The study of urban heat islands (UHI) is of great importance in the context of climate change (CC). The use of satellite images has helped considerably to understand UHI, especially in analyses of land surface temperature (LST). However, available sensors have a major limitation: their low spatial resolution (100 meters, Landsat; 1000 meters, MODIS) does not allow detailed analysis of UHI. Moreover, most remote sensors are limited to daytime data collection, while UHI mainly appear during the night. The images provided by satellites that study nocturnal thermal radiation, such as MODIS, have very low spatial resolution. There is abundant literature about the fusion of images from several satellites and sensors, especially information from MODIS and Landsat. However, most of these studies have concentrated on studying the combination of congruent images in the temporal plane, to extrapolate the results obtained to other temporal instances for which there is no detailed information. In general, few studies have focused on increasing the resolution of thermal images beyond the 100 meters/pixel of Landsat. The objective of this paper is to combine information from various sensors (Modis, Landsat 8 and Sentinel 2) by constructing a set of OLS models of daytime and nighttime LST. These models provide a detailed view of daytime UHI (10 meters) and a robust evaluation of the range of cooling produced during the night. A modelling exercise at 1 meter/pixel of resolution has also been developed, using information from more detailed sensors installed on aircraft in the Barcelona Metropolitan Area.
|
