Classification Metrics for Improved Atmospheric Correction of Multispectral VNIR Imagery

Area of Science:

• Remote Sensing
• Atmospheric Science
• Image Processing

Background:

• Multispectral Earth observation satellites (e.g., ALOS AVNIR-2) typically use 3-5 spectral bands.
• Standard atmospheric correction methods require shortwave infrared bands or hyperspectral data to determine aerosol optical thickness, which are often unavailable.

Purpose of the Study:

• To develop an improved atmospheric correction method for multispectral imagery.
• To enhance automatic processing systems for Earth observation data.

Main Methods:

• Development of novel classification metrics for identifying cloud, cloud over water, haze, water, and saturation.
• Application of these metrics within a prototype atmospheric processor for the ALOS AVNIR-2 payload.

Main Results:

• The new classification metrics improve the accuracy of atmospheric correction for multispectral images.
• The developed methods enhance the robustness of automatic processing systems.

Conclusions:

• The proposed classification metrics effectively address the limitations of standard atmospheric correction techniques for multispectral data.
• This advancement is crucial for reliable Earth observation data analysis, especially when specific spectral bands are absent.