面向气候降尺度的物理约束 Rectified Flow:投影可验证什么?
Towards Physics-Constrained Rectified Flow for Climate Downscaling: What Projections Can Certify?
AI 导读
提出物理约束 Rectified Flow(PC-RF)气候降尺度方法,训练惩罚项针对风散度、负降水和质量漂移,采样后投影可将非负性违反率(NVR)与质量守恒误差(MCE)精确降至零。
正文 · 原文
Abstract:Statistical downscaling restores the regional detail behind flooding and crop stress, but diffusion and flow-matching generators can produce sharp fields that break the conservation laws of their coarse input. Physics-Constrained Rectified Flow (PC-RF) is a conditional flow-matching downscaler. Its training penalties target wind divergence, negative precipitation, and mass drift from the coarse total, and a projection after sampling removes the last two exactly. On one corrupted synthetic field with no trained network, the projection drives the Non-negativity Violation Rate (NVR) from 0.236 and the Mass-Conservation Error (MCE) from 0.184 to zero, and the penalties lower the Divergence Error (DE) by about 30%. On ERA5 the projection gives PC-RF and RF-base, the penalty-free rectified-flow baseline, an NVR of exactly zero and an MCE of zero up to float32 rounding. The penalties add no perceptual gain that we can separate from training budget or run-to-run variation. At a matched budget on 2019-2021, PC-RF has no edge over RF-base (SSIM 0.939 vs. 0.942, PSNR 31.81 vs. 31.79, CRPS 0.771 vs. 0.772). Our unprojected CorrDiff reimplementation beats both flow models on PSNR (28.38) and CRPS (1.092) on 2020, yet 45% of its ensemble-mean precipitation cells are negative. On ERA5 a sign error in the divergence operator made that penalty and DE act on a deformation-like term. The fine reference is interpolated from ERA5 and accumulates precipitation over a longer window, so the perceptual scores cannot establish downscaling skill. NVR and MCE need one pass and no fine reference, so we argue they should run first as pass-or-fail checks, with DE as a diagnostic.
| Comments: | Accepted to AIAS+ 2026, ACM ICPS proceedings. 9 pages |
| Subjects: | Artificial Intelligence (cs.AI) |
| ACM classes: | I.2.6; J.2 |
| Cite as: | arXiv:2611.22283 [cs.AI] |
| (or arXiv:2611.22283v1 [cs.AI] for this version) | |
| https://doi.org/10.48550/arXiv.2611.22283 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Arun Sharma [view email]
[v1]
Tue, 17 Nov 2026 06:33:34 UTC (246 KB)
来源:arXiv cs.AI · arxiv.org