Data-driven weather forecast based on machine learning (ML) has experienced
rapid development and demonstrated superior performance in the global
medium-range forecast compared to traditional physics-based dynamical models.
However, most of these ML models struggle with accurately predicting extreme
weather, which is closely related to the extreme value prediction. Through
mathematical analysis, we prove that the use of symmetric losses, such as the
Mean Squared Error (MSE), leads to biased predictions and underestimation of
extreme values. To address this issue, we introduce Exloss, a novel loss
function that performs asymmetric optimization and highlights extreme values to
obtain accurate extreme weather forecast. Furthermore, we introduce a
training-free extreme value enhancement strategy named ExEnsemble, which
increases the variance of pixel values and improves the forecast robustness.
Combined with an advanced global weather forecast model, extensive experiments
show that our solution can achieve state-of-the-art performance in extreme
weather prediction, while maintaining the overall forecast accuracy comparable
to the top medium-range forecast models.

基于机器学习的数据驱动天气预报在全球中期预报方面表现出优于传统基于物理的动力模型的快速发展和卓越性能。通过数学分析，我们证明了使用对称损失如均方误差（MSE）会导致偏差预测和极值低估。为了解决这个问题，我们引入了 Exloss，一种执行非对称优化并突出显示极值以获得准确极端天气预报的新型损失函数。此外，我们引入了一种无需训练的极值增强策略 ExEnsemble，它增加了像素值的方差并提高了预报的鲁棒性。结合先进的全球天气预报模型，广泛的实验表明我们的解决方案在极端天气预测方面可以达到与顶级中期预报模型相当的准确性。

ExtremeCast：全球天气预报中的极值预测提升

ExtremeCast: Boosting Extreme Value Prediction for Global Weather  Forecast

We present FengWu, an advanced data-driven global medium-range weather
forecast system based on Artificial Intelligence (AI). Different from existing
data-driven weather forecast methods, FengWu solves the medium-range forecast
problem from a multi-modal and multi-task perspective. Specifically, a deep
learning architecture equipped with model-specific encoder-decoders and
cross-modal fusion Transformer is elaborately designed, which is learned under
the supervision of an uncertainty loss to balance the optimization of different
predictors in a region-adaptive manner. Besides this, a replay buffer mechanism
is introduced to improve medium-range forecast performance. With 39-year data
training based on the ERA5 reanalysis, FengWu is able to accurately reproduce
the atmospheric dynamics and predict the future land and atmosphere states at
37 vertical levels on a 0.25{\deg} latitude-longitude resolution. Hindcasts of
6-hourly weather in 2018 based on ERA5 demonstrate that FengWu performs better
than GraphCast in predicting 80\% of the 880 reported predictands, e.g.,
reducing the root mean square error (RMSE) of 10-day lead global z500
prediction from 733 to 651 $m^{2}/s^2$. In addition, the inference cost of each
iteration is merely 600ms on NVIDIA Tesla A100 hardware. The results suggest
that FengWu can significantly improve the forecast skill and extend the
skillful global medium-range weather forecast out to 10.75 days lead (with ACC
of z500 > 0.6) for the first time.

该研究论文介绍了一个基于人工智能的先进数据驱动的全球中程天气预报系统 FengWu，该系统从多模态和多任务的角度解决介质范围的天气预报问题，可在 0.25 度经纬度分辨率下预测未来 37 个垂直层次的地面和大气状态，与现有天气预报方法相比，FengWu 性能更佳。