Meteorologically consistent bias correction of climate time series for agricultural models

Conventional bias correction of simulated climate time series for impact models is done separately for climate variables and hence leads to inconsistencies between them. However, agricultural models mostly use several variables, and meteorological consistency is essential. The present work points out meteorological inconsistency due to quantile mapping and describes a new method of consistent bias correction by an optimization approach. Time series of hourly precipitation and global radiation from the regional model REMO5.7 (Run UBA C20/A1B_1) were corrected with site observations from the German Meteorological Service. The results urge to check conventionally corrected series for consistency before using them for multidimensional models. Here, quantile mapping resulted in underestimation of diffuse radiation at hours with precipitation. This deficit was minimized by the developed procedure.