Problem
Research questions and friction points this paper is trying to address.
Climate downscaling
Probabilistic modeling
Lightweight models
Machine learning
Lightweight Probabilistic Downscaling from a Deterministic Base Model
🤖 AI Summary
This study addresses the challenge of balancing computational overhead with distributional fidelity in climate data downscaling by proposing a lightweight probabilistic downscaling model based on an improved U-Net architecture. Methodologically, this work introduces a novel two-stage curriculum learning strategy that integrates deterministic pretraining with probabilistic fine-tuning, and conducts systematic evaluations using the CORDEX-ML-Bench benchmark. Experimental results demonstrate that the proposed model outperforms existing state-of-the-art methods in terms of RMSE while significantly reducing computational costs and effectively improving the accuracy of probabilistic distribution fitting. Ultimately, this approach establishes an efficient and reliable new paradigm for high-resolution climate prediction.
📝 Abstract
Climate data downscaling is the task of increasing the spatial resolution of climate data, typically by generating fine-resolution regional climate data from coarse global model output. Recent machine learning (ML) work in the related task of weather forecasting has seen significant improvements due to newly devised training methods and architectural components, but these have not yet benefited downscaling. We adapt two of these methods to create a family of lightweight probabilistic ML downscaling models built on a modified U-Net backbone and evaluate them on the CORDEX-ML-Bench suite for daily maximum temperature and precipitation across three geographic regions: the Alps, New Zealand and South Africa. We find that a two-stage training curriculum, combining deterministic pretraining with probabilistic tuning, transfers well to downscaling, beating the state-of-the-art for RMSE. Our work provides an advancement towards lightweight, probabilistic downscaling models, reducing the current trade-off between computational intensity and distributional fit.
Innovation
Methods, ideas, or system contributions that make the work stand out.
Probabilistic Downscaling
Lightweight Model
Two-stage Training Curriculum
Modified U-Net
Climate Data
🔎 Similar Papers
No similar papers found.
