Spatiotemporal downscaling and nowcasting of urban land surface temperatures with deep neural networks
arXiv:2605.13566v2 Announce Type: replace Abstract: Land Surface Temperature (LST) is a key variable for various applications, such as urban climate and ecology studies. Yet, existing satellite-derived LST products provide either high spatial or high temporal resolution, resulting in a fundamental trade-off between the two. To address this trade-off, we combine observations from a geostationary and a polar orbiting satellite and provide LST fields at high spatial and high temporal resolution (1 km at 15-min intervals). We demonstrate their application for intraday forecasting of LSTs. To estim
The increasing availability of advanced satellite data and deep learning techniques allows for more precise and granular environmental modeling, addressing long-standing data trade-offs.
Accurate, high-resolution spatiotemporal LST data enables better urban planning, climate resilience strategies, and impact assessments for infrastructure and public health.
The ability to generate high-resolution LST data (1km at 15-min intervals) shifts how urban climate and ecological studies can be conducted, moving from aggregated averages to dynamic, granular insights.
- · Urban planners
- · Climate researchers
- · Smart city developers
- · Environmental monitoring services
- · Traditional LST data providers (if they don't adapt)
- · Sectors reliant on less precise climate models
Improved understanding and predictive capabilities for urban heat island effects and localized climate phenomena.
More effective and targeted interventions for urban cooling, energy consumption management, and public health advisories during heatwaves.
Integration of dynamic LST data into AI-driven urban infrastructure management systems, leading to optimized resource allocation and enhanced city resilience.
This signal links to a primary source. Continuum Brief monitors and indexes it as part of the live intelligence stream — we do not republish source content.
Read at arXiv cs.LG