Yrjö Rauste

4.2. Feature Selection Here, we evaluated the potential of Sentinel-1 for GSV prediction using selected sets of images, in contrast to using images in the order of their acquisition dates as in . The idea was to achieve the greatest GSV prediction accuracy using a minimum number of images. In further analyses, we evaluated several feature selection approaches. When particular images were selected, their features were used in stand-level GSV prediction.

This is an early access version, the complete PDF, HTML, and XML versions will be available soon.