Oya: Deep Learning for Accurate Global Precipitation Estimation

2025-11-13

AI Summary

To address the low accuracy of sub-daily precipitation estimation over sparsely gauged regions in the Global South, this paper proposes a two-stage deep learning retrieval framework leveraging full-spectrum visible–infrared (VIS–IR) data from geostationary satellites. Stage one employs a U-Net architecture for rain/no-rain binary classification to mitigate class imbalance; stage two applies another U-Net—guided by the binary mask—to perform quantitative precipitation estimation. The model is supervised by GPM CORRA v07 ground-truth precipitation and pre-trained on IMERG-Final data to enhance generalizability. By fusing multi-satellite geostationary observations, it achieves near-global coverage. Evaluations demonstrate that our method significantly outperforms state-of-the-art satellite precipitation products in both regional and global benchmarks, particularly at temporal scales of 30–180 minutes. It delivers high spatiotemporal resolution and robustness, offering improved precipitation data for hydrological monitoring and nowcasting applications.

Technology Category

• Machine Learning: Calibration & Uncertainty Quantification
• Computer Vision: Remote Sensing / Geospatial AI
• Intelligent Robots: State Estimation

Application Category

• Search and Retrieval-Augmented AI: Vertical and domain-specific search
• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Web Mining and Content Analysis: Large pretrained models with web data

Abstract

Accurate precipitation estimation is critical for hydrological applications, especially in the Global South where ground-based observation networks are sparse and forecasting skill is limited. Existing satellite-based precipitation products often rely on the longwave infrared channel alone or are calibrated with data that can introduce significant errors, particularly at sub-daily timescales. This study introduces Oya, a novel real-time precipitation retrieval algorithm utilizing the full spectrum of visible and infrared (VIS-IR) observations from geostationary (GEO) satellites. Oya employs a two-stage deep learning approach, combining two U-Net models: one for precipitation detection and another for quantitative precipitation estimation (QPE), to address the inherent data imbalance between rain and no-rain events. The models are trained using high-resolution GPM Combined Radar-Radiometer Algorithm (CORRA) v07 data as ground truth and pre-trained on IMERG-Final retrievals to enhance robustness and mitigate overfitting due to the limited temporal sampling of CORRA. By leveraging multiple GEO satellites, Oya achieves quasi-global coverage and demonstrates superior performance compared to existing competitive regional and global precipitation baselines, offering a promising pathway to improved precipitation monitoring and forecasting.

Problem

• Develops deep learning algorithm for global precipitation estimation using satellite data
• Addresses data imbalance in rain detection with dual U-Net architecture
• Improves accuracy over existing products by leveraging full VIS-IR spectrum

Innovation

• Uses full spectrum VIS-IR from geostationary satellites
• Two-stage U-Net models for detection and estimation
• Trained with GPM CORRA data and IMERG pre-training