Great effort is made to address heat waves (HWs) in developed countries because of their devastating impacts on society, economy, and environment. However, HWs are still understudied over developing countries. This is particularly true in West Africa, and especially in the Sahel, where temperatures recurrently reach critical values, such as during the 2010 HW event in the western Sahel. This work aims at characterizing the Sahelian HWs during boreal spring seasons (April–May–June) and understanding the mechanisms associated with such extreme events. Over the last three decades, Sahelian HWs have been becoming more frequent, lasting longer, covering larger areas, and reaching higher intensities. The physical mechanisms associated with HWs are examined to assess the respective roles of atmospheric dynamics and radiative and turbulent fluxes by analyzing the surface energy budget. Results suggest that the greenhouse effect of water vapor is the main driver of HWs in the western Sahel, increasing minimum temperatures by enhanced downward longwave radiation. Atmospheric circulation plays an important role in sustaining these warm anomalies by advecting moisture from the Atlantic Ocean and the Guinean coasts into the Sahel. Maximum temperature anomalies are mostly explained by increased downward shortwave radiation due to a reduction in cloud cover. Interannual variability of HWs is affected by the delayed impact of El Niño–Southern Oscillation (ENSO), with anomalous temperature warming following warm ENSO events, resulting from an amplified water vapor feedback.

Trends in daily maximum (TX) and minimum (TN) temperatures and indices of warm extremes are studied in tropical North Africa, west of the eastern African highlands, from 1961 to 2014. The analysis is based on the concatenation and cross-checking of two observed databases. Due to the large number of missing entries (~25%), a statistical infilling using probabilistic principal component analysis was applied. Averaged over 90 stations, the linear trends of annual mean TX and TN equal respectively +0.021°C/yr and +0.028°C/yr. The frequency of very hot days (TX > 35°C) and tropical nights (TN > 20°C), as well as the frequency of daily TX and TN above the 90th percentile (p90) ("warm days" and "warm nights"), roughly follows the variations of mean TX and TN, respectively. Heat spells of TX or TN > p90 are often short (usually <2-3 days), and the interannual variation of their mean duration is noisier than for the other indices. Nevertheless, heat spells tend to last longer, with almost constantly positive anomalies since the mid-1990s. The trends in March-June, the warmest season across the Sahelian and Sudanian belts, show similar variations as annual means. Overall, the local-scale warming in annual temperatures, and in March-June, may be viewed merely as a simple shift of the probability distribution function of daily TX and TN. The correlations between the thermal indices and the 2 m temperatures suggest that the low-frequency (>8 years) variations may be viewed as a regional-scale fingerprint of the global warming, with largest correlations in the tropical Atlantic and Indian basins, while the high-frequency (<8 years) variations should be mostly viewed as a delayed remote impact of El Niño-Southern Oscillation (ENSO) events over the region, with warm (cold) anomalies tending to follow warm (cold) ENSO events.Several studies have analyzed the time variation of climate extremes at the global scale [Groisman et al., 1999;Frich et al.,