资源环境科技发展态势分析平台

Relationships between daily precipitation and daily maximum and minimum temperature (Tx and Tn, respectively) are analyzed at station level over the Greater Horn of Africa (GHA). Rainfall occurrence is associated with either above normal Tn (mostly in cool highland areas) or below normal Tn (especially lowland, hot environments and early parts of the rainy season). Tx generally displays a more consistent response to rainfall occurrence, with cooling peaking 1 day after the rainfall event. However there is often a persistence of this cooling several days after the rainfall event, and the amplitude of the cooling is also greater for heavy rainfall events. These temperature anomalies are thought to be a response to cloudiness (concurrent reduced Tx and concurrent enhanced Tn) and soil moisture (reduced Tx and Tn, suggested to reflect evaporative cooling). These relationships are of relevance to the interpretation of temperature trends. From 1973 to 2013, the GHA shows a clear warming signal, for both Tn (+0.20 to +0.25 A degrees C/decade depending on seasons) and Tx (+0.17 to +0.22 A degrees C/decade). Rainfall shows both negative (mostly between February and July) and positive trends (mostly in October-December). Given the superimposition of temperature and rainfall trends in parts of the GHA and the covariations between daily rainfall and both Tx and Tn, regression models are used to extract the rainfall influence on temperature, accounting for lag effects up to 4 days. The daily residuals from these models are used to depict temperature variations independent from precipitation variations. At some stations, trends computed on these residuals noticeably differ from the raw Tx trends. When averaged across the GHA, these effects do not exceed -0.06 to +0.03 A degrees C/decade (depending on the month) for Tx, and are marginal for Tn, thus do not strongly modify the magnitude of the warming in the last 40 years. Nevertheless, these results show that precipitation-temperature relationships must be addressed when analyzing temperature changes.