Evaluation and comparison of two ERA5 reanalysis variants for describing precipitation trends between 1950 and 2016 in West-Africa
- René Bodjrènou
, - Luc Ollivier Sintondji,
- Aymar Yaovi Bossa
René Bodjrènou
National Institute of Water, University of Abomey-Calavi, Benin
Corresponding Author:renebodjrenou@gmail.com
Author ProfileLuc Ollivier Sintondji
National Institute of Water, University of Abomey-Calavi, Benin
Aymar Yaovi Bossa
National Institute of Water, University of Abomey-Calavi, Benin
Abstract
Today, reanalyses are a viable alternative to ensure the continuity of hydrological studies and to assess climate variability in regions where the availability and/or quality of observational data is problematic. However, their performance and shortcomings need to be fully understood before their use can be recommended. In this study, the ERA5 (0.25° x 0.25° resolution) and ERA5-Land (0.1° x 0.1° resolution) reanalyses are evaluated and compared over West Africa, using point station data (measured in situ). The reanalysis time series is obtained by selecting the pixel closest to the point station where it was evaluated/compared. The results show that the ERA5-Land reanalysis provides good correlation at monthly (~0.79) and annual (0.60) scales. For the ERA5 reanalysis, the average correlation is approximatively similar (0.79 on a monthly scale and 0.59 on an annual scales). On a daily scale, both reanalyses show low precipitation correlation (between 0.18 and 0.39 for ERA5-Land vs. 0.18 and 0.34 for ERA5). ERA5 showed the lowest percentage of annual precipitation bias (~21%) compared to ERA5-Land (~23%), indicating that its average closer to observations. In terms of annual precipitation trends, we find a variation between -3.76 and 6.78 for ERA5, between -3.82 and 3.19 for ERA5-Land and between -9.45 and 1.52 for the observations. For 713 stations, observations show positive trends at 12.5%; ERA5 shows positive trends at 17.9% and ERA5-Land at 16.1%. The precipitation trends of the reanalyses mostly have the same sign as those of the point stations (75.0% of the stations have trends with the same sign as ERA5 compared to 76.2% for ERA5-Land). Finally, we consider it useful to adjust the biases of both reanalyses to further improve their ability to represent the climate in West Africa. 18 Jan 2025Submitted to 2024 AGU Annual Meeting Preprint Collection on ESS Open Archive 19 Jan 2025Published in 2024 AGU Annual Meeting Preprint Collection on ESS Open Archive
