Daily rainfall reconstruction in East Africa: application to the rainy season characteristics.

Abstract

The reconstruction of daily rainfall data at the station scale can be necessary to study the climate change or other applied climatology work. This is the case in tropical area to analyze rainy season characteristics (onset and cessation dates, duration, etc.). A classical methodological approach based on multiple linear stepwise regression is applied to reconstitute daily rainfall data. This methodology is developed to take advantage of all the available data and included a bias correction. Performances are assessed from various descriptors allowing to describe the temporal structure of the African rainy seasons. It is tested over East Africa, a challenging area where the complex topography strongly constrains spatial and temporal rainfall distribution. Results show that the rainy days occurrence and the daily rainfall amount are correctly reproduced in wettest and well-documented areas. Although the reconstruction skill is sometimes modest at the daily time step, satisfactory skill scores are obtained when the reconstruction method is evaluated through seasonal and intra-seasonal characteristics such as the total rainfall, the number of rain days and the dates of rainy season onset and cessation. The impact of temporal series – in-filling is directly associated with the amount and temporal distribution of missing values in the original series but it was not possible to determine how much missing values may lead to a significant decrease of the quality of the rainy season characteristics decreases. The temporal distribution of missing values has also an impact on the quality of the reconstructed rainy season so that the longer the sequences of missing values, the poorer the skill of the reconstruction.