Development of a Convolutional Neural Network to Predict Wintertime Orographic Precipitation Gradients of the Western US

Despite remarkable advances over the past few decades, many current operational modeling systems lack adequate grid spacing to sufficiently resolve fine-scaled precipitation patterns in regions of complex terrain. As a result, statistical or dynamical downscaling is often employed to produce higher-resolution precipitation forecasts. However, statistical downscaling can overlook non-stationary atmospheric relationships and often relies on a monthly or annual climatology, whereas dynamical downscaling features physically-based calculations, but is computationally expensive.

This study uses machine learning to predict orographic precipitation gradients (OPG) based on synoptic conditions to downscale precipitation over the western continental United States (CONUS). A machine learning algorithm can account for the non-linearity of the atmosphere while being computationally fast. Based on prior work by Bohne et al. (2020), we divide the western CONUS terrain into facets based on regional terrain orientation and use linear regression to quantify daily OPGs within each facet. Linear regression is applied to observed precipitation from the Global Historical Climatology Network-Daily dataset from 1979 to 2018. These daily OPGs frequently vary from the climatological means due to the effects of large-scale circulation patterns, suggesting the potential for a machine learning algorithm to predict OPG based on these patterns.

This presentation will present current results in developing a convolutional neural network (CNN) to predict OPG. Our CNN is trained using the ECMWF ERA5 Reanalysis and daily OPG from all suitable facets in the western CONUS. Current training results for the Northern Rockies region show the CNN model accounting for 50% of the OPG variance with a mean absolute error of about 2.5 mm (1000 m)^-1. The development of a combination CNN-OPG model enables insights into the relationships between large-scale circulations and western CONUS OPG and enables novel approaches for prediction.