Scholarlink

<p>This study leverages high-resolution remote sensing and data processing advanced techniques to improve ambient urban temperature monitoring. Using Landsat 8 and Sentinel-2A satellite imagery, the methodology combines spectral harmonization and correction techniques with Convolutional Neural Network (CNN)-based super-resolution models to achieve a high spatial accuracy in Land Surface Temperature (LST) and Air Temperature (Ta) estimations. The analysis integrates key environmental indices such as the Normalized difference vegetation, water, and built-area indices, and corrects for atmospheric and surface effects to refine LST data. Results demonstrate that CNN models improve temperature spatial detail significantly to a resolution of 1&#xa0;m with an R<sup>2</sup> above 0.85, and with estimations of aerial temperature optimized using Météo-France and validated against Météociel data, showing errors within 2°C. Regression models further estimate Ta from LST with R<sup>2</sup> values above 0.75, effectively mapping temperature distributions at fine resolutions for urban settings. This study bridges critical gaps in remote-sensing based temperature monitoring, hence offering a framework for high-resolution urban thermal analysis in regions with limited meteorological data.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Enhancing urban temperature monitoring through high-resolution remote sensing and advanced data processing techniques

  • 期刊论文

  • 发表日期：

    2025年7月2日

查看全文
  • Naji El Beyrouthy,
  • Mario Al Sayah,
  • Rita Der Sarkissian,
  • Rachid Nedjai

摘要

This study leverages high-resolution remote sensing and data processing advanced techniques to improve ambient urban temperature monitoring. Using Landsat 8 and Sentinel-2A satellite imagery, the methodology combines spectral harmonization and correction techniques with Convolutional Neural Network (CNN)-based super-resolution models to achieve a high spatial accuracy in Land Surface Temperature (LST) and Air Temperature (Ta) estimations. The analysis integrates key environmental indices such as the Normalized difference vegetation, water, and built-area indices, and corrects for atmospheric and surface effects to refine LST data. Results demonstrate that CNN models improve temperature spatial detail significantly to a resolution of 1 m with an R2 above 0.85, and with estimations of aerial temperature optimized using Météo-France and validated against Météociel data, showing errors within 2°C. Regression models further estimate Ta from LST with R2 values above 0.75, effectively mapping temperature distributions at fine resolutions for urban settings. This study bridges critical gaps in remote-sensing based temperature monitoring, hence offering a framework for high-resolution urban thermal analysis in regions with limited meteorological data.