U. Pérez Goya, A. Fernández Militino, M. D. Ugarte Martínez
Removing clouds or any other atmospheric effects is always an important task when working with satellite imagery.
A high presence of clouds can usually produce many images dropouts, but when it is only partially clouded, the missing or distorted data can be filled with multi-temporal images.
This work presents a new method of filling clouds in remote sensing data. The technique applies a thin-plate spline to a lower resolution of the mean residuals obtained in a neighbourhood of the target image.
The new procedure is evaluated in a simulation study with weekly and biweekly images captured in Navarre from 2011 to 2013 provided by MODIS TERRA and MODIS AQUA. The study compares the performance of the new method with some well-know procedures as TimeSat, Gapfill and Hants. Results show a precision increase of the new method compared with the rest of alternatives.
Palabras clave / Keywords: cloud-filling, time-series, satellite, images
Programado
Sesión V08 Estadística Espacial y Espacio-Temporal II
1 de junio de 2018 17:20
Sala 1
