Particularly dry or wet boreal summer monsoon seasons are major hazards affecting societal vulnerability in India and Africa. Several factors affect monsoon rainfall amount and limit the understanding of possible linkages between monsoon variability and mean climate changes. Here we characterize the multiscale variability of Indian and West African monsoon rain from two simulations of the last 6,000 years. Changes in Earth's orbit cause long-term monsoon drying trend in India and Africa, but the Indian monsoon is more sensitive to anthropogenic CO 2 . Variability is characterized by two major ranges of chaotic variability, each related to specific ocean-atmosphere modes present throughout the period. Combination of random 50-to 500-and 2-to 20-year variability leads to large events occurring at millennium scale. However, the two regions exhibit opposite trends in rainfall variability due to changes in teleconnection with Pacific sea surface temperature for India and Atlantic sea surface temperature for West Africa at interannual to decadal timescales.

Over the recent decades, Extreme Precipitation Events (EPE) have become more frequent over the Sahel. Their properties, however, have so far received little attention. In this study the spatial distribution, intensity, seasonality and interannual variability of EPEs are examined, using both a reference dataset, based on a high-density rain-gauge network over Burkina Faso and 24 precipitation gridded datasets. The gridded datasets are evaluated in depth over Burkina Faso while their commonalities are used to document the EPE properties over the Sahel. EPEs are defined as the occurrence of daily-accumulated precipitation exceeding the all-day 99th percentile over a 1°x1° pixel. Over Burkina Faso, this percentile ranges between 21 and 33 mm day-1. The reference dataset show that EPEs occur in phase with the West African monsoon annual cycle, more frequently during the monsoon core season and during wet years. These results are consistent among the gridded datasets over Burkina Faso but also over the wider Sahel. The gridded datasets exhibit a wide diversity of skills when compared to the Burkinabe reference. The Global Precipitation Climatology Centre Full Data Daily version 1 (GPCC-FDDv1) and the Global Satellite Mapping of Precipitation gauge Reanalysis version 6.0 (GSMaP-gauge-RNL v6.0) are the only products that properly reproduce all of the EPE features examined in this work. The datasets using a combination of microwave and infrared measurements are prone to overestimate the EPE intensity, while infrared-only products generally underestimate it. Their calibrated versions perform than their uncalibrated (near-real-time) versions. This study finally emphasizes that the lack of rain-gauge data availability over the whole Sahel strongly impedes our ability to gain insights in EPE properties.