1 minute read
PREDICTION OF 2020 DEVELOPMENT DEMAND IN DELAWARE VALLEY REGION
2. Data
The study adopted a quantitative approach, along with regression analysis to predict future development.
Advertisement
2.1.Land Cover Change as Development Indicators
The change of land cover from 2001 to 2011 indicates major developments in the decade.
1. Background
Delaware, USA
May. 2020 Group Work with Haitian Wang CPLN675
Mentor: Ken Steif
The DVR as core urban clusters of the Greater Philadelphia Region connects five adjacent states. While population grew from 5.1 million to 5.6 million, the developed area expanded by 1.5 times, signaling significant urban sprawl. The DVR region is forecasted to receive over 658,000 residents and more than 372,000 job opportunities from 2015 to 2045. (DVRPC, 2017) The region will face severe environmental damage if urban clusters continue sprawling at the preceding speed. Therefore, a projection study of future development demand will help observe development patterns and strategically alleviate sprawl through formulation of policy instruments.
2.2 Population Growth as a major Development Engines
Population is the foremost driving factor in development since an increase in population will bring about growth of density and employment. Intensive population growth has been observed in the north of Philadelphia.
2.1.3 Road network as a major Development Engines
In view of the high level of auto-dependency and consequently highway-oriented development patterns, our assumption is that development demand is likely to occur near highways to access better connection to destinations.
3. Method
To achieve a level of accuracy, the study used the roughly one-acre grids as units of analysis. By spatial join data including land cover change, population distribution and distance to highway to the grids of both 2000 and 2010, we used logic regression analysis to study the development patterns. Based on the hypothesis that future development demand shares continuity of current development, the patterns learned from the previous decade would serve as the model of development demands in the upcoming decades.
4. Exploratory Analysis
The spatial lag (i.e. the calculation of spatial weighted average) to 2001 development illustrates development probabilities. Higher chance of development occurs in the periphery of the region and close to certain sections of the highway networks. Land cover change also indicated positive correlation with population change from 2000 to 2001. Among different types of land cover, farmland has the highest conversion rate, follow by developed land and forests, at 1.16%, 1.07%, and 0.67% respectively.
4.1 2010 Prediction and evaluation
2010 prediction was based on analysis of six logic regression models. As observed, most developments have been predicted with probabilities of lower than 20%. The model failed to provide robust results. For example, 5% threshold correctly predicts a lower number of new development areas (Sensitivity) whereas incorrectly predicts a higher number of not changed areas (Specificity). Similarly, all counties demonstrate higher numbers of incorrect 5 predictions of not
Land Use and Envirnmental Modeling
ARC GIS + HECGEOHMS