GIS 4930 - Providence Food Deserts

After completing an analysis of Providence food deserts, I have produced a web map as well as a presentation document to communicate my results.

GIS 4930 - Further Adventures in Web Mapping

After completing the practice work with maps for Pensacola, this module moves on to doing similar work for my own particular study area - Providence, Rhode Island.

After creating an analysis of food deserts within the municipal boundaries of Providence, I created web tiles that showed the census tracts for Providence that I had determined were food deserts as well as another tileset for grocery stores in the area. I used the site mapbox.com to store tiles I created using Tilemill. This proved to be especially challenging as mapbox has undergone some updates recently and the instructions I had were out of date. However, I perservered and was able to get my data up and available. I did have to find a new set of code examples for my map page, though, before I could serve my data online. I'm quite excited at what these tools offer, although it would be good if APIs were less subject to radical change.

The image below is a simple screenshot showing my tiles appearing in a web browser.

Screenshot of Providence food deserts

GIS 4930 - Open Source Web Mapping

This week was involved in creating a web map using open source resources. Leaflet, a java mapping API, was used to create a live webmap stored on my UWF web drive. As this was practice to learn a bit about how to use leaflet with mapping tiles stored on mapbox, we were provided with extant tilesets and simply had to create an html file that would be able to load them properly as well as some bits of map functionality through plugin calls.

The map created shows Pensacola, FL and includes indicators for food deserts as well as for fictional (I believe) frisbee golf courses.

GIS 4930 - Open Source GIS - Food Desert Analysis

This module invovled using QGIS to create a map displaying food deserts in Pensacola, Florida as a preparation for creating a similar analysis for another region in the coming weeks. We were given the data that was used to process the map below, which displays food deserts and food oases in Pensacola, FL based on distance from grocery stores. Using US census tracts, a tract was determined to be 'desertified' if it's centroid was over a mile away from a grocery store.

Food Deserts in Pensacola, FL

GIS 4930 - Standard Residuals of Meth Lab Density Model

During the analysis phase of this module, a large number of potential explanatory variables were assessed to explain the density of meth labs found in and around the Charleston area. Most of them were gleaned from the US Census data for 2000 and 2010. The map below shows the standard deviation residuals for a linear regression model that includes variables for population density by census tract, percentage of households with a single male head of household and one or more children, and the ratio of males to females in the tract. The adjusted R-Squared value for this model is 0.572. The residuals below indicate that some tracts actual density did not match the model's predictions very well (the red ones were too high and the green, too low), but the distribution does not seem to be terribly biased, which is also confirmed by the Jarque-Bara result of 5.36.

Standard Residuals for Methamphetamine Density Model for census tracts in and around Charleston, WV

GIS 4930 - Statistical Analysis of Meth Lab Locations

We're starting a new project this week that involves doing statistical analysis of meth lab locations as an attempt to find correlations with population, geography and economic factors that may aid in predicting trouble spots for meth preparation.

The map below is a basic map of the study area showing methamphetamine labs that were interdicted over the course of 2004-2008 in and around Charleston, West Virginia. The study area layer has attributes for census tracts that will be correlated with the location of the labs. Preparations were made this week to create further attribute fields for use in the analysis stage of this project.

Meth Labs in Charleston, WV (2004-2008)

GIS 4930 - Completed MTR Story Map

The work of the past two weeks involved coordinating amongst my group as we all took a LandSat image from 2010 that covers a portion of our study area to analyze for evidence of mountain top removal. The analysis was fairly basic, and in all likelihood did not produce a robust set of MTR location data. See the previous entry for details of the analysis. I ended up doing my analysis over and got better, but still iffy, results.

Each of our team produced a layer package with our analysis as a polygon layer. I merged these into one unit and produced a calculation for acres covered by our MTR zones as well as an estimate of its accuracy, which was based on only a sparse number of randomly selected points. The sparsity was an artifact of the time it would take to do a thorough accuracy assessment.

Once the layers were merged, they were shared back with the entire group and we used them in our final story map.

Two layers [MTR zones & drainage] sans context - as they were uploaded to arcgis online.