Delft - Many of the world`s river basins are not subject to accurate monitoring, especially those in developing countries. The few measurements that have been taken in these areas are often uncertain, incomplete or chronologically inconsistent. But hydrological models are essential to predict drought, high water and the availability of water.

Delft researcher Hessel Winsemius was able to combine the scarce "ground-level data" from poorly monitored river basins with hydrological expertise and modern satellite data, enabling him to improve the existing hydrological models. He used satellite-based estimates of rainfall and evaporation rates and gravity measurements from the GRACE satellite and applied this to the Zambezi river basin, for example. "This river basin provides us with an excellent opportunity to put the new methods directly into practice."

"Fortunately, increasing amounts of relevant data are being made available free of charge. Satellite data is particularly interesting in this respect. Important factors such as rainfall, evaporation, radiation, soil moisture and water retention can now be estimated on the basis of raw satellite measurements,"

The methods have proved successful. From a case study on the Upper Zambezi, it has turned out that a combination of ground-level data, data from GRACE and an expert knowledge of the hydrology of the area have produced a model with a robust structure.

"A second case study on a sizeable tributary of the Zambezi – the Luangwa – demonstrated that the values of the model parameters can be indirectly adjusted on the basis of small amounts of data originating from the low-quality ground-level data on the one hand, and the evaporation rate estimates from the satellite measurements on the other. Of course, this leads to a considerable reduction in the uncertainty of the model," says Winsemius.

Original source: H. C. Winsemius, Satellite data as complementary information for hydrological modelling, available under vssd.nl

Quelle: Delft University of Technology

Zurück