Post-processing Multi-Model Medium-Term Precipitation Forecasts Using Convolutional Neural Networks

TL;DR

This study explores using convolutional neural networks to enhance medium-term precipitation forecasts by transforming combined model outputs into probabilistic images, but CNNs did not outperform traditional methods.

Contribution

It introduces a CNN-based approach for post-processing precipitation forecasts that combines multiple model outputs into probabilistic images, with an ablation analysis on model contributions.

Findings

CNNs did not outperform regularized logistic regression.

Combining global and regional model forecasts improved performance.

Fully convolutional neural networks can transform forecast images into probabilistic outputs.

Abstract

The goal of this study was to improve the post-processing of precipitation forecasts using convolutional neural networks (CNNs). Instead of post-processing forecasts on a per-pixel basis, as is usually done when employing machine learning in meteorological post-processing, input forecast images were combined and transformed into probabilistic output forecast images using fully convolutional neural networks. CNNs did not outperform regularized logistic regression. Additionally, an ablation analysis was performed. Combining input forecasts from a global low-resolution weather model and a regional high-resolution weather model improved performance over either one.