Evaluation of reanalysis rainfall estimates over Ethiopia

• Published in: International journal of climatology Vol. 29; no. 1; pp. 67 - 78
• Main Authors: Diro, G. T., Grimes, D. I. F., Black, E., O'Neill, A., Pardo‐Iguzquiza, E.
• Format: Journal Article
• Language: English
• Published: Chichester, UK John Wiley & Sons, Ltd 01.01.2009; Wiley
• Subjects: Earth, ocean, space; ERA‐40; Ethiopia; Exact sciences and technology; External geophysics; Geophysics. Techniques, methods, instrumentation and models; Meteorology; NCEP; rainfall; reanalysis; Water in the atmosphere (humidity, clouds, evaporation, precipitation); East Africa; Ethiopia; Africa; Weather forecast; rainfall; atmospheric precipitation; NCEP; climate variability; digital simulation; case studies; Hydroclimatology; Rainfall rate; Interannual variation; Dynamical climatology; Information use; Missing data; Spatial distribution; algorithm performance; ERA-40; Reanalysis; Numerical forecast; Incomplete information; seasonal variations; Ethiopia
• ISSN: 0899-8418; 1097-0088
• DOI: 10.1002/joc.1699

There is a pressing need for good rainfall data for the African continent both for humanitarian and climatological purposes. Given the sparseness of ground‐based observations, one source of rainfall information is Numerical Weather Prediction (NWP) model outputs. The aim of this article is to investigate the quality of two NWP products using Ethiopia as a test case. The two products evaluated are the ERA‐40 and NCEP reanalysis rainfall products. Spatial, seasonal and interannual variability of rainfall have been evaluated for Kiremt (JJAS) and Belg (FMAM) seasons at a spatial scale that reflects the local variability of the rainfall climate using a method which makes optimum use of sparse gauge validation data. We found that the spatial pattern of the rainfall climatology is captured well by both models especially for the main rainy season Kiremt. However, both models tend to overestimate the mean rainfall in the northwest, west and central regions but underestimate in the south and east. The overestimation is greater for NCEP in Belg season and greater for ERA‐40 in Kiremt Season. ERA‐40 captures the annual cycle over most of the country better than NCEP, but strongly exaggerates the Kiremt peak in the northwest and west. The overestimation in Kiremt appears to have been reduced since the assimilation of satellite data increased around 1990. For both models the interannual variability is less well captured than the spatial and seasonal variability. Copyright © 2008 Royal Meteorological Society