Africa has a paucity of long term reliable meteorological ground station data and reanalysis products are used to provide the climate estimations that are important for climate change projections. In this paper monthly observed precipitation records in the Logone catchment of the Lake Chad Basin are used to evaluate the performance of two global reanalysis products: the Climate Forecasting System Reanalysis (CFSR) and ERA Interim datasets.
The two reanalysis products reproduced the monthly, annual and decadal cycle of precipitation and variability relatively accurately albeit with some discrepancies. The catchment rainfall gradient was also well captured by the two products. There are good correlations between the reanalysis and rain gauge datasets although significant deviations exist, especially for CFSR. Both reanalysis products overestimated rainfall in 68% of the rain gauge stations. ERA Interim produced the lowest bias and mean absolute error with average values of 2% and 6.5 mm month−1 respectively compared to 15% and 34 mm month−1 for the CFSR. However, both reanalysis products systematically underestimated annual rainfall in the catchment during the period 1997–2002 for ERA Interim and 1998–2000 for CFSR. This research demonstrates that evaluating reanalysis products in remote areas like the Logone catchment enables users to identify artefacts inherent in reanalysis datasets. This will facilitate improvements in certain aspects of the reanalysis forecast model physics and parameterization to improve reanalysis dataset quality.
This study concludes that the application of each reanalysis product in the catchment will depend on the purpose for which it is to be used and the spatial scale required.
CORE (COnnecting REpositories)
CORE (COnnecting REpositories)