COPERNICUS SENTINEL-2 DATA FOR THE DETERMINATION OF GROUNDWATER WITHDRAWAL IN THE MAGHREB REGION
- 1Federal Institute for Geosciences and Natural Resources, Hannover, Germany
- 2University of Hildesheim, Hildesheim, Germany
- 3Observatory of the Sahara and Sahel, Tunis, Tunisia
Keywords: Copernicus, Sentinel-2, Classification, Multitemporal feature extraction, Evapotranspiration, Water
Abstract. Agriculture plays an important role in the economy of the Maghreb region. Most of the water needed for irrigation comes from pumping of the aquifers. A controlled pumping of the groundwater resources does not exist yet, thus, estimating the total water consumption for agricultural use only with in situ data is nearly impossible. In order to overcome this lack of information, Copernicus data are used for determining the groundwater withdrawal through agriculture in the Maghreb region. This paper presents an approach for estimating and monitoring crop water requirements in Tunisia based on multitemporal Sentinel-2 data. Using this multitemporal information, a thorough analysis of the different culture types over time is possible, from which a set of additional multitemporal features is deduced for crop type classification. In this paper, the contribution of those features is analyzed, showing a classification accuracy enhanced by 10 % with the multitemporal features. Furthermore, relying on existing methods and FAO standards for the estimation of crop water needs, the methodology aims to estimate the specific crop water consumption. The results of the water estimates are validated against delimited areas where estimates of the water consumption are available from the authorities. Finally, as the study is conducted within the framework of an international technical cooperation, the methodology aims to be reproducible and sustainable for local authorities. The particularity of the results presented here is that they are achieved through automatic processing and using exclusively Open Source solutions, deployable on simple workstations.