[A Comparative Study on the Calibration Accuracy of Landsat 8 Thermal Infrared Sensor Data]

Area of Science:

• Remote Sensing
• Earth Observation
• Satellite Radiometry

Background:

• Satellite thermal infrared imagery is crucial for monitoring Earth's surface temperature.
• Landsat 8's Thermal Infrared Sensor (TIRS) provides valuable data, but its calibration parameters have undergone revisions.
• Reprocessing of Landsat 8 data by USGS necessitates validation of its thermal calibration accuracy.

Purpose of the Study:

• To assess the calibration accuracy of reprocessed Landsat 8 TIRS data.
• To cross-compare Landsat 8 TIRS data with well-calibrated Landsat 7 ETM+ thermal infrared data.
• To investigate the differences in at-sensor temperature measurements between the two satellite sensors across varying vegetation covers.

Main Methods:

• Acquired three date-coincident image pairs of Landsat 8 TIRS and Landsat 7 ETM+ data over the western United States.
• Selected three test sites representing moderate vegetation, low vegetation, and bare soil cover.
• Converted thermal infrared data to at-sensor temperature and performed band-by-band comparison and regression analysis.

Main Results:

• High agreement (correlation coefficients > 0.95) was observed between Landsat 8 TIRS and Landsat 7 ETM+ thermal data.
• Landsat 8 TIRS Band 10 overestimated temperature by up to 1.37 K, while TIRS Band 11 underestimated by up to -3 K compared to Landsat 7 ETM+ Band 6.
• Temperature differences varied with vegetation cover: TIRS Band 10's overestimation increased with decreasing vegetation, while TIRS Band 11's underestimation decreased.

Conclusions:

• Despite reprocessing, Landsat 8 TIRS calibration remains unstable, particularly for Band 11.
• TIRS Band 11 shows better performance in bare soil areas, while TIRS Band 10 has overall better performance.
• Averaging TIRS Bands 10 and 11 reduces sensor temperature differences in low vegetation areas.