WindINR: Latent-State INR for Fast Local Wind Query and Correction in Complex Terrain

2026-05-10

AI Summary

This work addresses the need for rapid, high-resolution wind field estimation at arbitrary locations and elevations over complex terrain, where downstream applications often cannot rely on dense, fixed-grid forecasts. To this end, we propose WindINR, a framework based on implicit neural representations (INRs) that maps static terrain, low-resolution background fields, and continuous query coordinates to high-resolution wind fields, while supporting online correction using sparse observations. The key innovation lies in decoupling a reusable global representation from a compact, sample-specific latent state; by optimizing only the latter, the method efficiently adapts the wind field without compromising continuous query capability. Experiments over the Senja region demonstrate significant improvements in local wind accuracy, with online correction on CPU achieving 2.6× the speed of full network fine-tuning, confirming the approach’s efficiency and robustness.

Abstract

Many downstream decisions in complex terrain require fast wind estimates at a small number of user-specified locations and heights for a given forecast valid time, rather than another dense forecast field on a fixed grid. We present WindINR, a latent-state implicit neural representation framework for continuous high-resolution local wind query and sparse-observation correction. WindINR maps static terrain descriptors, a low-resolution background field, and continuous query coordinates to a high-resolution wind state through a latent-conditioned decoder. To enable rapid inference-time correction, WindINR separates reusable representation learning from sample-specific latent-state correction. During training, a privileged encoder infers a reference latent state from high-resolution supervision, a deployable latent predictor estimates an initial latent state from inference-time inputs alone, and their discrepancies are summarized into a dataset-adaptive Gaussian prior over latent corrections. At inference time, within the WindINR module, network weights remain fixed and only the latent state is updated by minimizing a regularized correction objective using sparse observations and their uncertainty. In controlled OSSEs over the Senja region, including a UAV-aided approach scenario and random-observation robustness tests, WindINR improves local high-resolution wind estimates by updating only a compact latent state rather than the full network. The corrected representation remains continuously queryable at arbitrary coordinates and, in our CPU benchmark, yields about a $2.6\times$ online-correction speedup over full-network fine-tuning, suggesting a practical interface between kilometer-scale background products, sparse local observations, and wind queries in complex terrain.

Problem

• wind estimation
• complex terrain
• local wind query
• sparse observation correction
• high-resolution wind

Innovation

• Implicit Neural Representation
• Latent-State Correction
• Fast Wind Query
• Sparse Observation Assimilation
• Complex Terrain Modeling