| dc.description.abstract | An accurate analysis of climate extremes is essential for impact assessment and devising appropriate adaptation measures. There is an urgent need to assess precipitation products in capturing the increasing occurrence of climate extremes. This study evaluates the ability of 20 observational datasets, including gauge-based, satellite-based and reanalyses, in representing different drought severity (moderate, severe and extreme drought) over Africa and its nine sub-regions at varying time scales (3-, 6- and 12-months) during 1983–2014. Drought is represented using the Standardized Precipitation Index (SPI). The results demonstrate that while most datasets are suitable for drought studies over the continent, the African Rainfall Climatology version 2 (ARC2) and Precipitation Estimation from Remotely Sensed Information using Artificial Neural Networks Climate Data Records (PERSIANN_CDR_v1r1) are less fitted for such investigations. Moreover, regions such as the Sahara (SAH), Central Africa (CAF) and North Eastern Africa (NEAF) show a larger disparity among the datasets, requiring more caution when selecting a dataset for use in such areas. Generally, the datasets present low agreement toward the lower end of the range (5–30%) because the individual datasets estimate varying drought severities at different grids and months. This is observed in the coefficient of variation of 20–25% of the datasets falling outside the ± 1 standard deviation range. Therefore, using an ensemble to represent the datasets remains an indispensable tool. The datasets present better agreement in the timing of drought events than the spatial distribution. The findings provide valuable insights into the complexity of drought assessment using diverse precipitation datasets. Furthermore, the results highlight the significance of considering spatial and temporal dimensions, as datasets may capture drought events at varying locations and times, revealing subtle variations in drought impact. | en_US |
