Measuring water availability with limited ground data: assessing the feasibility of an entirely remote-sensing-based hydrologic budget of the Rufiji Basin, Tanzania, using TRMM, GRACE, MODIS, SRB, and AIRS

This study explores the feasibility of an entirely satellite remote sensing (RS)-based hydrologic budget model for a ground data-constrained basin, the Rufiji basin in Tanzania, from the balance of runoff (Q), precipitation (P), storage change (S), and evapotranspiration (ET). P was determined from the Tropical Rainfall Measuring Mission, S from the Gravity Recovery and Climate Experiment, and ET from the Moderate Resolution Imaging Spectroradiometer, the surface radiation budget, and the Atmosphere Infrared Radiation Sounder. Q was estimated as a residual of the water balance and tested against measured Q for a sub-basin of the Rufiji (the Usangu basin) where ground measurements were available (R-2=0.58, slope=1.9, root mean square error=29 mm/month, bias=14%). We also tested a geographical information system (GIS)-driven (ArcCN-runoff) runoff model (R-2=0.64, slope=0.43, root mean square error=39 mm/month). We conducted an error propagation analysis from each of the model's hydrologic components (P, ET, and S). We find that the RS-based model amplitude is most sensitive to ET and slightly less so to P, whereas the model's seasonal trends are most sensitive to S. Although RS-GIS-driven models are becoming increasingly used, our results indicate that long-term water resource assessment policy and management may be more appropriate than instantaneous' or short-term water resource assessment. However, our analyses help develop a series of tools and techniques to progress our understanding of RS-GIS in water resource management of data-constrained basins at the level of a water resource manager. Copyright (c) 2012 John Wiley & Sons, Ltd.