Review PaperEvaluation of extreme precipitation estimates from TRMM in Angola

- Comparison of TRMM annual maximum daily precipitation with ground observation data.
- TRMM 3B42 products underestimate the annual maximum daily precipitation.
- TRMM 3B42 estimates can be corrected and the results improved significantly.
- TRMM means are not statistically different to the means based on rain gauged data.
- Pearson type III distribution with regionalized higher order moments led to best fit.

SummaryIn situ ground observation measurement of precipitation is difficult in vast and sparsely populated areas, with poor road networks. This paper examines the use of remote sensors installed in satellites and evaluates the accuracy of TRMM 3B42 annual maximum daily precipitation estimates in Angola, in West Africa, a region where ground monitoring networks are generally. TRMM 3B42 estimates of annual maximum daily precipitation are compared to ground observation data from 159 locations. As a direct comparison between the two datasets for a common specific period and sites is not possible, a statistical approach was adopted to test the hypothesis that the TRMM 3B42 estimates and the ground monitoring records exhibit similar statistical characteristics.The study shows that the annual maximum daily precipitation estimates obtained from TRMM 3B42 slightly underestimate the quantiles obtained from the in situ observations. The use of remote sensing products to estimate extreme precipitation values for engineering design purposes is however promising. A maximum daily precipitation map for a return period of 20 years was computed and in the future, as the length of the remote sensing data series increases, it may be possible to estimate annual maximum daily precipitation estimates exclusively from these datasets for larger return periods.The paper also presents maps of the PdT/PDT ratios, where PdT is the annual maximum precipitation for a duration d and a return period of T years, and PDT is the annual maximum daily precipitation for a return period of T years. In conjunction with these maps it is possible to estimate the maximum precipitation for durations between 3 h and 5 days.