Averaging Strategy to Improve SAR-To-NDVI Estimations in A Region of Interest

The Normalized Difference Vegetation Index (NDVI) is an index for quantifying the health and density of vegetation, as it has a high correlation with the actual state of the vegetation on the ground. However, cloud cover is a major limitation in data acquisition. A possible solution is to estimate NDVI from radar images by means of a deep learning model.This study aims to improve quality of the NDVI estimations in a Region of Interest (ROI) with an averaging strategy that combines different available approximations from the SARtoNDVI estimator, a Conditional Generative Adversarial Network (cGAN) that approximates NDVI from processed Synthetic Aperture Radar (SAR) images.