Ensemble‐based Reconstructed Forcing (ERF) for regional climate modeling: Attaining the performance at a fraction of cost

Abstract Multimodel ensemble (MME) is considered the most reliable for both present‐day and future climate simulations. However, the conventional MME approach can create physical inconsistencies between climate variables. It can also be computationally inhibiting when using regional climate models (RCMs) due to uncertainties originating from both RCMs and the driving global climate models (GCMs) which must be accounted for. Here we propose the Ensemble‐based Reconstructed Forcing (ERF) approach which derives the RCM's initial and boundary conditions from the ensemble average of multiple GCMs. Using ensembles of up to six GCMs as examples, we show that the ERF approach with a single RCM integration performs comparably to the MME average in bias reduction. The ERF simulation provides unaltered solutions to the RCM's underlying equations, making it theoretically superior to the conventional MME approach which averages multiple solutions and may result in a physically implausible climate state.