- Satellite precipitation estimates are crucial for managing climate-related risks such as droughts and floods. However, these datasets often contain systematic errors due to the observation methods used. The accuracy of these estimates can be enhanced by integrating spatial and temporal resolution data from in situ observations. Nevertheless, the accuracy of the merged dataset is influenced by the density and distribution of rain gauges, which can vary regionally. This paper presents an approach to improve satellite precipitation data (SPD) over Burkina Faso. Two bias correction methods, Empirical Quantile Mapping (EQM) and Time and Space-Variant (TSV), have been applied to the SPD to yield a bias-corrected dataset for the period 1991–2020. The most accurate bias-corrected dataset is then combined with in situ observations using the Regression Kriging (RK) method to produce a merged precipitation dataset. The findings show that both bias correction methods achieve similar reductions inSatellite precipitation estimates are crucial for managing climate-related risks such as droughts and floods. However, these datasets often contain systematic errors due to the observation methods used. The accuracy of these estimates can be enhanced by integrating spatial and temporal resolution data from in situ observations. Nevertheless, the accuracy of the merged dataset is influenced by the density and distribution of rain gauges, which can vary regionally. This paper presents an approach to improve satellite precipitation data (SPD) over Burkina Faso. Two bias correction methods, Empirical Quantile Mapping (EQM) and Time and Space-Variant (TSV), have been applied to the SPD to yield a bias-corrected dataset for the period 1991–2020. The most accurate bias-corrected dataset is then combined with in situ observations using the Regression Kriging (RK) method to produce a merged precipitation dataset. The findings show that both bias correction methods achieve similar reductions in RMS error, with higher correlation coefficients (approximately 0.8–0.9) and a normalized standard deviation closer to 1. However, EQM generally demonstrates more robust and consistent performance, particularly in terms of correlation and RMS error reduction. On a monthly scale, the superiority of EQM is most evident in June, September, and October. Following the merging process, the final dataset, which incorporates satellite information in addition to in situ observations, demonstrates higher performance. It shows improvements in the coefficient of determination by 83%, bias by 11.4%, mean error by 96.7%, and root-mean-square error by 95.5%. The operational implementation of this approach provides substantial support for decision-making in regions heavily reliant on rainfed agriculture and sensitive to climate variability. Delivering more precise and reliable precipitation datasets enables more informed decisions and significantly enhances policy-making processes in the agricultural and water resources sectors of Burkina Faso.…
