Statistical analysis of hyperspectral data from two Swedish lakes.

Area of Science:

• Environmental Remote Sensing
• Aquatic Ecosystem Monitoring
• Geospatial Data Analysis

Background:

• Accurate water quality assessment is crucial for lake management.
• Remote sensing offers a scalable approach to monitor large water bodies.
• Statistical analysis of spectral data can reveal water quality parameters.

Purpose of the Study:

• To analyze statistical properties of Compact Airborne Spectrographic Imager (CASI) spectral data from Swedish lakes.
• To develop and validate algorithms for chlorophyll mapping and water quality assessment using remote sensing data.
• To evaluate the transferability of these algorithms between different lake environments.

Main Methods:

• Principal Component Analysis (PCA) for dimensionality reduction and feature extraction.
• Development of regression models for chlorophyll mapping based on spectral data.
• Application of the radiative transfer code 6S for atmospheric correction.
• Cross-validation of algorithms between two distinct Swedish lakes.

Main Results:

• PCA effectively reduced data dimensionality and enabled chlorophyll map generation.
• The quality of spectral reconstruction from principal components was high.
• The radiative transfer code 6S demonstrated accuracy in atmospheric correction.
• Algorithms developed for one lake showed robustness when applied to another, with comparable water quality mapping.

Conclusions:

• Statistical analysis of CASI data, particularly using PCA, is effective for chlorophyll mapping.
• Atmospheric correction using models like 6S is vital for accurate spectral data analysis.
• Developed algorithms exhibit robustness and transferability for mapping water quality parameters in different lake ecosystems.
• Proposed spectral band characteristics can enhance chlorophyll mapping accuracy.