A Geospatial Analysis of the G3 Program

A geospatial analysis of the environmental restoration outcomes from the Chesapeake Bay Trust's Green Streets, Green Jobs, Green Towns (G3) program was conducted. The goal of this project was to 1) determine where future projects should be awarded that will have the greatest impact on environmental health hazard mitigation and 2) create web maps that visualize the G3 data and environmental justice indices at the county level. 22 restoration metrics from 267 awarded projects were aggregated and mapped using ArcGIS Pro. 6 EPA EJScreen indices and the CDC Environmental Justice Index for each Chesapeake Bay watershed county were compared to µ, a parameter representing each county's total awarded amount divided by the county's population. It was determined that Baltimore City (MD), Prince George's county (MD), and Richmond City (VA) were among the most vulnerable counties (as determined by environmental justice indices) that have been well-funded by the Trust. Dorchester (MD), Hampshire (WV), and Kent (MD) had the highest µ values (23.2, 14.3, and 13.7 respectively), i.e. the Trust has awarded the highest dollar amount per person in these counties. Based on the environmental justice indices examined in this analysis, it is recommended that the Trust fund future projects in other populous and vulnerable areas such as Petersburg City (VA), Washington, D.C., Northampton (VA), and Hopewell City (VA).

The Chesapeake Bay Trust's  Green Streets, Green Jobs, Green Towns (G3) grant program  has funded numerous environmental restoration projects across the Chesapeake watershed. These projects have had both environmental and human health benefits. States within the Chesapeake Bay Watershed include Maryland, Virginia, Delaware, Pennsylvania, West Virginia, and New York; Washington, D.C. is also included. Since the program's inception in 2010, 274 projects have been completed; 233 of these projects are currently plotted on the  Trust's Awarded Projects Map  (figure 1).

One can filter the projects to show the approximate locations of the G3 awarded projects. By looking at the point locations of the G3 projects alone, however, one cannot draw any conclusions about the tangible impacts the program has made thus far. In this research project, the following questions regarding the G3 program were explored:

1. What are the cumulative environmental restoration impacts by county?
2. Which counties would most benefit from future G3 program interventions?
3. What is the total dollar amount awarded per county?
4. What is the ratio of total awarded dollars to population per county?
5. How can GIS mapping be used to better visualize the impacts of the G3 program?

Not all these questions necessarily require geospatial tools (i.e. ArGIS Pro) for analysis. However, maps are intuitive data visualization tools and can help communicate research results. Therefore, this project also involved creating an interactive web map to embed in this Storymap.

Data was pulled from multiple sources. G3 awarded project data (e.g. restoration metrics, award amounts, and project latitude/longitude ^* ) was queried from the Chesapeake Bay Trust's internal Blackbaud Grantmaking database. To answer research question (2) above, two geospatial datasets were used: EPA's  EJScreen  indices and the CDC's  Environmental Justice Index  (EJI).  US County data  polliwas sourced from ESRI. Extensive data hygiene and (geo)processing were conducted in order to ensure that these disparate data sets could be used in conjunction with one another.

 _{*It is worth noting that not every project's impact is best represented by a singular latitude/longitude data point. For the purposes of this analysis, however, the latitude/longitude point data approximates the location of each project's intervention.} 

267 G3 projects were included in the final data set. Projects that were closed after approval were not included in the final dataset. 214 of these projects have already been completed; 53 projects are approved and in progress. Awardees are required to provide quantitative estimates of their projects' impacts as well as, ultimately, quantitative actuals (i.e. key performance indicators (KPIs)). For the purposes of this analysis, the 53 in progress projects' estimates were treated as actuals. Geoprocessing in ArcGIS Pro was used to assign a county to each of the 267 projects.

An Excel pivot table was created to summarize the KPIs at the county level. A subset of the G3 program's metrics/KPIs—22 in total—were chosen for this analysis: banks stabilized (linear feet), living shoreline created (linear feet), number of marsh grasses planted, number of native plants installed, number of pollinator-friendly plants installed, number of rain barrels installed, number of rain cisterns installed, number of trees planted, trash debris removed (lbs.), bay grasses / subaquatic vegetation planted (ft ² ), impervious surface removed (ft ² ), impervious surface treated (ft ² ), pervious surface installed (ft ² ), invasive species removed (ft ² ), nontidal wetlands created (ft ² ), area reforested or afforested (ft ² ), streamside forest buffers planted (ft ² ), wetlands enhanced (ft ² ), rain gardens / bioretentions created (ft ² ), total Nitrogen pollution reduced (lbs./year), total Phosphorous pollution reduced (lbs./year), and total suspended solids pollution reduced (tons/year). These 22 metrics were chosen since they were the most direct environmental restoration metrics.

EJScreen provides environmental hazard data at the census tract level. 6 out of 13 of the EJScreen indices were chosen for this analysis: PM2.5, Diesel PM, Ozone, Air Toxics Cancer Risk, Air Toxics Respiratory Hazard, and Wastewater Discharge. These 6 indices were most relevant to G3 program's environmental restoration impacts. ArcGIS Pro was used to aggregate census tract level data to the county level; each county-level index was an average of its constituent census tracts' indices (for each individual EJScreen index).

EJI provides environmental hazard data at the census tract level. The overall EJI value ranges from 0-1.0; a value of 0 denotes no environmental injustice while 1.0 indicates the highest possible value of environmental injustice. ArcGIS Pro was used to aggregate census tract level data to the county level; each county-level index was an average of its constituent census tracts' indices.

ArcGIS Pro was used to plot all data layers. 267 G3 projects were plotted as point data. 22 layers were created to show the geospatial distribution of the G3 projects' restoration impacts. EJI and 6 EJScreen layers were plotted, both at the county level. ArcGIS Pro was used to calculate the field µ = Awarded total / county population for each county.

The ArcGIS Pro map was published as a web map to ArcGIS Online.

The 22 environmental restoration KPIs were aggregated at the county level to determine quantitative impact totals (figure 2). The G3 program has awarded projects in 53 out of the 203 (26.1%) counties within the Chesapeake Bay Watershed.

The EJScreen indices and the EJI were sorted descending. µ was calculated for each county by dividing the total amount awarded to that county by the county population; this value represents how much money has been spent per person. µ was then sorted descending. Each environmental justice index was then compared to µ. The comparison tables follow below (figures 3-9). Only the top 20 counties were examined for these comparisons.

EJScreen PM2.5 index. 4 counties overlapped with µ: Baltimore City (MD), Prince George's (MD), Richmond City (VA), and Lancaster (PA).

EJScreen Diesel PM index. 3 counties overlapped with µ: Prince George's (MD), Baltimore City (MD), and Richmond City (VA).

EJScreen Ozone index. 4 counties overlapped with µ: Baltimore City (MD), Prince George's (MD), Kent (MD), and Dorchester (MD).

EJScreen Air Toxics Cancer Risk index. 3 counties overlapped with µ: Richmond City (VA), Prince George's (MD), and Baltimore City (MD).

EJScreen Air Toxics Respiratory Hazard index. 3 counties overlapped with µ: Prince George's (MD), Baltimore City (MD), and Richmond City (VA).

EJScreen Wastewater Discharge index. 1 county overlapped with µ: Sussex (DE).

CDC Environmental Justice Index. 3 counties overlapped with µ: Petersburg (VA), Baltimore City (MD), and Morgan (WV).

GIS mapping provides an intuitive means of visualizing data. Research questions (1), (3), and (4) could be answered without ArcGIS Pro, but in this project ArcGIS Pro was useful for processing the Grantmaking data in order to answer these questions. Research question (5) was addressed by creating an interactive web map that allows users to browse through the data layers by themselves. The environmental restoration web map could be used as a quick reference tool; currently, there is no easy way to get county-level data on the G3 program's restoration metrics out of the Grantmaking software. The environmental restoration web map—particularly the Awarded Total layer—also provides transparency into the G3 program, which is important for an organization working to increase wellbeing/health for the general public and the environment. 

The environmental justice indices web map provided a means of visualizing where G3 projects have been conducted in relation to regional environmental hazard levels. From inspecting these maps, it is evident that hazards such as diesel PM and respiratory air toxics are most prevalent in urban areas. Wastewater discharge poses the greatest risk in waterfront counties. Ozone risk is significantly higher in the eastern half of the watershed and around the Chesapeake Bay, as opposed to the western half. PM2.5 shows a striking pattern of elevated pollution throughout central PA, the Washington D.C. and Baltimore metropolitan areas, and further south in the watershed. The CDC EJI shows that rural areas are generally at elevated risk overall and therefore should not be ignored, even if those counties have smaller populations. The map also shows that most G3 projects are clustering in urban areas. These conclusions make intuitive sense from looking at the data layers; the web map is a communication tool that can drive home these conclusions in a more concrete way than simple two-dimensional data tables.

GIS mapping as a data visualization tool does have its limitations. The data layers created for this project are high-level abstractions of the G3 project work; looking at the data at the watershed scale cannot elucidate what the G3 projects look like on the ground. Nonetheless, these restoration outcome data layers provide a snapshot of what the G3 program has achieved thus far.

Answering research question (2) required the usage of external EJ GIS data sets. The comparison tables show that the G3 program has been able to reach some (usually 3-4) of the most vulnerable counties for each EJScreen metric examined. Only the Wastewater Discharge index had 1 county overlapping with the top 20 counties reached by G3 (i.e. µ). Baltimore City (MD), Prince George's County (MD), and Richmond City (VA) were repeatedly among the top 20 most vulnerable for the 6 metrics examined and were also within the top 20 counties reached by G3. These areas are quite populous, so it makes sense that they are being targeted—and they should be targeted. However, other cities such as Petersburg City (VA), Washington, D.C., Northampton (VA), and Hopewell City (VA) were often among the most vulnerable but were not in the set µ. These counties/cities are thus examples of where G3 could award more projects in the future. On a per-person basis, G3 has been very successful at awarding money in counties such as Dorchester (MD), Hampshire (WV), Kent (MD), Morgan (WV), and Baltimore City (MD).

In an ideal world, it would make sense to target all of the most vulnerable counties for all 6 EJ indices; however, not all of these counties are populous, and reach is an important factor for the G3 program. There are also many variables that determine where projects take place—funding sources, external and internal organizational capacities, and other priorities. Nonetheless, the G3 program has already made a significant, tangible difference for environmental restoration in the Chesapeake Bay (figure 2).

The G3 program has been a boon for environmental restoration within the Chesapeake Bay watershed. Almost $15 million has been awarded to date. 53 out of 203 counties in the watershed have been reached. Plotting the environmental restoration metrics at the county level and visualizing the data using ArcGIS Online's web maps made the restoration data accessible and interactive. The environmental justice indices used enabled the visualization of the locations of G3 projects in relation to various environmental health hazards. These indices ultimately elucidated which counties are most vulnerable, and these counties were compared with the counties where the Trust has awarded projects. The Trust has many options for where it can award future G3 projects, but projects are not awarded solely on the basis of where environmental hazard is the greatest (e.g. other variables such as population and organizational capacities need to be taken into account). In summary, the G3 project has had tangible impacts on environmental restoration in the Chesapeake Bay watershed, and web mapping is a powerful data visualization tool to facilitate data exploration and foster program transparency. 

 _{Thank you to Sadie Drescher, Nguyen Le, and Whitney Vong for their input on this project. Many thanks to everyone else at the Trust for their support as well! This work was also made possible by UMBC's Shriver Center's Maryland Public Service Scholars program / Sondheim Nonprofit Leadership program.}