Forest mapping and monitoring in Africa using Sentinel-2 data and deep learning

Download PDF
Open Webpage

Anders U. Waldeland, Øivind Due Trier, Arnt-Børre Salberg

Published: 01 Jan 2022, Last Modified: 05 Nov 2025Int. J. Appl. Earth Obs. Geoinformation 2022EveryoneRevisionsBibTeXCC BY-SA 4.0
Abstract: Highlights•A UNet type convolutional neural network is used to estimate vegetation height from Sentinel2-data.•The method achieves a mean average error (MAE) of 4.6 m.•A change detection scheme is proposed to detect clear cutting events in vegetation height time series.