Suchergebnisse

Abstract
Climate variability directly affects traditional low input and rain-fed farming systems, but few studies have paid attention retrospectively to the cropping system's ability to mitigate climate risk. This study analyzes the impacts of rainfall variability on farmers' seed variety losses over time, considering changes in smallholder farming systems. The cropping system dynamics, in favoring maize at the expense of sorghum and pearl millet, have induced an increasing risk of seed loss during drought. Combining ecological anthropology and climatology, a retrospective survey asking farmers about the period 1961–2006 was carried out at three altitudinal levels (750, 950, and 1100 m) on the eastern slope of Mount Kenya. Over that period, based on 3204 seed loss events reported orally and independently by 208 farmers, the probability to lose sorghum seed (0.056–0.065) was significantly lower than the probability to lose maize seed (0.071–0.087). All crops were more impacted by droughts than by very wet years. Seed loss probability increased for rainy seasons shorter than 50 days, with less than 28 rain days, and with a precipitation amount under 400 mm. Losses are almost linearly related to the frequency of rain days. Logistic regression confirmed that a change in cropping systems, favoring maize at the expense of sorghum and pearl millet, increased the risk of seed losses due to drought over the 46-yr period.

Ces dernières décennies, l'Afrique de l'Est souffre d'une longue et lente détérioration pluviométrique. De 2007 à 2011, la ville de Djibouti a enregistré un déficit de 73% des précipitations moyennes annuelles par rapport à la moyenne de 30 ans (1981-2010). Cette sécheresse sans précédent va mettre de nombreuses familles des zones rurales (de Djibouti, d'Ethiopie et de Somalie) sur la route. Cette migration a pour point de chute la ville de Djibouti où des quartiers neufs se créent, comme Buldhuqo. Totalement inexistant en 2004, il explose après 2009. Les derniers arrivants s'installent dans le fond de l'oued, là même où la zone était inondée en 2004 et 2009 suite à des pluies courtes mais intenses. La prochaine pluie extrême, l'exposition de ces populations précaires aux risques hydrologiques sera maximale. Cette analyse montre que le déficit pluviométrique actuel est exceptionnel et historiquement unique. Les migrations importantes de population induites par cette sécheresse vers la ville de Djibouti doivent être encadrées, notamment lors de leur installation spontanée. Il est nécessaire de diminuer l'exposition aux risques hydrologiques de ces populations sinistrées, pour que les victimes de la sécheresse ne soient pas emportées par un excès pluviométrique. ; Peer reviewed

An intercomparison of seven gridded rainfall products incorporating satellite data (ARC, CHIRPS, CMORPH, PERSIANN, TAPEER, TARCAT, TMPA) is carried out over Central Africa, by evaluating them against three observed datasets: (a) the WaTFor database, consisting of 293 (monthly records) and 154 (daily records) rain-gauge stations collected from global datasets, national meteorological services and monitoring projects, (b) the WorldClim v2 gridded database, and (c) a set of stations expanded from the FAOCLIM network, these two latter sets describing climate normals. All products fairly well reproduce the mean rainfall regimes and the spatial patterns of mean annual rainfall, although with some discrepancies in the east–west gradient. A systematic positive bias is found in the CMORPH product. Despite its lower spatial resolution, TAPEER shows reasonable skills.When considering daily rainfall amounts, TMPA shows best skills, followed by CMORPH, but over the central part of the Democratic Republic of the Congo, TARCAT is amongst the best products. Skills ranking is however different at the interannual time-scale, with CHIRPS and TMPAperforming best, though PERSIANN has comparable skills when only fully independent stations are used as reference. A preliminary study of Southern Hemisphere dry season variability, from the example of Kinshasa, shows that it is a difficult variable to capture with satellite-based rainfall products. Users should still be careful when using any product in the most data-sparse regions, especially for trend assessment. ; Peer reviewed