Climate-model-informed deep learning of global soil moisture distribution

Abstract. We present a deep neural network (DNN) that produces accurate predictions of observed surface soil moisture, applying meteorological data from climate model. The was trained on daily satellite retrievals moisture the European Space Agency (ESA) Climate Change Initiative (CCI). predictors precipitation, temperature and humidity were simulated with ECHAM/MESSy atmospheric chemistry–climate model (EMAC). Our evaluation shows DNN are highly correlated observations, both spatially temporally, free bias. This offers an alternative for parameterisation schemes in models, especially simulations use but may not focus which we illustrate threshold wind speed mineral dust emissions. Moreover, can provide proxies missing values observations to produce realistic, comprehensive high-resolution global datasets. As approach presented here could be similarly used other variables study is proof concept basic expedient machine learning techniques modelling, motivate additional applications.