Figure 1: Left: Green Space Change from 1985-2020. Right: Median Household Income by Zip CodeFigure 1: Left: Green Space Change from 1985-2020. Right: Median Household Income by Zip Code

Figure 1: Left: Green Space Change from 1985-2020. Right: Median Household Income by Zip Code

Introduction

New York City, traditionally referred to as the concrete jungle, is a city known for its iconic skyline, urban geography, and diverse population. With some of the most concentrated ethnic and socio-economic diversity in the world, New York City is a prime example for the ways that development and public health disparities follow closely along the lines of income disparity and racial stratification. According to World Population Review, New York City has a population density of roughly 27,000 people per square mile, demonstrating the ways that the urban landscape has choked out much of the green space that once covered the city’s five boroughs (World Population Review 2022). Extensive studies have been conducted regarding the positive effects of green space provision on public health, both mental and physical, and New York City urban planners have long sought out new ways to bring green space to the city’s population (Kuo 2015). Although, while green space provision has garnered more support in city planning initiatives, the benefits of these initiatives have not been equally born (Hu & Schweber 2020). With green space provision being a primary environmental justice concern as cities continue to develop, analysis regarding equitable green space provision is necessary to ensure public health is improved as the climate crisis worsens.

Figure 2: NDVI Change in relation to Median Income
Figure 2: NDVI Change in relation to Median Income

Data and Research Question

Utilizing tools provided by Google Earth Engine, in this study I was able to derive a Mean Normalized Difference Vegetation Index (NDVI) value for each New York City Zip Code, as delineated by the United States Census, from the years 1985 and 2020 (Shepardson 2022, Figure 1). Ordered on a scale of -1 to 1, NDVI is used by geographers to visualize and quantify the amount of land that is covered by vegetation in a given geometry (GISGeography 2021). From these values, I derived an “NDVI Difference” Value, which captured the difference between each zip code’s NDVI value from 1985 and 2020. Using median household income data, also organized by zip code from the US Census, I set out to draw a relationship between change in NDVI value from 1985 to 2020 and 2019 median household income by zip code. Greenspace provision has long been an environmental justice issue in urban environments, New York City chief among them. Using these merged datasets, I set out to draw a causal relationship between green space change and median income, with zip code being the unit of observation. In Figure 1, areas that are more blue have higher median household income in the image on the right, while on the left areas that are green have experienced green space gain and areas that are purple have experienced green space loss (Figure 1).

Figure 3: Green space Change in relation to Median Income
Figure 3: Green space Change in relation to Median Income

Analysis & Conclusions

Any municipal investment, particularly green space investment which is so intrinsically tied to public health, by local governments has long been tied to the median income of the local populations benefitting most from these investments. As is demonstrated in Figure 2, there is a clear linear correlation between more-positive change in mean NDVI by zip code as Median Income increases, showcasing the disparities in green space investment across New York City. While there is significant clustering occurring at the lower income values, there are clear positive outliers at the higher income values, showing just how much green space gain has occurred in the wealthiest areas of the city (Figure 2). When grouping zip codes by their relation to the median city income, which falls at roughly $70,000 in 2019, these disparities are made even more apparent. As is demonstrated in Figure 3, zip codes with above citywide median household income experienced marginal green space gain, while those that were below citywide median household income experienced green space loss at a rate 15x greater than their counterpart’s comparitive gain. This analysis helps to highlight the significant inequities in New York City green space provision, and highlights the need for environmental justice initiatives to be implemented across New York City to address these issues.

In this analysis, there are two possible discernible outcomes that are supported by the data presented here. Because of limits to the data readily available, the median income data that was utilized for this analysis was data as of 2019, whereas the NDVI data was a variable highlighting change from 1985-2020. Because the median income data was only a snapshot of 2019, we are not able to analyze if median income change was linked to NDVI change. As such, there are two possible causal relationships that are able to be derived. First, which was the study’s intended outcome, is that zip code’s with above-city-median median household income experienced, on average, green space gain, whereas zip-codes with below-city-median median household income experienced green space loss. When a linear regression was performed, the effect of median income on NDVI change was shown to be statistically significant at the 0% level, showing the relationship between these two variables. This indicates the heightened likelihood of the New York City municipal government to invest in areas with higher median household income with regard to green space provision, and highlights the disparities in green space access between high-income and low-income neighborhoods. Conversely, the second conclusion that could be derived, but one that can not be fully supported due to the limits of the data, suggests that investment in green space may have raised median household income in certain neighborhoods. Had a variable been available that would’ve allowed a correlation to be drawn, whether positive or negative, between median household income change and NDVI change, a more conclusive analysis with fewer omitted variables would have been possible. This study does not control for income change as a result of greenspace investment. This is a story more of gentrification and the effects of city investment on the whitewashing of traditionally low-income and majority BIPOC neighborhoods, which have long experienced heavy income inequality. In a study conducted by Stacey Sutton in 2018, the legacy effects of gentrification and racial income disparity in New York City was observed, showcasing the ways that the city’s neighborhoods have started to change demographically as city investment has been shifted (Sutton 2018). With income data from 1985, this study could have been improved by drawing a causal relationship with median income change and NDVI change, but as this data was not available, this conclusion could not be drawn.

While these limits to the study are present, the conclusion remains the same, which is that green space and higher household income are intrinsically linked, and the provision of green space is a key issue for agitators and advocates for environmental justice to prioritize in urban settings. As such, the causal relationship shown by this study indicates that the zip code’s with the highest current median income have also experienced the highest amount of green space change, while those with the lowest current median income have experienced green space loss, indicating the disparities in city investment in green space maintenance from 1985 to 2020. With green space looming large as a public health necessity as summers grow hotter and the New York City’s heat-island effect worsens (Nunez & Dobins 2021), these inequities in green space provision must be addressed in order to ensure equitable access to these spaces is ensured.

Works Cited

Closson, Troy. “The Disparities in Access to New York’s Parks.” The New York Times, The New York Times, 27 May 2021, https://www.nytimes.com/2021/05/27/nyregion/parks-access-nyc.html.

GISGeography. “What Is NDVI (Normalized Difference Vegetation Index)?” GIS Geography, 29 Oct. 2021, https://gisgeography.com/NDVI-normalized-difference-vegetation-index/.

Handy, Ben. “Green Gentrification in New York City.” JCRED, Journal of Civil Rights and Economic Development, 13 Oct. 2020, https://www.jcred.org/shortreads/green-gentrification-in-new-york-city.

Hu, Winnie, and Nate Schweber. “New York City Has 2,300 Parks. but Poor Neighborhoods Lose out.” The New York Times, The New York Times, 15 July 2020, https://www.nytimes.com/2020/07/15/nyregion/nyc-parks-access-governors-island.html.

Kuo, Ming. “How Might Contact with Nature Promote Human Health? Promising Mechanisms and a Possible Central Pathway.” Frontiers in Psychology, vol. 6, 2015, https://doi.org/10.3389/fpsyg.2015.01093.

“New York City, New York Population 2022.” New York City, New York Population 2022 (Demographics, Maps, Graphs), World Population Review, https://worldpopulationreview.com/us-cities/new-york-city-ny-population.

Nunez, Fabiola, and Chris Dobens. “Report Reveals Inequities in the Exposure to the Urban Heat Island in New York City.” WE ACT for Environmental Justice, 20 July 2021, https://www.weact.org/2021/07/report-reveals-inequities-in-the-exposure-to-the-urban-heat-island-in-new-york-city/.

Nyxt. “NYC Environmental Protection: Advocating for Green Spaces.” NYXT, 1 May 2020, https://www.nyxt.nyc/blog/green-spaces/.

Shepardson, Aidan. “Disparities in New York City Green Space.” Disparities in New York City Green Space, 2022, https://ashepardson.users.earthengine.app/view/disparities-in-new-york-city-green-space.

Sutton, Stacey. “Gentrification and the Increasing Significance of Racial Transition in New York City 1970–2010.” Urban Affairs Review, vol. 56, no. 1, 2018, pp. 65–95., https://doi.org/10.1177/1078087418771224.

Appendix

Data Description

  1. US Census Bureau Household Median Income by Zip Code (Source: U.S. Census Bureau, American Community Survey 1-Year Estimates, Summary Table B19001 (2005-2019); retrieved from https://data.census.gov/.) This dataset contains the median household income by New York City zip code in 2019. This dataset initally contains 6068 observations, but filtering reduces it to 181 observations.

  2. Mean NDVI Value (Source: https://ashepardson.users.earthengine.app/view/disparities-in-new-york-city-green-space) Using Google Earth Engine, a platform I used in a previous class at Middlebury College (GEOG0150 with Jeff Howarth), I was able to derive a mean Normalized Difference Vegetation Index (NDVI) value for each zip code within the New York City limits. By using Landsat 5 and Landsat 8 satellite imagery, datasets provided by Google for use within this platform, and producing imagery that visualized the amount of ground covered by vegetation, I was able to derive these mean values and merge them with the US Census Bureau median household income dataset, creating a dataset of 181 observations.

Data Manipulation

Because this research only focused on zip code data, the Median Household income data was filtered by year to reduce inconsistencies, as well as by ‘All Households’, as there was additional data included in the initial dataset that included households with children as a separate category. Furthermore, there were double identical entries for each zipcode, so another filter was applied to ensure only one entry was provided for each zip code.

The final merged and filtered dataset had 181 rows, with each zip code being the unit of observation. The final variables were: FIPS, Location, Household Type, TimeFrame, DataFormat, Data, mean_1985, mean_2020, and NDVI_difference.

Statistical Appendix

1. Summary Statistics for each variable:

Summary Statistics for each variable
Summary Statistics for each variable

2. Detailed values for the graph ‘“Mean NDVI Change based on relation to Median Income’

3. Hypothesis being tested: Higher median household income increases greenspace gain

Omitted Variables: This study does not account for the ways that median income has changed between the two years being observed in the NDVI_difference value derivation. Because median household income by zip code data was not available for 1985, a “Median Income Difference” variable could not be derived, which could have had a correlation with the NDVI_difference variable.

4. Univariate regression: Regress NDVI_difference against med_income

Median Income’s effect on NDVI change in each zip code is statistically significant at 0%
Median Income’s effect on NDVI change in each zip code is statistically significant at 0%