Evaluation of homogeneity of daily ambient temperature within several regional boundaries in New York state

Keywords: time series, homogeneity, regional boundaries, environmental variables
Session Type: Virtual Poster Abstract
Day: Friday
Session Start / End Time: 2/25/2022 05:20 PM (Eastern Time (US & Canada)) - 2/25/2022 06:40 PM (Eastern Time (US & Canada))
Room: Virtual 21

Abstract

Time series regression studies have been widely used in environmental epidemiology, notably for the investigation of the association between health outcomes and short-term exposure to air pollution or temperature. Typically, for both health outcome and exposure, data are available at a regular time intervals over an area, such as a city or county. Time series regressions are known for its robustness by explicitly accounting for delayed effects and long-term trend, although an assumption on the homogeneity of exposure within the study area is rarely tested. In the present study, we will evaluate the homogeneity of daily ambient temperature between 2009-2016 within several regional boundaries delineated by different stake holders in New York State. Specifically, we considered labor market regions (LMR), core-based statistical areas (CBSA), plant harness zones (PHZ), and air quality health advisory regions (AQHR). We compared the distribution of daily ambient temperature variability quantified by standard deviation over the study period to identify the regional definition that yields the smallest variance, i.e., homogeneous boundary. The mean and standard deviation of spatial variation of temperature over 8 years for LMR, CBSA, PHZ, and AQHR are 4.42 (1.59), 2.33 (0.87), 2.37 (1.43), 3.73 (1.37), respectively. We concluded that core-based statistical areas boundary yields the most homogeneous distribution of the ambient daily temperature. In the future work, we will include other exposure variables, i.e., PM2.5 or Ozone, and will determine the boundary for the homogeneous zone of environmental variables.