Monitoring urban landscape thermal features and change using time series Landsat surface temperature data

Abstract:

Urban development modifies landscape conditions and surface energy balance by replacing non-urban land with various amounts of developed impervious surface. The conversion is known to generate local warming effects by altering surface biophysical properties of the land surface, resulting in urban areas usually exhibiting relatively higher air and surface temperatures than surrounding non-urban lands. The phenomenon is usually named urban heat island (UHI). The most common assessment of UHI uses satellite remotely sensed observation to measure land surface temperature (LST). The LST data can be used to quantify UHI intensity and spatiotemporal variation by incorporating land cover dynamic information. In this work, we introduce a novel framework that integrates Landsat Analysis Ready (ARD) LST and land cover change product produced by the USGS Land Change Monitoring, Assessment, and Projection (LCMAP) project. The framework is built on two main components: time series analysis of LST and urban-rural extent dynamics. The time-series analysis reveals LST annual averages and temporal variations from 1980s to the current. The annual land-cover data leads to reducing uncertainties in characterizing urban land cover, delineating their spatial extents, detecting change trends, and having potential to be used to quantify their biophysical conditions. The framework integrates both LST and land cover data to quantify UHI intensities and change trends in the prototype areas. The framework also identifies multi-scale clusters of urban hotspots using various thresholds within cities. We present prototype results of UHI assessments by implementing the framework in three metropolitan areas cities: Atlanta, Minneapolis, and Sioux Falls. The results suggest that the UHI intensities vary from 2 to 3 ⁰C as annual averages. The temporal trends of the UHI intensity differ from 0.02 to 0.04 ⁰C/year in these areas. We also show UHI intensities derived from other remotely sensed data including Ecostress that has a relatively higher spatial resolution than Landsat. Furthermore, we demonstrate the capability to assess and monitor UHI intensities and their historical trends in major metropolitan areas in the United States using this framework through collaboration with USGCRP and USEPA.