Some Validation Results of a Spatially Fine Scale Air Temperature Statistical Model in New York City

Since mortality during heat waves is a sensitive function of temperature, prediction of temperature variations in densely populated areas is crucial. Here we focus on daytime air temperature as satellite thermal imagery can retrieve the radiative component of heat exposure. In previous conferences we have reported on a fine scale temperature variability model based on a series of street level field campaigns in Manhattan. The model blends fine scale spatial measurements with weather dependent temperature variability estimates to predict the daily amplitude of fixed spatial afternoon temperature patterns in this highly urbanized environment. The model has been running daily since the summer of 2015, and has been tested in hindcast mode against both a set of fixed instruments in Manhattan and against a set of volunteer stations throughout the city. The model shows modest improvement over a uniform temperature forecast when compared to instruments carefully placed in the shade, but shows no improvement when compared to the more randomly placed volunteer stations.