Building spectral libraries for wetlands land cover classification and hyperspectral remote sensing

Area of Science:

• Ecology
• Remote Sensing
• Geospatial Analysis

Background:

• Remote sensing technology, particularly hyperspectral imagers, offers advanced capabilities for mapping wetland vegetation at relevant scales.
• Existing space-based sensors have limitations in spectral resolution compared to airborne hyperspectral imagers.
• Developing spectral libraries is essential for advanced analytical techniques to monitor wetland ecosystems.

Purpose of the Study:

• To create a common spectral library for mapping wetland vegetation using remotely sensed data.
• To assess the potential of airborne hyperspectral sensors for mapping wetland composition.
• To highlight the importance of diverse, comprehensive wetland spectral libraries for global applications.

Main Methods:

• Collected canopy and leaf spectra (n=1336) from five coastal wetland sites in California, Texas, and Mississippi.
• Developed a spectral library encompassing various bioclimatic, edaphic, and disturbance conditions.
• Utilized the spectral library to classify and delineate vegetation in the Pacheco Creek wetland using PROBE-1 airborne hyperspectral data (5m resolution, 128 bands).

Main Results:

• Demonstrated the successful classification and delineation of wetland vegetation using the developed spectral library and hyperspectral data.
• Illustrated the potential of advanced hyperspectral sensors for detailed mapping of wetland composition.
• Highlighted that comprehensive spectral libraries provide a robust physical basis for data interpretation.

Conclusions:

• Comprehensive wetland spectral libraries are vital for accurate vegetation mapping and monitoring.
• Airborne hyperspectral sensors, combined with robust spectral libraries, show significant potential for wetland research.
• Standardized criteria and compilation of spectral libraries are recommended for future global applications and space-based hyperspectral missions.