Spatial-temporal Fusion Of Land Surface Temperatures Retrieved From Landsat-8 And MODIS Imageries:Improvement And Programming Realization Of SADFAT Algorithm

Multi-scale land surface temperature (LST) is an indispensible parameter in studying the urban thermal fields, ecological processes and climate changes. It also values a lot in many studies, such as surface energy balance, soil moisture conditions, and urban heat islands, et al. Thermal infrared (TIR) images with both high spatial and temporal resolutions are highly required to retrieve LSTs in these studies. So far, due to the technical limitations, there is no single satellite sensor can deliver TIR imagery wtih both high temporal and spatial resolution. In recent years, various models were put forward to spatially downscale the thermal imagery, however, few of them was able to downscale the thermal imagery at temporal and spatial level simultaneously. Therefore, study of spatio-temporal data fusion model for temperature mapping has great importance to applications and developments of thermal remote sensing.In this thesis, the newly presented spatial-temporal fusion model SADFAT (Spatio-temporal Adaptive Data Fusion Algorithm for Temperature mapping) was introduced and modified, taking Yanping district, Nanping city and Youxi county, Sanming city, Fujian province, China as the study area. Four pairs of Landsat-8 and MODIS data over the study area of were used. By introducing a new edge moving window algorithm together with a new clouds removal method, the original SADFAT model was improved. The modified SADFAT presented in this study was fully tested from three different levels including single band level, vegetation index level and LST level. A number of qualitative indices were also introduced to make the assessment of the fusion effects of modified SADFAT more reliable and convincible. In addition, the modified SADFAT was realized by combining IDL (Interactive Data Language) and C# programming. When compared with conventional ways of realization of the spatio-temporal fusion of thermal infrared imageries, the system constructed in this study showed to be more powerful and convenient, thus has great potential in future.