The effect of remapping techniques in assessing extreme precipitation events in CMIP6 models over West Africa

Quantitatively assessing climate simulations across models and observational datasets often requires mapping fields to a common spatial grid, a procedure commonly referred to as remapping.  This procedure can substantially alter key statistical properties of simulated variables, with impacts that depend on both the variables and the interpolation method under consideration. While some remapping techniques smooth extremes, others preserve the integral properties of the variable fields, leading to different conclusions in model evaluation.

A variable highly sensitive to these techniques is precipitation, due to its high spatial variability and intermittency.  In this study, we examine the effect of bilinear and conservative remapping techniques on precipitation statistics in CMIP6 simulations over West Africa. We quantify spatially explicit differences between original and remapped fields, with particular emphasis on changes in the representation of extreme precipitation events associated with floods and droughts.

Our results highlight that remapping-induced distortions can significantly influence assessments of extreme precipitation events and model performance, underscoring the need for careful selection and reporting of remapping strategies in climate analysis.