How changing grain size affects the land surface temperature pattern in rapidly urbanizing area: a case study of the central urban districts of Hangzhou City, China.

Abstract

Urbanization has led to the rapid and large-scale changes in land use and land cover and has affected the spatial distribution of land surface temperature (LST) in urban areas. Studying the LST pattern and their spatial heterogeneity characteristics at different scales can help understand the dynamic mechanism of the thermal landscape and provide insights into urban ecological planning. We utilized transfer matrixes, landscape metrics, and spatial autocorrelation analyses to study the transfer of LST classes, changes in the LST pattern, and changes in LST clusters with varying grain sizes by taking the central urban districts of Hangzhou City in China as a case study. Results indicate that (1) the transfer proportion of the LST classes increased, except for high-temperature class, and each LST class shifted to the adjacent dominant LST class with the increase in grain size. (2) The landscape metrics remarkably changed as the grain size increased, indicating that the LST pattern was scale-dependent. As the grain size increased, the small patches gradually merged into large patches; the fragmentation, complexity, and ductility of the urban thermal landscapes decreased; and the shape of the patches became simple and regular. (3) The LST pattern exhibited a positive spatial autocorrelation. The area of low-low cluster decreased, whereas that of non-significant clusters substantially increased with the grain size. The area of high-high cluster remained steady when the grain size exceeded 90m. (4) Patch density, mean patch fractal dimension, clumpiness index, and contagion index exhibited predictable responses to changing grain size, whereas Shannon's diversity and Shannon's evenness indexes showed erratic responses, indicating that the diversity and evenness of the LST pattern were not scale-dependent. (5) The suitable domain of scale for the analysis of LST pattern was (60, 120), and the optimal grain size was 120m. The selection of domains of scale and optimal grain size need to be determined according to the changes in thermal landscape patterns at different grain sizes and regional environments.