1 Statewide Land-Use Allocation Model for Florida Stephen Lawe, John Lobb & Kevin Hathaway Resource Systems Group
2 Presentation Outline 1.Land Use Allocation Model (LUAM) Implementation & Overview 2.Data on Land Use Patterns Challenges of Estimation 3. Model Structure & Calibration 4.Results - Sensitivity to Accessibility
3 LUAM Process – Turnpike Integrated Model Structure Truck Trip Table Land Use Allocation Model (LUAM) Auto Trip Table Auto Toll Trips Auto Toll- Free Trips Truck Toll Trips Truck Toll- Free Trips Network Volumes and Travel Times Described in this presentation As described in a presentation by Songer, et al. on Wednesday and Adler, et al. on Thursday 5-year lag
4 LUAM Process – Model Fundamentals Written in C++ and integrated with transportation models: Currently integrated into Florida’s Turnpike Enterprise State Model Built to integrate into MPO models Allocates housing and employment growth at TAZ-level Growth total inputs at user- defined geography State, County, MPO, other. LUAM Model runs in 2-4 minutes St Petersburg
5 LUAM Process – Model Development Plan Goal: Incremental improvements while always having a model that supports analysis 4-step process 1. Estimate and calibrate model using aggregate town & county trends 2. Re-estimate statewide model with parcel-level data from selected counties 3. Complete statewide model estimation with parcel-level data for all counties in state 4. Continue to refine input data and policy control hooks
6 Land Use Data – Overview of Process Legal & Physical Constraints Total Land Available Generate Travel Impedance Matrix TAZ Land Use Forecasts 5 Year Dynamic Land Use Lag Florida - Calibrated Parameters Determine Remaining Developable Land Accessibility and Overall Attractiveness Forecasting Analysis Procedures Base Year Zonal Land Use Households Employment Future Growth Increment Allocate new Increment of Land Use Households Employment Households Employment
7 Land Use Data – Development History (parcel level) Cape Coral Fort Myers
8 Land Use Data – Lumpy by Year
9 Land Use Data – Spatial Variability (parcel variability)
10 Land Use Data – Urban Growth Boundaries (legal constraints)
11 Model Structure – Parameters of Model VariableSourceEffect on Development AccessibilityTravel ModelNon-Linear, generally positive Proximity to CoastGISPositive Distance from Arterials & InterchangesGIS/NetworkNegative Density of Current UseParcelPositive Mix of Land-UsesParcelHomogeneity is positive Urban Growth BoundaryGIS layerLess development outside UGB Undeveloped AreaParcel Positive Res. & Non-Res. Development HistoryParcel(dependent variable)
12 Model Structure – Two Stage Logit / Linear 1. What percent of the available land is consumed? Logit Model Estimates: P ij = Probability of TAZ i, land use j, being developed e P ij = Pr(Y ij = 1 | X ij ) = ἀ + β 1 x 1 + β 2 x 2 +…+ β k x k 1 + e Density ij = β 1 x 1 + β 2 x 2 +…+ β k x k + ε 2. How many houses & emp. will build on the consumed land? Tobit Model Estimates: Density ij = amount of land use j in TAZ i
13 Model Structure – Calibration Results (Residential Growth ) Pearson’s Correlation TAZ=.58 ZIP =.81 Observed Growth Modeled Growth
14 Statewide Application – 2015 households Run model and look for outliers Run without controlling and compare to county control totals Correlation of.93 with 2015 medium BEBR growth totals Bureau of Economic and Business Research
15 Model Sensitivity Test - Accessibility Approximate Population to Employment Ratio Cape Coral: 3 pop to 1 emp Fort Myers: 1 pop to 2 emp Doubling of bridge capacity during 1990s Large subsequent observed increase in development in Cape Coral Population increase by 75% between ’95 – ’05 Twice the rate of growth in ’80s and ’90s Modeled removal of new bridge capacity Population increase by 55% between ’95 – ’05 (as compared to 75% with the bridges)
16 Summary and Conclusions FL TSM v1 completed in 2006, has been used for major planning applications (Doherty, Fennessy & Songer, Wednesday presentation – Session 17) Land-use model built on a high-quality, maintained statewide GIS database Simplified initial model structure supports focus on critical data quality and maintenance issues Work is continuing on estimating one set of land-use models for the State Statewide Land Use Model creates growth forecasts in a consistent manner which is important to the Florida’s Turnpike Enterprise