Improving regional weather forecasts with neural interpolation
This work addresses a domain-specific challenge in weather forecasting by proposing a method to enhance regional model accuracy, but it appears incremental as it builds on existing techniques like CNNs and residual networks without claiming major breakthroughs.
The paper tackles the problem of improving boundary data for regional weather models by designing a neural interpolation operator to map multi-scale dynamics between grid resolutions, using a simplified model to generalize results to dynamical cores.
In this paper we design a neural interpolation operator to improve the boundary data for regional weather models, which is a challenging problem as we are required to map multi-scale dynamics between grid resolutions. In particular, we expose a methodology for approaching the problem through the study of a simplified model, with a view to generalise the results in this work to the dynamical core of regional weather models. Our approach will exploit a combination of techniques from image super-resolution with convolutional neural networks (CNNs) and residual networks, in addition to building the flow of atmospheric dynamics into the neural network