Resolution Revolution: A Physics-Guided Deep Learning Framework for Spatiotemporal Temperature Reconstruction

2025-07-13

AI Summary

Earth observation faces a fundamental spatiotemporal resolution trade-off in temperature monitoring: satellite remote sensing (e.g., Landsat) offers high spatial resolution (~100 m) but suffers from long revisit intervals (16 days) and cloud contamination, whereas Earth system models provide high temporal resolution (hourly) yet coarse spatial granularity (9–31 km). To bridge this gap, we propose a physics-guided deep learning framework that integrates multi-source heterogeneous data—GOES-16 (hourly, 2 km), cloud-sensitive Landsat (~100 m), and model outputs—via a novel annual thermal cycle prior and a linear upscaling mechanism. Built upon a ConvLSTM architecture, our method enables end-to-end spatiotemporal super-resolution. It achieves, for the first time, hourly, 100-meter-resolution, all-weather global land surface temperature retrieval. Evaluated on multiple benchmark datasets, it reduces RMSE by 23–37% over state-of-the-art methods, establishing a new paradigm for high-resolution land surface process modeling and disaster monitoring.

Technology Category

• Computer Vision: Remote Sensing / Geospatial AI
• Planning, Routing, and Scheduling: Optimization of Spatio-temporal Systems
• Knowledge Representation and Reasoning: Geometric, Spatial, and Temporal Reasoning

Application Category

• Search and Retrieval-Augmented AI: Large language models for search
• Graph Algorithms and Modeling for the Web: Algorithms and analysis for heterogeneous, signed, attributed, multi-relational, temporal, higher-order, and annotated Web-related graphs
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications

Abstract

Central to Earth observation is the trade-off between spatial and temporal resolution. For temperature, this is especially critical because real-world applications require high spatiotemporal resolution data. Current technology allows for hourly temperature observations at 2 km, but only every 16 days at 100 m, a gap further exacerbated by cloud cover. Earth system models offer continuous hourly temperature data, but at a much coarser spatial resolution (9-31 km). Here, we present a physics-guided deep learning framework for temperature data reconstruction that integrates these two data sources. The proposed framework uses a convolutional neural network that incorporates the annual temperature cycle and includes a linear term to amplify the coarse Earth system model output into fine-scale temperature values observed from satellites. We evaluated this framework using data from two satellites, GOES-16 (2 km, hourly) and Landsat (100 m, every 16 days), and demonstrated effective temperature reconstruction with hold-out and in situ data across four datasets. This physics-guided deep learning framework opens new possibilities for generating high-resolution temperature data across spatial and temporal scales, under all weather conditions and globally.

Problem

• Bridging spatiotemporal resolution gap in temperature data
• Integrating coarse and fine-scale temperature data sources
• Reconstructing high-resolution temperature under all weather conditions

Innovation

• Physics-guided deep learning for temperature reconstruction
• Integrates satellite and Earth system model data
• Convolutional neural network with annual cycle term