DEF: Diffusion-augmented Ensemble Forecasting

TL;DR

This paper introduces DEF, a diffusion-based method for creating initial condition perturbations that enhance stochastic weather forecasting by transforming deterministic models into probabilistic ones, improving long-term forecast accuracy.

Contribution

The paper presents a simple conditional diffusion model that generates structured perturbations, applicable iteratively, and controllable, enabling stochastic extensions of neural weather prediction models.

Findings

Reduced long-term forecast error with stochastic models

Generated meaningful and structured perturbations

Achieved improved predictive performance on ERA5 dataset

Abstract

We present DEF (\textbf{\ul{D}}iffusion-augmented \textbf{\ul{E}}nsemble \textbf{\ul{F}}orecasting), a novel approach for generating initial condition perturbations. Modern approaches to initial condition perturbations are primarily designed for numerical weather prediction (NWP) solvers, limiting their applicability in the rapidly growing field of machine learning for weather prediction. Consequently, stochastic models in this domain are often developed on a case-by-case basis. We demonstrate that a simple conditional diffusion model can (1) generate meaningful structured perturbations, (2) be applied iteratively, and (3) utilize a guidance term to intuitivey control the level of perturbation. This method enables the transformation of any deterministic neural forecasting system into a stochastic one. With our stochastic extended systems, we show that the model accumulates less error over long-term forecasts while producing meaningful forecast distributions. We validate our approach on the 5.625$^\circ$ ERA5 reanalysis dataset, which comprises atmospheric and surface variables over a discretized global grid, spanning from the 1960s to the present. On this dataset, our method demonstrates improved predictive performance along with reasonable spread estimates.