Efficient Baseline for Quantitative Precipitation Forecasting in Weather4cast 2023

Abstract

Accurate precipitation forecasting is indispensable for informed decision-making across various industries. However, the computational demands of current models raise environmental concerns. We address the critical need for accurate precipitation forecasting while considering the environmental impact of computational resources and propose a minimalist U-Net architecture to be used as a baseline for future weather forecasting initiatives.

Abstract: arXiv descriptive metadata (CC0). Thank you to arXiv for use of its open access interoperability. This service was not reviewed or approved by, nor does it necessarily express or reflect the policies or opinions of, arXiv.