Thermal remote sensing over heterogeneous urban and suburban landscapes using sensor-driven super-resolution

Area of Science:

• Earth and Space Sciences
• Remote Sensing
• Environmental Monitoring

Background:

• Thermal remote sensing is crucial for climate and urban heat island monitoring but often has low spatial resolution.
• Existing data-driven super-resolution methods enhance visual appeal but may compromise radiometric consistency.
• Sensor-specific characteristics are often overlooked in traditional empirical super-resolution techniques.

Purpose of the Study:

• To evaluate a sensor-driven super-resolution approach for thermal imagery.
• To ensure radiometric consistency during high-resolution thermal image retrieval.
• To improve the spatial resolution of thermal remote sensing data.

Main Methods:

• Developed and applied a sensor-driven super-resolution algorithm explicitly modeling the sensor blurring process.
• Utilized Moderate Resolution Imaging Spectroradiometer (MODIS) data for a heterogeneous urban and suburban landscape.
• Validated the downscaled image against Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) data.

Main Results:

• The sensor-driven algorithm successfully downscaled MODIS thermal imagery to 250-m resolution.
• Maintained high statistical consistency with both original MODIS and reference ASTER thermal images.
• Demonstrated effectiveness in diverse landscapes including urban, forest, and water bodies.

Conclusions:

• The sensor-driven approach offers a radiometrically consistent method for enhancing thermal image resolution.
• This technique is valuable for accurate environmental monitoring, especially for urban heat islands.
• Components like radiometric offset correction show potential for integration with existing methods.