A platform for research: civil engineering, architecture and urbanism
A platform for research: civil engineering, architecture and urbanism
Downscaling of Coarse Resolution Land Surface Temperature Through Vegetation Indices Based Regression Models
Abstract In geoscience and remote sensing necessitate thermal imagery having high-resolution for various applications like estimation of the Land surface temperature (LST) analysis, thermal comfort, urban energy resources, forest fire, assessment of evapotranspiration, drought prediction, etc. We need accurate and sharp thermal images to explore surface temperature related phenomenon on frequent basis. The present physical and technological constraints have not allowed us to dig up remote sensing thermal data at high temporal and spatial resolution simultaneously. Hence, it is obligatory to construct a dynamic relation between low- and high-resolution satellite data to acquire enhanced thermal images. The present study evaluates three downscaling algorithms in our study area, namely, disaggregation of radiometric surface temperature (DisTrad), sharpening thermal imagery (TsHARP), and local model using seasonal Landsat 8 and MODIS data thermal imagery. The aggregated Landsat 8 LST of 1000 m resolution has been downscaled to 400, 300, 200, and 100 m using DisTrad, TsHARP, and the local model and compared with original Landsat 8 and resampled LST of matching level. The results have shown that LST downscaling technique performance varies over climate, surface feature and earth surface moisture conditions. The models have not performed well in surface having highest and lowest water content i.e. water bodies and arid sandy areas. Alternatively, regression-based downscaling accuracy is higher for NDVI > 0.3. For example, the accuracy of all algorithms is higher for the growing seasons (February and October) unlike the harvesting season (April). The root means square error of the downscaled LST increases from 400 to 100 m spatial resolution in all seasons. The downscaling algorithms gave realistic results of MODIS satellite thermal band to a spatial resolution of 200 m. The present study is an attempt to rationalize coarse resolution thermal image by using the association between earth facade vegetation indices and land surface temperature. The study aims to develop a robust LST downscaling algorithm for MODIS data at LANDSAT resolution. The downscaling methods successfully operate over a heterogeneous landscape and reduced thermal mixture effect to monitor the daily basis long-term environmental phenomena.
Downscaling of Coarse Resolution Land Surface Temperature Through Vegetation Indices Based Regression Models
Abstract In geoscience and remote sensing necessitate thermal imagery having high-resolution for various applications like estimation of the Land surface temperature (LST) analysis, thermal comfort, urban energy resources, forest fire, assessment of evapotranspiration, drought prediction, etc. We need accurate and sharp thermal images to explore surface temperature related phenomenon on frequent basis. The present physical and technological constraints have not allowed us to dig up remote sensing thermal data at high temporal and spatial resolution simultaneously. Hence, it is obligatory to construct a dynamic relation between low- and high-resolution satellite data to acquire enhanced thermal images. The present study evaluates three downscaling algorithms in our study area, namely, disaggregation of radiometric surface temperature (DisTrad), sharpening thermal imagery (TsHARP), and local model using seasonal Landsat 8 and MODIS data thermal imagery. The aggregated Landsat 8 LST of 1000 m resolution has been downscaled to 400, 300, 200, and 100 m using DisTrad, TsHARP, and the local model and compared with original Landsat 8 and resampled LST of matching level. The results have shown that LST downscaling technique performance varies over climate, surface feature and earth surface moisture conditions. The models have not performed well in surface having highest and lowest water content i.e. water bodies and arid sandy areas. Alternatively, regression-based downscaling accuracy is higher for NDVI > 0.3. For example, the accuracy of all algorithms is higher for the growing seasons (February and October) unlike the harvesting season (April). The root means square error of the downscaled LST increases from 400 to 100 m spatial resolution in all seasons. The downscaling algorithms gave realistic results of MODIS satellite thermal band to a spatial resolution of 200 m. The present study is an attempt to rationalize coarse resolution thermal image by using the association between earth facade vegetation indices and land surface temperature. The study aims to develop a robust LST downscaling algorithm for MODIS data at LANDSAT resolution. The downscaling methods successfully operate over a heterogeneous landscape and reduced thermal mixture effect to monitor the daily basis long-term environmental phenomena.
Downscaling of Coarse Resolution Land Surface Temperature Through Vegetation Indices Based Regression Models
Sharma, Kul Vaibhav (author) / Khandelwal, Sumit (author) / Kaul, Nivedita (author)
2019-06-20
12 pages
Article/Chapter (Book)
Electronic Resource
English
Downscaling land surface temperatures at regional scales with random forest regression
| Online Contents | 2016
Downscaling land surface temperatures with multi-spectral and multi-resolution images
| Online Contents | 2012
Downscaling species occupancy from coarse spatial scales
| Wiley | 2012