Optical classification of an urbanized estuary using hyperspectral remote sensing reflectance

Area of Science:

• Environmental Optics
• Remote Sensing
• Aquatic Biogeochemistry

Background:

• Optical water classification using remote sensing reflectance (Rrs(λ)) is crucial for understanding aquatic systems.
• Complex estuarine environments like Long Island Sound (LIS) require robust bio-optical algorithms.
• In situ data are essential for developing and validating these algorithms.

Purpose of the Study:

• To optically classify waters in the Long Island Sound (LIS) using hyperspectral Rrs(λ) data.
• To investigate the relationship between spectral shape, biogeochemistry, and water properties.
• To evaluate continuous spectral indices and atmospheric correction methods for satellite data.

Main Methods:

• Collected in situ hyperspectral Rrs(λ) and biogeochemical data over five years in LIS.
• Applied k-means clustering to Rrs(λ) data for discrete optical water types.
• Utilized the Apparent Visible Wavelength (AVW) for continuous spectral indexing.
• Assessed Sentinel-3 OLCI satellite data using the Quality Water Index Polynomial (QWIP).

Main Results:

• LIS waters were best described by three Rrs(λ) spectral clusters, two dominated by phytoplankton.
• The third cluster represented river and river plume waters.
• AVW effectively captured water quality gradients and spectral changes.
• The Polymer atmospheric correction approach was preferred for Sentinel-3 OLCI data in LIS.

Conclusions:

• Discrete and continuous optical indices effectively characterize LIS water properties and biogeochemistry.
• Continuous indices like AVW are valuable for assessing nearshore variability.
• Integrative indices aid in evaluating in situ and satellite bio-optical data quality for water management.