Fine-Scale Structure Reconstruction of Weather Radar Echoes via Blind Super-Resolution (opens in new tab)
Due to the unavailability of real high-resolution (HR) radar data, most super-resolution (SR) methods for weather radar rely on low-resolution (LR) samples synthesized using fixed and simplified degradation schemes. However, such excessively simplified schemes are inadequate to capture the complex degradations introduced by hardware and acquisition constraints. As a result, HR-LR training sets generated in this manner deviate from real-world observations, leading to poor generalization: model...
Read the original article