ABSTRACT
For evaluation of the Climate Prediction Center-MORPHing (CMORPH) satellite rainfall product in the Zambezi Basin, daily time series (1998–2013) of 60 rain gauge stations are used. Evaluations for occurrence and rain rate are at sub-basin scale and at daily, weekly, and seasonal timescale by means of probability of detection (POD), false alarm ratio (FAR), critical success index (CSI) and frequency bias (FBS). CMORPH predicts 60% of the rainfall occurrences. Rainfall detection is better for the wet season than for the dry season. Best detection is shown for rainfall rates smaller than 2.5 mm/day. Findings on error decomposition revealed sources of Hit, Missed and False rainfall bias. CMORPH performance (detection of rainfall occurrences and estimations for rainfall depth) at sub-basin scale increases when daily estimates are accumulated to weekly estimates. Findings suggest that for the Zambezi Basin, errors in CMORPH rainfall should be corrected before the product can serve applications such as in hydrological modelling that largely rely on reliable and accurate rainfall inputs.
Acknowledgments
The authors are grateful to the University of Zimbabwe’s Civil Engineering Department for the platform to carry out this research.
Disclosure statement
No potential conflict of interest was reported by the authors.
Author Contributions
Webster Gumindoga was responsible for the data analysis and article development. Tom Rientjes was responsible for the research approach and conceptualization. Tom and Alemseged Haile were responsible for synthesising the methodology and made large contributions to the manuscript write-up. Hodson Makurira and Reggiani assisted in interpreting results. All authors have approved the final article.