Spatio-temporal non-uniformity of urban park greenness and thermal characteristics in a semi-arid region

Thermal characteristics of urban parks often vary city-wide due to different landscape properties of parks themselves or their surrounding environments. Understanding such heterogeneity is critical for strategic use of urban forest elements to mitigate extreme heat and provide various ecological amenities; however, relatively few studies to date have assessed such variability at the whole-city scales and across different seasons. This study investigated seasonal variation, statistical association and local spatial clustering in satellite image-based proxies of vegetation greenness and surface temperature (normalized difference vegetation index (NDVI) and at-sensor surface brightness temperature (Tb), respectively) among 135 parks in a part of California, USA on ten different-season dates in 2014. Both metrics varied among parks and dates and exhibited a significant negative linear relationship which was stronger on warmer-season dates. Regressions with NDVI explained 2-17% more variance in Tb when they also included the proportion of woody plant cover (negative effect on Tb) or the proportion of grass cover (positive effect on Tb) on all dates, and park area on some dates (negative effect on Tb). The analysis of spatial variation in park properties revealed several significant local clusters of parks with higher Tb that persisted among warmer dates and had significantly smaller area and warmer neighborhoods than did significant clusters of greener or cooler parks. These results highlight potential under-provisioning of park microclimatic benefits in the associated neighborhoods and calls for further research on environmental and social implications of these results to inform mitigation of urban heat in this region and similar climates.