WRF-DL v1.0: A Bridge between WRF v4.3 and Deep Learning Parameterizations and its Application to Atmospheric Radiative Transfer

Abstract. In numerical weather prediction (NWP) models, physical parameterization schemes are the most computationally expensive components, despite being greatly simplified. In the past few years, an increasing number of studies have demonstrated that deep learning (DL) parameterizations of subgrid physics have the potential to accelerate and even outperform conventional physic-based schemes. However, as the DL models are commonly implemented using the DL libraries written in Python, very few DL-based parameterizations have been successfully integrated with NWP models due to the difficulty of embedding Python functions into Fortran-based NWP models. To address this issue, we developed a coupler to allow the DL-based parameterizations to be coupled with a widely used NWP model, i.e., the Weather and Research Forecasting (WRF) model. Similar to the WRF I/O methodologies, the coupler provides the options to run the DL model inference with exclusive processors or the same processors for WRF calculations. In addition, to demonstrate the effectiveness of the coupler, the DL-based radiation emulators are trained and coupled with the WRF model successfully.