Machine learning evapotranspiration (ET) estimation using Sentinel-2 and ECOSTRESS data
Abstract
The proposed project goal is to extend the development of the ET estimation service, which was already done in the ASAP project, VASE3 and Hedwig project4. In the VASE project freely available Copernicus Sentinel and other geoinformation data was used to create an evapotranspiration map for Vienna and the surrounding area, which provides information about open green spaces that exhibit particularly good evapotranspiration. The idea of the proposed service is to develop a methodology for estimating ET at the scale of 10 meters, by using an innovative machine learning data fusion to combine the data from the Copernicus Sentinel-2 mission at 10m, with NASA ECOSTRESS mission at 70 m, resulting in a complete ET map over Austria. Sentinel-2 capabilities will be used as input and NASA ECOSTRESS mission as reference. The deep learning model will have as objective to retrieve the relationship between thermal infrared (TIR) and normalized vegetation index (NDVI) and by using additional variables estimating ET. The validation will be done by the partner project by setting up ET experiments in climate chambers and measuring ET of different plant communities throughout the year in different weather conditions to ensure the accuracy of the final product. They will also help in engaging decision-makers across the entire value chain, conducting extensive public outreach efforts, and considering the requirements of target audiences and potential stakeholders.
Project staff
Bernhard Scharf
Dipl.-Ing.Dr. Bernhard Scharf
bernhard.scharf@boku.ac.at
Tel: +43 1 47654-87413
Project Leader
01.11.2024 - 30.04.2027
Rosemarie Stangl
Univ.Prof. Dipl.-Ing.Dr. Rosemarie Stangl
rosemarie.stangl@boku.ac.at
Tel: +43 1 47654-87401, 87426
Sub Projectleader
01.11.2024 - 30.04.2027
BOKU partners
External partners
SISTEMA GmbH
coordinator
EOX IT Services GmbH
partner
Grünstattgrau Forschungs- und Innovations GmbH
partner