x 23 506 11901 95.1% 4.50 l4 231.2 3706 106.6 34.76 23.4 130 .0 23.4 x 30 702 44o8 84.1# 3.00 15 1*1*0 .8 1*195 112.3 37.35 2 .0 114.3 2 x 31 62 4257 98.5# 2.75 16 — 3l*2.2 6391 202.6 31.54 202.6 _«— 6391 100 % 2 .0 0 17 191.1 1*033 107.7 37.45 — 107.7 4033 100 % 1 .5 0 18 199.1* 1*711 125.5 37.54 0.5 126.0 0.5 x 39 20 4731 99.6# 1.50 19 302.6 31*1*5 95.8 35.96 5.0 100 .8 5 x 37 185 3630 94.92 1.75 20 1*09.3 3799 111.3 34.13 —. 111.3 —— 3799 100 # 2 .0 0 21 326.1 61*70 197.3 32.79 — 197.3 6470 100 # 2 .0 0 22 229.8 5311 119.4 44.48 — 119.4 —• __ 5311 100 # 1.50

23 151.0 1*032 66.7 60.45 — 66.7 — 4032 100 # 1 .0 0 177 24 5ll*.8 7856 227.3 34.56 4.0 231.3 4 x 30 120 7976 98.52 3.00 TABLE 22— Continued

1 2 3 1* 5 6 7 8 9 10 11 12

25 278^3 6213 11*9.2 1*1.61* 11*9.2 —mm 6213 100 % 2.25 26 280.1 61*1*5 159.7 1*0.36 —— 159.7 -- -- 61*1*5 100 % 1.50 27 151*. 3 2291 29.1 78.73 - 29.1 - - 2291 100 % 1.00 28 301*. 6 1787 23.3 76.69 7.1 30.1* 7.1 x 30 213 2000 89. b% 0 .00 29 225.2 31*05 6 6.6 51.13 12.9 79.5 12.9- x 1*6 595 1*000 89.1% 0.75 30 380.3 1*938 120.1 1*1 .1 2 - 120.1 — 1*938 100 % 1 .0 0 31 189.0 21*09 76.1* 31.53 15.0 91.1* 15 x 30 1*50 2859 81*. 3# 3.00 32 196.6 3305 111.3 29.69 O .76 112.06 .76 x 30 22.8 3328 99.3% 2.50 33 201.3 51*28 126.6* 12.88 _ 126 .6 51*28 100 % 1.50 3b 173.1 3878 67.2 57.70 6 7.2 —» 3878 100 % 0.75 35 615.0 2736 97.9 27.91* 1U.1 112.0 llt.l X 30 1*21* 3160 3 6.6% 2 .50 36 327.2 5397 130.5 1*1.36 — 130.5 — 5397 100 % 1.75 37 163.6 2075 1*6.3 1*1*. 82 — 1*6.3 « 2075 100 % 1 .00 38 209.3 1506 35.1 1*2.91 5.0 1*0 .1 5 x 30 150 1656 90,9% 1.25 39 520.7 7278 268.7 27.09 268.7 _ 7278 100 % 3.50 1*0 337.6 1*038 116.9 3l*.5l* — 116.9 -- — 1*038 100 % 2.50 1*1 31*9.6 3090 75.6 1*0.87 — 75.6 —— •— 3090 100 % 1.50 1*2 165.6 3518 92.5 38.03 6.7 99.2 6.7 x 30 203 3721 9b.5% 2.00 1*3A 357.1 1*18 10.3 1*0.58 8 7 .0 97.3 87 x 25 2175 2593 1 6.1 /S lt.00 1*3B 287.0 3657 ll*7.0 2U.87 1 2 .0 159.0 12 x 23 276 3933 93 % lt.50 1*1* 303.3 1*1*78 ll*2.9 31.31* __ 11*2.9 1*1*78 100 % 3.00 1*5A 568.9 391*8 181.3 21.78 lit. 2 195.5 lit. 2 x 25 355 1*303 91.7# lt.00 1*5B 1127.6 7790 1*96.2 15.69 1*2.9 539.1 1*2.9 x 17 730 8520 91.b% 6 .00 51* 1*57.3 5385 11*5.9 36.91 - 11*5.9 -- -- 5385 100 % 2.25 55 1395.9 111*85 1*86.0 23.63 156.9 61*2.9 156 .9x 17 2667 11*152 81.2/S 6.00 56 21*32.2 2798 196.0 lit, 28 21*2 .0 1*38.0 21*2 x 21 5082 7880 35* 5% 5.00 57 310U.9 10065 32U.3 31.01* 258.3 582.6 258.3x 21 5l*2l* 151*89 65 % 5.00 58 151*0.0 6506 363.0 17.92 502.9 865.9 502.9x 19 9555 16061 1*0.5SS 5.50 59 966.8 73l*2 1*1*2 .6 16.58 100.0 51*2.6 100 x 16 1600 89U2 82.15S 6.25 60 370.8 3567 199.1 17.92 26.12 225.2 26.12x 1 9 1*96 1*063 87.8# 5.50 TABLE 22- Continued

1 2 3 1* 5 6 7 8 9 10 11 12

6l 309.2 1*259 166.2 25.63 1*.7 170.9 1*.7 X 21 99 1*358 97.7% 5.00 62 58*1.8 3386 2l*l*.l* 13.85 2 2 .2 266.6 2 2 .2 X 22 1*88 387I* 87.k% 4.75 63 21*1 .1 1*329 ll*1 .2 30.65 0 .6 ll*1 .8 0 .6 X ll* 8 1*337 99.8% 1* .25 6k 210. k 337U 111*. 8 29.39 17. oi* 131.81* 17.0l*x 23 392 3766 89.656 1*. 50 65 33U.6 261*8 109.8 21*. 12 29. H 139.2 29.1* X 22 61*7 3295 80.1*56 U.75 66 799.1 3670 180.2 20.37 207.0 387.2 207 X 23 1*761 81*31 1*3.55? 1*. 50 67 703.3 2721* 108.9 25.01 2 2 .2 131.1 22.2 X 26 577 3301 82.556 3.75 68 1017.2 2516 ll*5.1 17.31* 261* 1*09.1 261* X 17 1*1*81* 7000 35 . 956 5.00 69 523.1 1*292 185.0 23.20 37.3 222.3 37.3 X 19 708 5000 85.8$ 5.50 72 2591.1 6006 509.1 11.79 635.0 111*1*. 1 635 X 11 699k 13000 1*6.256 7.50 73 2782.2 2861 223.7 12.79 500.0 723.7 500 X 11 5500 8361 3l*.2# 7.50 7k 1252.3 3285 198.8 16.52 155.0 353.8 155 X 17 2635 5920 55.5# 6.00 75 1009.6 2851* 259.8 10.99 383.0 61*2.8 383 X 17 6511 9365' 30 . 556 6.00 77 980.3 1*21*1 372.2 11.39 100 .0 1*72.2 100 X 19 1900 6lHl 69.1% 5.50 78 70 ^.8 1*989 301.1* 16.55 86.55 371.k 86.55 X 16 1385 6371* 78.3? 6.25 79 1113.0 7252 526.0 13.78 295.0 821.0 295 X 12 351*0 10792 67.256 7.00 85 101*9.3 31*36 285I* 12.0l* 251.0 536.U 251 X 12 3012 61*1*8 53.3?5 7.00 86 1169.6 1*1*91 l*32l* 10.39 29l*.0 726.1* 29I* X 12 3528 8019 56 % 7.00 179 l8o

TABLE 23

1965 RESIDENTIAL DENSITY BY DISTANCE BANDS

Distance...... Population/ ' (miles) Residential Acre

0 .0 0 7 6.70 1 .0 0 1+5.51 2 .0 0 33.35 3.00 31.29 1+.00 2 6 .83 5.00 19.21 6 .0 0 1 7 .28 7.00 12.17 TABLE 2k

1965 POPULATION AS A PERCENTAGE OF MAXIMUM HOLDING CAPACITY BY DISTANCE BANDS

Distance 1965-Pop./ Max. Holding ' (mil6s)...... Capacity Pap. ' ' •

0 .0 0 . 1787 _ nQ

2000 ~

1.00 • 511^3 _ Qfl , 51908 " 98*5^ 2.00 67022 _ s m = 97- ® 3.00 39739 U1768 = 95.1^

k.OO 58625 _ WI 90 ~ 86%

5.00 kk2lQ ^ 73161 " 60*5^ 6.00 377^5 = 70 Q, 53273 ' 97°

! i i - » • «

Total l § 5 i w = T9.9rt 182

TABLE 25

197k FORECAST OF POPULATION BY DISTANCE BAND AND CENSUS TRACT

% Holding 1974 Dist. Satura- Cap. Forecast Distribution to Census Tracts from tion from ■ from ■ by Dist. (given-in the same sequence HVC fig. l'T '' table 2 4 Band as in'tabia 21) '

0 .0 0 80# 2000 1600 1600 0.75 4000,. 3775 1 .0 0 86# 51908 44679 3195, 229 1, 4890, 1550 1.25 1656 1 -5 0 3670, 3895, 3980, 4695, 4012, 3090 1.75 2849, 5000 2 .0 0 88.3# 68699 60641 5325, 3225, 4850, 3721 2.25 4480, 5675, 5385 2.50 3300, 2925, 3380, 3328, 3160, 4038 2.75 . 3275 3 .0 0 85.6# 41768 35756 2500, 4400, 7380, 360, 4295 3.25 546 3.50 6200, 6800 * 3.75 4822, 1850 4.00 8 3.3# 68190 56774 7025, 1 5 0, 4303 {*•25 5350, 6750, 4330 4 .5 0 9700, 3821, 3766, 4907 4.75 3400, 3295 5 .0 0 79.6# 73161 58260 3525, 14050, 4350, 6879 5.25 5.50 11040, 3300, 4801, 3620 5.75 6 .0 0 74.7# 53273 39787 7220, 13100, 1565, 3502 6 .2 5 9000, 5400 6.50 6.75 7.00 74.5# 46620 34743 9590, 4900, 3500 7.25 7.50 12650, 4103 TABLE 26

197U ACTUAL POPULATION BY DISTANCE BANDS AND CENSUS TRACTS

Census Tract Pop...... •Dist. (given-in the same 197^ 'Pop. ' (Actual) (miles) sequence as table 21)• Max. Holding Cap.

0.00 1955 '^222- 07 8f 2000 “ 9 > •

0 .7 5 3956, 3052- 1.00 3027, 2289, 1*71*6, 1291 1*21*60 1 .2 5 1656 • 51908 = 81*! 1 .5 0 3U9 6, 3717, 386U, 1*530, 3791, 301+5 1.75 2238, 1*1*23- 2.00 5126, 2967, U6ll, 3721 56811* = 82.1% 2.25 1*18U, 5375, 5350 68699 2.50 2986, 26U9, 3073, 3328, 3160, 3623

2.75 3370 3.00 2613, 1*1*0 8, 8056, 360, 36955 1*338 1*1768 = 88.5# 3.25 582 3.50 6325, 6903 3.75 1*722, 17^5 U.00 6933, 63, 1*303 55386 = 8 1.2# 1*.25 5205, 66U8 , 1+290 68190 I*.50 9591, 3721, 3766, 1*399

1*.75 30l*U, 3217 5.00 3011*, 13752, 1*023, 651*3 57Ql*0 5.25 73161 = 7 8.0# 5.50 10659, 3055, 1*51*8, 5185

5.75 6.0 0 8520, 11*152, 1*858, 3955 1*7161 6.25 95l*7, 6129 53^73 *5^ 6.50

6.75 7.00 9729, 1*328, 3611 3l*l*7l* 13.9% 7.50 12l*1 9, 1*387 1*6620 33221*0 Total 81.91% 1*05619 isi*

TABLE 27

DISTRIBUTION OF POPULATION TO CENSUS TRACTS (DSGM)

Fore­ Fore­ Actual cast r . Actual cast ...Pop...... Pap. ... I 97I*.... • Pop.■ ..Pop. .. 1971+ CT' ' -197V ' 1965 Pop. • CT ’ ' 1971+ 1965 ' Pop.

2 5205 5700. . 5350 38 1656 1506 1656 3 66U8 7179 6750 39 6903 7278 6800 1+ 1*722 5162 1*822 1*0 3623 1+038 1+038 5 582 821 5U6 1*1 301*5 3090 3090 6 6325 5630 6200 1*2 3721 3518 3721 7 6933 7069 7025 1+3A 63 1*18 150 8 2986 3381 3300 1+3B 3721 3657 3821 9 261 *1* 3513 2925 1*1* 1+338 1*1+78 1*295 10 1*181 * 1*91*1 1*1+80 1+5A 1*303 391*8 1*303 11 3073 1*1*90 3380 1+5B 8520 7790 7220 12 2613 3366 2500 51* 5350 5385 5385 13 9591 11395 9700 55 ll*152 111*85 13100 Ik 1+1*08 3706 1*1*00 56 3011* 2798 3525 15 3370 1*195 3275 57 13752 10065 11*050 16 5126 6391 5325 . 58 10659 6506 liol+o 17 3U96 1*033 3670 59 95l*7 73l*2 9000 18 3717 1*711 3895 60 3055 3 567 3300 19 2238 31*1*5 28U9 61 1*023 1*350 1*259 20 2967 3799 3225 62 30i*l+ 3386 31*00 21 1*611 61*70 1*850 63 1*290 1*329 1*330 22 3861* 5311 3960 61* 3766 3371+ 3766 23 3027 1*032 3195 65 3217 261*8 2k 3295 8056 7856 7380 66 1*399 3670 1*907 25 5375 6213 5675 67 171*5 2721+ 1850 26 1*530 61*1*5 1*695 68 65U3 2516 27 6879 2289 2291 2291 69 1*5U8 1*292 1*801 28 1955 1787 1600 72 12l*19 6006 12650 29 3956 31*05 1*000 73 1*387 2861 1*103 30 1*71*6 1*938 1*890 71* 1+858 3285 1565 31 360 21*09 360 75 32 3955 285I+ 3502 3328 3305 3328 77 5185 1*21*1 3620 33 3791 51*28 1*012 78 3k 6129 1*989 51*00 3052 3878 3775 79 9729 7252 9590 35 3160 2736 3160 : 85 1*328 31*36 1+900 36 1*1*23 5397 5000 86 3611 1+1*91' •■ 3500 37 1291 2075 1550 3322^*0 APPENDIX C

DATA AND CALCULATIONS FOR THE TOLEDO METHOD

TABLE 28

PLANNING AREA TO CENSUS TRACT (i9 60) EQUIVALENCY TABLE (TOLEDO METHOD)

Planning Area Census Tract (i9 60) 'Numbers

Old Orchard 13, lk9 15, 2h, 31, 67 Nest Toledo A 2,-3, 6, 7 Nest Toledo B 6 0, 6 1, 6 2, 63, 6k, 65 Mayfair V-57 Trilby 58, 59, 79 Talmadge 77, 78 ...... Heatherdowns ^3A, U3B, U5A, k5B, 68p, 6 9, 72 Reynolds Corner 6 6, 7U, 75, 85, 86 Airport Highway 73 Point Place 55, 56p Fort Industry lip, 56p Inner Core Center City 2 8, 3%>, 37p, 38p LaGrange 5, 9, 1 0 , lip, 17, 1 8, 19 North End 12 20 29 30 South. Side 35p, 38p, 39, Uo, Ul, k2 , hk, 5k, 68p Dorr 25, 26, 32, 33, 3kp, 35p, 36, 37p Old Nest End 8 , 1 6, 2 1, 2 2, 2 3, 27

185 186

TABLE 29

DISTRIBUTION OF 197*+ FORECAST POPULATION TO CENSUS TRACTS, TOLEDO METHOD

Raw Final Raw Final 197^ 197^ 1971+ 1971+ 1971+ 191b •Forecast Forecast •Actual Forecast •• Forecast Act. CT • ’Pop. ' ' Pop. Pep. ■ CT' ' Pop. POP. POP.

2 5025 5282 5205 38 1718 I806 1656 3 621+0 6559 66U8 39 7025 7381+ 6903 k 1+71+0 1+982 1+722 1+0 1+000 1+201+ 3623 5 600 631 582 1+1 1+020 1+225 30U5 6 5985 6291 6325 1+2 3820 1+015 3721 7 65bB 6883 6933 1+3A 125 131 63 8 2B96 30U1+ 2986 1+3B 2895 301+3 ' 3721 9 2620 2751+ 261+1+ 1+1+ 5008 5261+ 1+338 10 5020 5277 1+181+ 1+5A 3t+25 3600 1+303 11 2982 313b 3073 1+5B 7680 8072 8520 i2 2502 2630 2613 5U 51+82 5762 5350 13 9700 10196 9591 55 11+200 11+926 11+152 ll+ 1+750 1+993 1+1+08 56 3135 3295 3011+ 15 3U25 3600 3370 57 13800 11+505 13752 16 5200 5U65 5126 58 8950 91+07 10659 17 3500 3679 31+96 59 7663 8055 95*+7 18 3680 3668 3717 60 2825 2969 3055 19 1798 1890 2238 61 3960 1+162 1+023 20 2795 2938 2967 62 2811 2955 301+1+ 21 1+321+ 1+51+5 1+611 63 1+000 1+20U I+290 22 3560 371+2 3861+ 61+ 3530 3710 3766 23 3008 3162 3027 65 3030 3185 3217 2b 8120 8535 8056 66+ 11+21 1+61+7 4399 25 5*+50 5728 5375 67 2356 21+76 17U5 26 1+1+25 1+651 1+530 68 5807 6101+ 65U3 27 1837 1931 2289 69 31+08 3582 I+5I+8 28 1525 1603 1955 72 9851+ 10358 121+91 29 1+020 1+225 3956 73 3187 3350 1+387 30 3183 331+6 1+71+6 71+ 1+869 5118 1+858 31 i+oo 1+20 360 75 1+028 1+231+ 3955 32 3102 3261 3328 77 1+782 5 026 5185 33 380U 3998 3791 78 5283 5553 6129 3*+ 2980 3132 3052 79 7858 8260 9729 35 3182 331+5 3160 85 1+709 1+950 1+328 36 1+700 1+91+0 1+1+23 86 3677 3865 36ll 37 1121 1178 1291 BIBLIOGRAPHY

Akron Metropolitan Area Transportation'Study (AMATS). "Landuse and Development Forecast, 1970-2000, Technical Report." Akron, Ohio: AMATS, Jan., 1976. Association of Central Oklahoma Governments (ACOG). "Transportation Technical Reports, 1971-7^." Oklahoma: ACOG, 197^.

Atchiey,.Robert C.,. Population Projections and Estimates for Local Areas'. Oxford, Ohio: Miami University, 1970.

Boyce, David E.; Cote, Roger W. "Verification of Landuse Forecasting Models." Research Record No. 126. Washington, D.C.: Highway Research Board, 1966.

Boyce, David E.; Day, Norman D.; and McDonald, Chris. Metropolitan Plan Making. Monograph No. Philadelphia: Regional Science Research Institute, 1970. Brand, Daniel. "Theory and Method in Landuse and Travel Forecasting." Research Record No. U22. Washington: Highway Research Board, 1973. pp. 10-20. Chadwick, George. A Systems View of Planning. Oxford, U.K.: Per- gamon Press, 1971.

Chapin, F. Stuart. Urban Landuse Planning. Urbana, 111.: University of Illinois Press, 196 5.

Chicago Area Transportation Study (CATS). "Vol. II Data Projections." Chicago, 111.: CATS, i96 0.

______. "Vol. Ill, Transportation Plan." Chicago, IL: CATS, 1962. City of Austin. "Austin Tomorrow: Landuse, An Analysis of Urban Development in Austin." Austin, TX: City of Austin, June, 1975. Creighton Hamburg Planning Consultants. Data Requirements for Metro­ politan Transportation Planning. NCHRP Report No. 120. Wash­ ington, D.C.: Highway Research Board, 1971.

Echenique, M. Models: A Discussion, Landuse and Built Form Studies, Working Paper No. 6. Cambridge, U.K.:University of Cambridge, March, 1968.

187 *88

Federal Highway Administration, Instructional Memorandum.' 50-^-68: Operations Plan for Continuing Urban Transportation Planning, Washington, D.C.: Federal Highway Administration, May, 1968.

Goldner, William,. -"The Lowry Model -Heritage.11' Journal~of'the American ‘ Institute ' of ' Planners, (March, 19T1): 1QQ-110.

Grecco, W. L.; Wegmana, F. J.; Spencer, J.A.; and Chatter^ee, A. Transportation Planning for Small Urban Areas.' NCHRP'Report N o . 167. Washington, D.C., Transportation Research Board, T97ST:

Hamburg, John R.; and Sharkey, Robert H. "Landuse Forecast." Chi­ cago, 111.: Chicago Area Transportation Study, 1961.

Highway Research Board, Urban Development Models, Proceedings of a Conference, June, 1967, Dartmouth College, Hanover, N.H.' Spe­ cial Report No. '97. Washington, D.C.: Highway Research Board, 19W.

ITE Technical Committee 6f6. "Informational Report, Landuse and Demography: Growth Versus Forecast." Traffic Engineering, March, 1977, pp. U2-M*.

King, Leslie J. "Models of Urban Landuse Development," in Models of Urban Structure, ed., David C. Sweet. Lexington, Mass.: Lexington Books, 1972.

Krueckeberg, Donald A.; and Silvers, Arthur L. Urban Planning Analy­ sis, Methods and Models. New York, N.Y.: John Wiley and Sons, Inc., 1971*. Lee, Douglas B. Models and Techniques for Urban Planning. Buffalo, New York: Cornell Aeronautical Lab., 1968.

..... "Requiem for Large-Scale Models." Journal of the American Institute of Planners. May, 1975, pp. 163-178.

Lowry, Ira S. A Model of a Metropolis. Santa Monica, Ca: The Rand Corporation, 196^.

' "Seven Models of Urban Development: A Structural Com­ parison," in Urban Development Models, Special Report No. 97. Washington, D.C.: Highway Research Board, 1968, pp. 121-l63.

McNulty, Robin C. "An Analysis of the Procedures for Forecasting and Distributing Future Landuse in Small Urban Areas." Unpublished M.C.P. thesis, The University of Tennessee, 197^. 189 Marks, Harold. Protection of Highway Utility, NCHRP Report No. 121. Washington, D.C.: Highway Research Board, 1971. Mid-America Regional Council. (MARC). "Methodology for Landuse, Population and Employment Distribution." Kansas City, Mo.: MARC, 1973. Mid-Ohio Regional Planning Commission (MORPC). "A Technical Report on the Year 2000 Landuse and Trip Generation Variables." Columbus, Ohio: MORPC, 1977.

"Franklin County Regional Transportation Plan." Columbus, Ohio: MORPC, 1972.

Mills, Edwin S.. Studies in the Structure'of the Urban EOOnOmy. Baltimore, Md.: Johns Hopkins Press, 1972.

Mitchell, Robert E.; and Rapkin, Chester. Urban Traffic:''A'Function of LandUse. New York, N.Y.: Columbia University Press, 195^+.

Moore, Will Terry. An Introduction'to Urban Development'Models. Washington, D.C.: U.S. Department of Transportation, October, 1975. Ohio Department of Natural Resources (ODNR). "A Manual Land Capa­ bility Evaluation and Scoring System for Five Development Related Uses." Technical Report No. 6. Columbus, Ohio: ODNR, 1975.

' "Land Capability Analysis." County Report No. 3, Stark County. Columbus, Ohio: ODNR, June, 1975.

Putman, Stephan H. "Urban Landuse and Transportation Models: A State-of-the-Art Summary," Transportation Research. Vol. 9, 1975, pp. 187-202.

Reif, Benjamin, Models in Urban and Regional Planning. New York, N.Y.: Intertext Educational Publishers, 1973.

Richardson, Harry W. Elements of Regional Economics. Baltimore, Md.: Penguin Books, 1969.

Roberts, Paul 0. "Model Systems for Urban Transportation Planning: Where Do We Go From Here?"' Res earch Rec ord‘No. 309. Washington, D.C.: Highway Research Board, 1970, pp. 34-44.

Rockford Area Transportation Study (RATS). "A Dwelling Unit Allocation . Technique for Winnebago County." Rockford, 111.: RATS, December, 1971. 190 Southwestern Pennsylvania Regional Planning Commission (SPRPC). "Forecasting Framework: -Jobs, People and Land." Pitts­ burgh, Pa.: SPRPC, 197^. Stark County Area Transportation Study (SCATS). "Population and Economic Study, Small Unit Forecast." Canton, Ohio: SCATS, 19771 Swerdloff, Carl N.; and Stowers, Joseph R. "A Test of Some First ■ Generation Residential Landuse Models." ResCarCh ReCCrd No. ' 126. Washington, D.C.: Highway Research Board, 1966, pp. 38- 59.

Toledo Regional Area Plan for Action (TRAPA). "1965 Regional Landuse Inventory/Analysis and 1985 Forecasts." Toledo, Ohio: TRAPA, 1973.

U. S. Department of Transportation. Federal Highway Administration, Urban Transportation Planning, Policy'and Procedure Memorandum 50-9. Transmittal 162, Washington, D.C.: Government Printing Office, November, 1962.

______. General Information and Introduction to System 360. Washington, D.C.: Government Printing Office, March, 1965.

_____ . Computer Programs for Urban Transportation Planning, General Information. Washington, D.C.: Government Printing Office, 1977.

Vogt, Ivers and Associates. Landuse: Ohio Procedure Manual No. 3. Cincinnati, Ohio: Vogt, Ivers and Associates, 1965.