标题： VA-MoE：用于增量天气预报的变量自适应专家混合模型 显示英文标题

标题： VA-MoE: Variables-Adaptive Mixture of Experts for Incremental Weather Forecasting

摘要： 本文提出了变量自适应专家混合（VAMoE），这是一种新颖的增量天气预报框架，能够实时动态适应实时数据中的演变时空模式。 传统的天气预测模型常常在高昂的计算开销以及随着新观测数据到达而持续更新预测的需求上面临挑战。 VAMoE通过利用专家的混合架构来解决这些问题，其中每个专家专门用于捕捉大气变量（温度、湿度、风速）的不同子模式。 此外，所提出的方法采用变量自适应门控机制，根据输入上下文动态选择和组合相关专家，从而实现高效的知识蒸馏和参数共享。 这种设计显著降低了计算开销，同时保持了高预测准确性。 在真实世界ERA5数据集上的实验表明，VAMoE在短期（1天）和长期（5天）预测任务中表现与最先进模型相当，仅使用约25%的可训练参数和50%的初始训练数据。 显示英文摘要

摘要： This paper presents Variables Adaptive Mixture of Experts (VAMoE), a novel framework for incremental weather forecasting that dynamically adapts to evolving spatiotemporal patterns in real time data. Traditional weather prediction models often struggle with exorbitant computational expenditure and the need to continuously update forecasts as new observations arrive. VAMoE addresses these challenges by leveraging a hybrid architecture of experts, where each expert specializes in capturing distinct subpatterns of atmospheric variables (temperature, humidity, wind speed). Moreover, the proposed method employs a variable adaptive gating mechanism to dynamically select and combine relevant experts based on the input context, enabling efficient knowledge distillation and parameter sharing. This design significantly reduces computational overhead while maintaining high forecast accuracy. Experiments on real world ERA5 dataset demonstrate that VAMoE performs comparable against SoTA models in both short term (1 days) and long term (5 days) forecasting tasks, with only about 25% of trainable parameters and 50% of the initial training data.