Quantifying land surface temperature change from LISA clusters: An alternative approach to identifying urban land use transformation

Despite the potential to provide new insight into the underlying heterogeneity of urban thermal landscapes, local indicators of spatial autocorrelation (LISA) remains underutilized in land surface temperature (LST) related study. The present research applies local Moran's I as a unique approach to recognize the pattern of statistically significant LST increase by detecting clusters of localized hot spots. The single channel algorithm has been applied to extract LST from nine temporal Landsat datasets ranging from 2001 to 2013. LST difference maps generated with 2001 as the base year have been filtered by local Moran's I at four different spatial weights (90 m, 120 m, 150 m and 200 m) to represent a spatially homogeneous form of LST clusters. Smaller spatial weights are useful to map more localized and specific pattern of urban LST change, whereas larger spatial weights produce a somewhat generalized pattern, useful for studying the coarse scale phenomenon. Integrating LISA clusters to the temperature-vegetation (TVX) feature space validates the hot spots as areas with significant land use transformations in the last decade. Further, the spatial sector-wise concentration of hot spots throughout the time series determines the direction and magnitude of urban expansion. The study provides innovative insights on how the integrated application of local Moran's I on LST change and TVX can be used to delineate the regions of acute land transformation and, in turn, act as a decision-making tool for urban planning. (C) 2016 Elsevier B.V. All rights reserved.