Downscaling: UNet model¶
This video shows two versions of the same precipitation field, one from a physical model (left) and one from a machine-learning UNet model trained on the physical model outputs (right). The aim is to judge the quality of the ML model - how well does it reproduce the physical model results.
The precipitation field is a sequence of daily averages. For the physical model that’s a daily average of high-frequency model output - the ML model calculates daily averages directly. To make a video, the daily averages are interpolated to hourly values, and then the hourly values are shown as a sequence of frames.
Data source is the same as that used in the diffusion model example.
