In a groundbreaking study published in IEEE Access, a journal that translates to ‘IEEE Open Access’, researchers led by Jangho Lee from the University of Illinois Chicago have developed a novel approach to estimating near-surface air temperature (T2M) using satellite-derived land surface temperature (LST) and deep learning techniques. This research could significantly impact the energy sector by enhancing predictive capabilities and optimizing resource management.
The study, which compares three deep learning methods—Long Short-Term Memory (LSTM), Temporal Convolutional Network (TCN), and Neural Basis Expansion Analysis Time Series (N-BEATS)—shows that incorporating temporal context through varying look-back windows can substantially improve the accuracy of T2M estimates. By reducing the root-mean-square error (RMSE) from around 2.6–2.8°C to below 1.8°C, these models demonstrate the value of historical LST observations in capturing the evolving surface-air temperature relationship.
“By leveraging deep learning, we’ve shown that historical LST data can significantly enhance the accuracy of near-surface air temperature estimates,” said Lee. “This has direct implications for sectors like energy, where precise temperature predictions are crucial for efficient resource management and grid stability.”
The research highlights that longer lags generally improve accuracy, although N-BEATS performance plateaus beyond a certain window. This reflects both diminishing returns and practical limitations linked to missing cloud-free satellite data. Seasonal and diurnal evaluations reveal higher errors in spring and midday hours, likely due to rapid vegetation changes and stronger physical and dynamical processes that make T2M less predictable.
Spatially, stations with denser vegetation exhibit elevated errors, suggesting that transpiration and canopy effects complicate the LST-T2M linkage. For extreme-event detection, LSTM provides the fewest false alarms (highest precision), N-BEATS captures the most extremes (highest recall), and TCN offers the best overall balance in precision and recall (highest F1).
The implications for the energy sector are profound. Accurate T2M estimates can help energy providers better predict demand, optimize power generation, and reduce the risk of blackouts during extreme weather events. “This framework has direct applications in heat-warning systems, resource management, precision agriculture, and urban climate adaptation,” Lee noted. “It stands to benefit further from ongoing advancements in satellite sensing technology.”
Future work could explore adaptive lag strategies, additional data sources, and more advanced data-fusion techniques. As satellite sensing technology continues to evolve, the integration of deep learning with satellite data could revolutionize how we monitor and predict near-surface temperatures, benefiting not only the energy sector but also agriculture, urban planning, and climate research.