Evaluation of Satellite Derived Rainfall Data for Mt Kenya Region

Abstract

Rainfall information is crucial more so for an economy that heavily relies on rain fed agriculture like Kenya. However, the spatial distribution of rain gauges is low and is only representative of location of measurement. Satellite based precipitation products can be used to supplement the data acquired from the ground stations available. There is need of validation of the satellite products by establishing a level of confidence and subsequently the probability of use of the data in areas where ground observations are not available. The performance of RFE V2 and GPCP 1DD was evaluated against observations for the region around Mt. Kenya. The aim was to determine the level of correlation between the satellite-based precipitation and the ground observations. Based on the analysis, resolve on the use of the data in cases where ground observations are not available. Observation data from six ground stations for the period between 2001 and 2012 were used in the analysis. Over estimation or underestimation was initially calculated. Weights were assigned based on their range of deviation from the observations. The coefficient of determination was also computed and weights assigned based on their precision to the observed values. Weights from the two parameters were then multiplied to give a combined goodness of fit. RFE performed better in most of the stations. An analysis of the performance of the satellite estimates should be done using daily observation data as opposed to the monthly data used.