Articles | Volume 16, issue 2
Research article
14 Feb 2012
Research article |  | 14 Feb 2012

Comparison and evaluation of satellite derived precipitation products for hydrological modeling of the Zambezi River Basin

T. Cohen Liechti, J. P. Matos, J.-L. Boillat, and A. J. Schleiss

Abstract. In the framework of the African DAms ProjecT (ADAPT), an integrated water resource management study in the Zambezi Basin is currently under development. In view of the sparse gauging network for rainfall monitoring, the observations from spaceborne instrumentation currently produce the only available rainfall data for a large part of the basin.

Three operational and acknowledged high resolution satellite derived estimates: the Tropical Rainfall Measuring Mission product 3B42 (TRMM 3B42), the Famine Early Warning System product 2.0 (FEWS RFE2.0) and the National Oceanic and Atmospheric Administration/Climate Prediction Centre (NOAA/CPC) morphing technique (CMORPH) are analyzed in terms of spatial and temporal repartition of the precipitations. They are compared to ground data for the wet seasons of the years 2003 to 2009 on a point to pixel basis at daily, 10-daily and monthly time steps and on a pixel to pixel basis for the wet seasons of the years 2003 to 2007 at monthly time steps.

The general North-South gradient of precipitation is captured by all the analyzed products. Regarding the spatial heterogeneity, FEWS pixels are much more inter-correlated than TRMM and CMORPH pixels. For a rainfall homogeneity threshold criterion of 0.5 global mean correlation coefficient, the area of each sub-basin should not exceed a circle of 2.5° latitude/longitude radius for FEWS and a circle of 0.75° latitude/longitude radius for TRMM and CMORPH considering rectangular meshes.

In terms of reliability, the correspondence of all estimates with ground data increases with the time step chosen for the analysis. The volume ratio computation indicates that CMORPH is overestimating the rainfall by nearly 50%. The statistics of TRMM and FEWS estimates show quite similar results.

Due to its lower inter-correlation and longer data set, the TRMM 3B42 product is chosen as input for the hydraulic-hydrologic model of the basin.

Further work will focus on the calibration of the hydraulic-hydrological model of the basin, including the major existing hydraulic structures with their operation rules.