Datensatz

Zusammenfassung

Water supplies in northeastern Brazil strongly depend on the numerous surface water reservoirs of various sizes there. However, the seasonal and long-term water surface dynamics of these reservoirs, particularly the large number of small ones, remain inadequately known. Remote sensing techniques have shown great potentials in water bodies mapping. Yet, the widespread presence of macrophytes in most of the reservoirs often impedes the delineation of the effective water surfaces. Knowledge of the dynamics of the effective water surfaces in the reservoirs is essential for understanding, managing, and modelling the local and regional water resources. In this study, a two-year time series of TerraSAR-X (TSX) satellite data was used to monitor the effective water surface areas in nine reservoirs in NE Brazil. Calm open water surfaces were obtained by segmenting the backscattering coefficients of TSX images with minimum error thresholding. Linear unmixing was implemented on the distributions of gray-level co-occurrence matrix (GLCM) variance in the reservoirs to quantify the proportions of sub-populations dominated by different types of scattering along the TSX time series. By referring to the statistics and the seasonal proportions of the GLCM variance sub-populations the GLCM variance was segmented to map the vegetated water surfaces. The effective water surface areas that include the vegetation-covered waters as well as calm open water in the reservoirs were mapped with accuracies >77%. The temporal and spatial change patterns of water surfaces in the nine reservoirs over a period of two consecutive dry and wet seasons were derived. Precipitation-related soil moisture changes, topography and the dense macrophyte canopies are the main sources of errors in the such-derived effective water surfaces. Independent from in-situ data, the approach employed in this study shows great potential in monitoring water surfaces of different complexity and macrophyte coverage. The effective water surface areas obtained for the reservoirs can provide valuable input for efficient water management and improve the hydrological modelling in this region.