U. Pérez-Goya, A. Fernández Militino, M. D. Ugarte Martínez, P. M Atkinson
The use of high resolution images or fine images in small regions is still scarce in many applications because free access satellite platforms mainly provide small or medium resolution images. Nowadays, Sentinel-2 satellite provides fine images every 10 days, yet frequently these images are dropout because the presence of clouds. In this work we propose a new downscaling method that merges fine images from Sentinel-2 and daily coarse images from Modis to generate high resolution time series of images. The procedure is based on averaging the fine images on a spatio-temporal window to accommodate spatio-temporal dependence, and smooth the anomalies generated from the difference of the coarse images and fine mean images. The procedure is evaluated in a simulation scenario using both the Sentinel-2 and the Modis images, as fine and coarse images respectively.
Palabras clave: satellite imagery, downscaling, data fusion
Programado
EET-2 Estadística Espacial y Espacio Temporal
5 de septiembre de 2019 10:40
I2L5. Edificio Georgina Blanes