Accurate forecasting of small-scale (0.1 - 10 km) atmospheric gradients in the world's urbanized coastal regions is challenging, but can be of great societal utility with dangerous weather events, air quality, security, as well as helping support commerce and a good quality of life. Temperature gradients, for example, can be large – differences across suburban-urban and inland-coastal boundaries around New York City (NYC) are frequently 5-10°C. Here, temperature measurements from hundreds of meteorological stations over multiple seasons are analyzed to better understand and map out NYC microclimates. Observations are also compared with a high-resolution numerical model with the goal of evaluating and improving our modeling capabilities.

The Naval Research Laboratory's Coastal Ocean-Atmosphere Mesoscale Prediction System (COAMPS) is utilized, focusing analyses on the inner two nested grids with resolutions of 1000 and 333 m. The model includes an urban canopy parameterization of land-surface impacts on radiation, momentum and turbulent mixing. It is also initialized by the Navy's global reanalysis of sea surface temperature from satellite and in situ data, and has previously been shown to capture sharp coastal atmospheric gradients. The observations are from a wide range of amateur, professional and governmental meteorological stations, so statistical analyses are utilized to objectively define outliers. Clustering and principal components analysis are used to delineate microclimatic regions and examine how well COAMPS and the observations compare in their perspectives on the primary modes of variability and their spatial patterns. Lastly, the observations are utilized to examine how well the region's historical meteorological stations, which are primarily at airports, capture weather and climate in the city's neighborhoods.