Adding machine learning to the MIP toolkit: Predictor importance for
hydrological fluxes of Global Hydrological and Land Surface Models
Abstract
Global Hydrological and Land Surface Models (GHM/LSMs) embody numerous
interacting predictors and equations, complicating the diagnosis of
primary hydrological relationships. We propose a model diagnostic
approach based on Random Forest feature importance to detect the input
variables that most influence simulated hydrological processes. We
analyzed the JULES, ORCHIDEE, HTESSEL, SURFEX and PCR-GLOBWB models for
the relative importance of precipitation, climate, soil, land cover and
topographic slope as predictors of simulated average evaporation,
runoff, and surface and subsurface runoffs. The machine learning model
could reproduce GHM/LSMs outputs with a coefficient of determination
over 0.85 in all cases and often considerably better. The GHM/LSMs
agreed precipitation, climate and land cover share equal importance for
evaporation prediction, and mean precipitation is the most important
predictor of runoff. However, the GHM/LSMs disagreed on which features
determine surface and subsurface runoff processes, especially with
regards to the relative importance of soil texture and topographic
slope.