Performance of three reflectance calibration methods for airborne hyperspectral spectrometer data

Area of Science:

• Remote Sensing
• Spectroscopy
• Geospatial Analysis

Background:

• Accurate conversion of airborne hyperspectral data to reflectance factors is crucial for quantitative analysis.
• Existing calibration methods like reflectance mode (RM) and linear-interpolation (LI) exhibit limitations in accuracy and consistency.
• Variations in time of day and flight duration significantly impact the performance of these calibration methods.

Purpose of the Study:

• To characterize and compare the performance and accuracy of three airborne hyperspectral data calibration methods.
• To identify the most reliable method for converting raw spectrometer data to reflectance factors.
• To provide quantitative estimates of uncertainty for the most effective calibration method.

Main Methods:

• Evaluated the reflectance mode (RM) method, involving pre-flight calibration against a white reference panel.
• Assessed the linear-interpolation (LI) method, using pre- and post-flight reference panel readings.
• Investigated the continuous panel (CP) method, employing a multi-band radiometer for continuous reference measurements during flight.

Main Results:

• The RM method produced biased and distorted reflectance spectra, highly dependent on time of day and flight duration.
• The LI method yielded precise but inaccurate reflectance spectra with biases varying by flight duration.
• The CP method demonstrated superior accuracy and reliability, providing consistent reflectance factor retrievals across different conditions with an estimated uncertainty of 0.0025 ± 0.0005 reflectance units.

Conclusions:

• The continuous panel (CP) method is the most accurate and reliable approach for airborne hyperspectral reflectance calibration.
• The RM method is suitable only for very short flights (<15 min) near solar noon.
• The LI method requires flight durations under 30 minutes for reasonable accuracy, while CP is effective for longer campaigns (1-2 hours).