Recent Submissions

  • Testing Park Cool Island GIS Analysis Methods For Use In Semi-Urban Conservation Planning

    Luna, Marcos (2020-01-01)
    The human and environmental impacts of urban heat islands (UHI) have become an increasingly relevant issue to city planners. This topic has spurred research into the relationships between land cover, ambient temperature, and the role of greenspace in emitting cooler air to its surrounding area, now known as the ‘park cool island’ effect (PCI). While ample research has been given to this phenomenon in dense urban areas, much less has been dedicated to semi-urban communities who may wish to inform their development practices as they expand their footprints. This research used satellite-derived Landsat Level-2 Provisional Surface Temperature data, MassGIS 2016 Land Cover / Land Use data, and MassGIS Standardized Assessors’ Parcels data to analyze parcels in Essex County, Massachusetts, for PCI intensity and the influence of land cover and parcel characteristics on PCI. LST data from July, 2016, were used to evaluate the mean temperature difference between parcels and their surrounding area to derive PCI. Replicating methods demonstrated by Cao et al. [Landscape and Urban Planning, 96(4):224-231 (2010)], linear regression analyses were undertaken to determine the relationships between PCI, parcel land cover and geometry. The 500 meter buffer distance used by Cao et al. to calculate PCI was also analyzed. Twenty iterations of the linear regression model were run based on a changing buffer value to calculate PCI. Two sensitivity analyses were performed based on these model iterations: 1) change in model performance, as expressed by its R2 value, across the range of PCI buffer distances and 2) the change in the coefficient strengths of the independent variables across the range of PCI buffer distances. The linear regression model underperformed as compared to Cao et al.’s study, however, it affirmed the 500 meter buffer distance as a parameter for calculating PCI, with that model iteration returning the highest R2 value (0.587). Buffer distances greater than 500 meters performed relatively well, however, smaller buffer values were associated with weak model performance. Among land cover coefficients, there were scale-sensitivities observed, with some variables changing in strength and polarity across the model iterations. It was determined that PCI could effectively evaluate cooling intensity in the study area, however, using it as a dependent variable within a linear regression model had only moderate performance. This was due to heterogeneity among the makeup of land cover within parcel buffer areas which inhibited the regression model’s ability to build consistent relationships between land cover and PCI.
  • Improving Compactness Measures For Political Districts

    Luna, Marcos; Ratner, Keith; Strohschein, David (2019-05-01)
    Political redistricting plans often need to consider the compactness of the district’s shape. For states requiring districts to be compact, there is a need to quantify compactness. Existing measures of compactness unfairly penalize districts with coastlines and islands or whose geography itself is not compact. By incorporating information about the underlying geography into the calculation of a modified compactness score, it would be possible to use a compactness test more effectively and fairly across all districts. Several methods of incorporating such data were explored with test districts. A Python script was created to apply the calculations behind the selected method to any polygon shapefile. The script was run on the 436 districts of the 114th Congress of the United States to consider and analyze the modified compactness calculation and its usefulness. Scores for districts covering areas with a significant amount of water were improved when the modified compactness calculation was applied.
  • Which Urban Residents Vote and Why? A Geospatial Analysis of Voting Behavior in Worcester, MA

    Ratner, Keith (2018-08-01)
    This study investigates the relationship between voter travel distance to polling places in Worcester, MA and voter turnout. Linear and geographically-weighted regression are used to evaluate the significance of travel distance and demographic control variables. Worcester appears to be unique when compared to previous studies investigating travel distance and voter turnout. Travel distance to polling place does not reliably predict voter turnout in Worcester, but vehicle ownership, race, and age do.