Skip to main content

Impact of Extreme Weather on Hard-to-Capture, Vulnerable Populations: Evidence from Hart Island — New York’s Public Burial Ground

Investigators:

Frank W. Heiland, Deborah Balk, Selen Ozdogan, Jennifer Brite, Peter Marcotullio and Christian Braneon

Funding:

NIAID R03 Grant AI166809-01, NIA R61AG086854 (CACHE)

Data sources:

  • Daily Unclaimed Death Records: The burial records from Hart Island, New York City’s public cemetery located on an island near the Bronx in the Long Island Sound, are publicly available through a searchable online database named Hart Island Lookup Service (New York City Department of Correction, 2024). We scrape the data between 1977 and 2022 for the present study. The Hart Island records include exact date of death (day-month-year) and, for most records, first and last names, age at death, sex (male/female/unknown), and place of death (homes, residential care facilities, hospitals or on the street). A small number of records are available before 1977, but the counts suggest the data are not complete and were therefore excluded. On Hart Island, graves are numbered and organized in sections within plots. For bodies buried on Hart Island in more recent years the records typically also include a medical examiner case number, which reveals information about the location at death (Heiland et al., 2023; New York City Department of Correction, 2024). There have been very few child burials on Hart Island since 1977, but there are many records of infants and fetuses. The present analysis looks at adults and children only.
  • Air temperature: The National Oceanic and Atmospheric Administration (NOAA) collects daily weather-station meteorological data in the Global Historical Climatology Network (GHCN) (Menne, Durre, Korzeniewski, et al., 2012). The records for New York City are reliable historically and the data are provided in year-month-day format. We use daily maximum and minimum dry bulb (air) temperature measured at three stations (or point locations) in NYC (LaGuardia airport in northern Queens county, Central Park in Manhattan (New York county), and JFK airport in southern Queens county) in degrees Celsius. We average the temperature extremes over these three locations to obtain the city-wide maximum and minimum for each day and then convert to Fahrenheit for ease of interpretation.
  • Wet bulb temperature: We use the ERA5-Land hourly data (Muñoz-Sabater et al., 2021) to calculate the wet bulb temperature, which measures how warm it feels based on evaporative cooling (i.e. taking air humidity into account). These satellite-based, reanalysis raster-format data indicate hourly temperature and dewpoint in Celsius at a spatial resolution of 0.25 by 0.25 latitude-longitude degrees. Using the underlying variables, we obtain the hourly wet bulb temperature for each grid between 1977 to 2022 and then convert to Fahrenheit. We then temporally aggregate it by calculating the daily maximum and minimum wet bulb temperatures. Then, we spatially aggregate the raster data by retrieving the maximum wet bulb temperature within the boundaries of New York City, obtained from the Borough Boundaries vector data (NYC OpenData, 2024). We call this measure of the maximum value of the daily maximum temperature the “max-max” wet bulb temperature. We obtain the “min-min” in a similar manner.

Measures:

  • Mortality Measures: Unclaimed death count in New York City since 1977 (daily number of deaths where the body remained unclaimed and was subsequently buried on Hart Island — the City’s public burial ground). Recent research shows the importance of understanding unclaimed and public burial deaths (Brite et al. 2024, Sohn et al. 2020), and to the extent that daily death records are reported, these data are structurally similar to other types of administrative data that indicate daily demographic events.
  • Aging/older adults Measures: Age at death information is available for most adult decedents (the analysis is restricted to persons over age 1 though very few persons under 18 are buried on Hart Island). We create a dichotomous death count measure for population under 65, and 65 and above.
  • Climate Measures: Daily data allow for the construction of measures of temperature and total precipitation so that we can examine the associations between climate and mortality over 50 years among the especially vulnerable. NOAA data allow for measures of air temperature and the ERA5 reanalysis data allow for the construction of a wet bulb measure. We use temperature maximum and minimum rather than average temperatures in the main analysis, in order to isolate the effects of changes in minimum and maximum separately; both are commonly used in the literature (Benmarhnia et al., 2015; Bunker et al., 2016; Weilnhammer et al., 2021)

Project Summary:

We examine the association between weather extremes and persons buried in the nation’s largest indigent burial ground—Hart Island, New York, where more than one million unclaimed individuals are interred. Public burial records provide a window into mortality among particularly vulnerable populations. We connect public-burial deaths with extreme temperatures, and precipitation, in first of their kind analyses. This study takes place in two parts.

In the first part of the analysis, we link New York City daily air and wet bulb temperature patterns for the period 1977-2022 to daily deaths that remain unclaimed and were subsequently buried on Hart Island. We find robust evidence linking peak summer temperatures to unclaimed mortality. On average, a 1-degree Fahrenheit (.556-degree Celsius) higher maximum peak daily air temperature in a 7-day (3-day) summer period predicts 1.2% (1%) more daily unclaimed deaths on day 7 (3), controlling for precipitation and season, year, decade and holiday effects. Further analysis suggests that 5.1% of all extreme heat related deaths were unclaimed city-wide, implying New Yorkers who died due to extreme heat were more than three times as likely to be unclaimed. (The measures based on the weather data indicated above are for this long-term analysis only.)

In the second part of this analysis (beginning in Spring 2025), we consider the spatial variation in temperatures within New York City, for a more recent time period. That analysis will use measures of temperature based on land-surface temperature as well as air temperature in order to capture variation in the built-environment (Hrisko et al., 2020; Naserikia et al., 2023). Especially for large cities, built-environment features are considered important to understanding how people experience heat. These data will be cross-validated with a set of temperature data collected at schools, public housing campuses and selected other public locations throughout New York City (https://crest.cuny.edu/uHMT/index.html). We will integrate the spatial temperature data with geocoded unclaimed death records from Hart Island and study vulnerability hotspots and periods across New York City neighborhoods.  

This research highlights the undue burden that climate change places on deeply vulnerable populations, and thus the need to use innovative demographic and climate data with integrated methods to better capture and understand these impacts.

On the creation of the weather variables:

Questions and answers supplied here have been excerpted from Heiland et al. 2025 [link forthcoming].

Q:   Why didn’t we use long-term historical data on extreme heat that is made available by the Centers for Disease Control on extreme heat days and daily maximum summer temperatures for all counties and census tracts?
A:   The analysis of public burial deaths (on Hart Island) examines daily temperature mins and maxes in all seasons, so we needed data that included all days, and both minima and maximum temperatures.

Q:   Why did we choose temperature windows of 1-, 3-, 7- and 10-days?
A:   Evidence from studies on extreme temperatures and mortality and morbidity suggest extreme temperature spells impact with a lag of up to 10 days and differential sensitivity to heat versus cold spells. While extreme temperatures can have immediate health impacts measurable in single-day models, extended exposure periods may produce additional and distinct effects beyond these immediate impacts. The most relevant exposure duration can vary significantly based on factors, such as age, access to temperature control systems, built environment, and local climate. By examining multiple time windows, we can provide sensitivity analysis to the choice of the exposure window but also determine which duration is most significant for understanding temperature effects in the Hart Island unclaimed deaths sample.

Q:   Why choose both wet bulb and air temperature?
A:   We examine both wet bulb and air temperature measurements because they capture different aspects of how heat affects human health. While air temperature alone is important, humidity plays a crucial role in how the body experiences and responds to heat. Heat indices like wet bulb measures combine temperature and humidity to indicate human discomfort (Anderson & Bell, 2009; Ashcroft, 2002; Oudin Åström et al., 2015). The body primarily cools itself through sweat evaporation, but high humidity reduces this cooling mechanism’s effectiveness. This is particularly relevant because humans have a physiological limit for heat tolerance at a wet bulb temperature of 35°C (Raymond et al., 2020), and certain populations, especially older adults, have a reduced ability to regulate their body temperature (Brazaitis et al., 2017; Cramer et al., 2022; Meade et al., 2020). Research suggests that there is an upper limit on heat and humidity tolerance for all mammals that could be breached in the future due to climate change (Sherwood & Huber, 2010).

By analyzing both wet bulb and air temperature measurements, we can assess whether incorporating humidity into our temperature metrics meaningfully affects our findings for the Hart Island population. While previous research emphasizes humidity’s role in heat stress, comparing results using both metrics allows us to determine if this theoretical importance translates to practical differences in mortality patterns in our specific sample. Our analysis reveals that the relationship between temperature and mortality follows similar patterns across both measures, though the air temperature models show slightly stronger associations in magnitude.

Outputs:

Link to Poster

Paper (R&R TBD)

Code (TBD)

References:

Adeyeye TE, Insaf TZ, Al-Hamdan MZ, Nayak SG, Stuart N, DiRienzo S, Crosson WL. Estimating policy-relevant health effects of ambient heat exposures using spatially contiguous reanalysis data. Environ Health. 2019 Apr 18;18(1):35. doi: 10.1186/s12940-019-0467-5. PMID: 30999920; PMCID: PMC6471902

Brite, J., Heiland, F.W. and Balk, D., 2025. Understanding deep disadvantage at the end of life: A nationwide analysis of unclaimed deaths. Social Science & Medicine, 365, p.117551. https://doi.org/10.1016/j.socscimed.2024.117551

Hrisko et al. (2020). Urban air temperature model using GOES-16 LST and a diurnal regressive neural network algorithm, Remote Sensing of the Environment https://www.sciencedirect.com/science/article/abs/pii/S0034425719305140

Menne, M. J., Durre, I., Korzeniewski, B., McNeill, S., Thomas, K., Yin, X., Anthony, S., Ray, R., Vose, R. S., Gleason, B. E., & Houston, T. G. (2012). Global Historical Climatology Network—Daily (GHCN-Daily), Version 3 [Dataset]. NOAA National Centers for Environmental Information. https://doi.org/10.7289/V5D21VHZ

Muñoz-Sabater, J., Dutra, E., Agustí-Panareda, A., Albergel, C., Arduini, G., Balsamo, G., Boussetta, S., Choulga, M., Harrigan, S., & Hersbach, H. (2021). ERA5-Land: A state-of-the-art global reanalysis dataset for land applications. Earth System Science Data, 13(9), 4349–4383.

Naserikia et al. (2023). Land surface and air temperature dynamics: The role of urban form and seasonality Science of the Total Environment https://www.sciencedirect.com/science/article/pii/S0048969723059338

New York City Department of Correction. (2024). Hart Island Lookup Service [Dataset].
https://a073-hartisland-web.nyc.gov/hartisland/pages/home/home.jsf

NYC OpenData. (2024). Borough Boundaries [Dataset]. https://data.cityofnewyork.us/City-Government/Borough-Boundaries/tqmj-j8zm

Sohn, H., Timmermans, S. and Prickett, P.J., 2020. Loneliness in life and in death? Social and demographic patterns of unclaimed deaths. Plos one, 15(9), p.e0238348.