MeltwaterBench: Deep learning for spatiotemporal downscaling of surface meltwater

2025-12-12

AI Summary

Monitoring the spatiotemporal dynamics of meltwater from the Greenland Ice Sheet is hindered by the fundamental trade-off between temporal and spatial resolution in remote sensing data. To address this, we propose the first spatiotemporal joint super-resolution framework specifically designed for glacial meltwater mapping. Our approach fuses synthetic aperture radar (SAR), passive microwave, and digital elevation model (DEM) data within a multimodal deep learning architecture—UNet/DeepLabv3+—trained under SAR supervision and refined using regional climate model outputs to reconstruct daily meltwater distributions at 100 m resolution. We introduce MeltwaterBench, the first multi-source, temporally aligned benchmark dataset for meltwater mapping. Validation over Helheim Glacier achieves 95% accuracy—exceeding conventional methods by over 10 percentage points. The source code and the complete time-aligned dataset are publicly released.

Technology Category

• Computer Vision: Remote Sensing / Geospatial AI
• Machine Learning: Multimodal Learning
• Planning, Routing, and Scheduling: Optimization of Spatio-temporal Systems

Application Category

• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Security and Privacy: Data transparency and provenance

Abstract

The Greenland ice sheet is melting at an accelerated rate due to processes that are not fully understood and hard to measure. The distribution of surface meltwater can help understand these processes and is observable through remote sensing, but current maps of meltwater face a trade-off: They are either high-resolution in time or space, but not both. We develop a deep learning model that creates gridded surface meltwater maps at daily 100m resolution by fusing data streams from remote sensing observations and physics-based models. In particular, we spatiotemporally downscale regional climate model (RCM) outputs using synthetic aperture radar (SAR), passive microwave (PMW), and a digital elevation model (DEM) over the Helheim Glacier in Eastern Greenland from 2017-2023. Using SAR-derived meltwater as "ground truth", we show that a deep learning-based method that fuses all data streams is over 10 percentage points more accurate over our study area than existing non deep learning-based approaches that only rely on a regional climate model (83% vs. 95% Acc.) or passive microwave observations (72% vs. 95% Acc.). Alternatively, creating a gridded product through a running window calculation with SAR data underestimates extreme melt events, but also achieves notable accuracy (90%) and does not rely on deep learning. We evaluate standard deep learning methods (UNet and DeepLabv3+), and publish our spatiotemporally aligned dataset as a benchmark, MeltwaterBench, for intercomparisons with more complex data-driven downscaling methods. The code and data are available at $href{https://github.com/blutjens/hrmelt}{github.com/blutjens/hrmelt}$.

Problem

• Develops a deep learning model for high-resolution daily meltwater mapping
• Downscales climate model outputs using remote sensing data fusion
• Creates a benchmark dataset to evaluate data-driven downscaling methods

Innovation

• Deep learning fuses remote sensing and climate model data
• Spatiotemporally downscales to daily 100m resolution meltwater maps
• Uses SAR-derived meltwater as ground truth for accuracy improvement