Abstract
The African continent has a very low density of rain gauge stations, and long time-seriesfor recent years are often limited and poorly available. In the context of global change, it is veryimportant to be able to characterize the spatio-temporal variability of past rainfall, on the basis ofdatasets issued from observations, to correctly validate simulations. The quality of the rainfall datais for instance of very high importance to improve the efficiency of the hydrological modeling,through calibration/validation experiments.The HydroSciences Montpellier Laboratory (HSM) has a long experience in collecting andmanaging hydro-climatological data. Thus, HSM had initiated a program to elaborate a referencedataset, in order to build monthly rainfall grids over the African continent, over a period of 60 years(1940/1999). The large quantity of data collected (about 7,000 measurement points were used in thisproject) allowed for interpolation using only observed data, with no statistical use of a referenceperiod. Compared to other databases that are used to build the grids of the Global HistoricalClimatology Network (GHCN) or the Climatic Research Unit of University of East Anglia, UK (CRU),the number of available observational stations (a was significantly much higher, including the end ofthe century when the number of measurement stations dropped dramatically, everywhere.Inverse distance weighed (IDW) was the chosen method to build the 720 monthly grids and amean annual grid, from rain gauges. The mean annual grid was compared to the CRU grid. The gridswere significantly different in many places, especially in North Africa, Sahel, the horn of Africa, andthe South Western coast of Africa, with HSM_SIEREM data (database HydroSciencesMontpellier_Système d’Information Environnementales pour les Ressources en Eau et leurModélisation) being closer to the observed rain gauge values. The quality of the grids computed waschecked, following two approaches—cross-validation of the two interpolation methods, ordinarykriging and inverse distance weighting, which gave a comparable reliability, with regards to theobserved data, long time-series analysis, and analysis of long-term signals over the continent,compared to previous studies. The statistical tests, computed on the observed and gridded data,detected a rupture in the rainfall regime around 1979/1980, on the scale of the whole continent; thiswas congruent with the results in the literature. At the monthly time-scale, the most widely observedsignal over the period of 1940/1999, was a significant decrease of the austral rainy season betweenMarch and May, which has not earlier been well-documented. Thus, this would lead to a furtherdetailed climatological study from this HSM_SIEREM database.
Highlights
On a global scale, the first climate observations began during the second part of the 19th century.Among these climate variables, rainfall data sets have been the most complete, since the beginning of the 1950s [1]
Of data collected during a long period, mainly in Western and Central Africa, by the ORSTOM and i.e., it showed a boreal tropical rainy season but changed much more during
This paper is the over resultAfrica, of a long duration of workshowing on the collection, criticism, and assessment of a in describing rainfall with some regions very important differences, especially very and high-quality database of rainfall
Summary
The first climate observations began during the second part of the 19th century Among these climate variables, rainfall data sets have been the most complete, since the beginning of the 1950s [1]. To study rainfall changes at the scale of the whole African continent, rainfall grids can be downloaded from several institutions, but it is well-known that the African continent is less documented than other parts of the world [5,6] This has led to some discrepancies between rainfall calculated from different databases, as has been shown for Western Africa, by Mahé et al [7]. For instance, for Western and Central Africa, rainfall data are not very well-documented [10,11]
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
