downscaling

This paper introduces CLAM, a method for estimating localized causal effects from coarse-resolution data by jointly learning causal mechanisms and a disaggregation mapping, with applications in public health and environmental policy.

本文介绍了EddyFlow，一个用于公里级海面温度降尺度的深度学习框架，它在预测精度、尺度相关结构和区域泛化之间取得平衡。该框架在多个海洋区域实现了强大的零样本性能和近乎理想的频谱保真度。