Optical classification of an urbanized estuary using hyperspectral remote sensing reflectance

Abstract

Optical water classification based on remote sensing reflectance (Rrs(λ)) data can provide insight into water components driving optical variability and inform the development and application of bio-optical algorithms in complex aquatic systems. In this study, we use an in situ dataset consisting of hyperspectral Rrs(λ) and other biogeochemical and optical parameters collected over nearly five years across a heavily urbanized estuary, the Long Island Sound (LIS), east of New York City, USA, to optically classify LIS waters based on Rrs(λ) spectral shape. We investigate the similarities and differences of discrete groupings (k-means clustering) and continuous spectral indexing using the Apparent Visible Wavelength (AVW) in relation to system biogeochemistry and water properties. Our Rrs(λ) dataset in LIS was best described by three spectral clusters, the first two accounting for the majority (89%) of Rrs(λ) observations and primarily driven by phytoplankton dynamics, with the third confined to measurements in river and river plume waters. We found AVW effective at tracking subtle changes in Rrs(λ) spectral shape and fine-scale water quality features along river-to-ocean gradients. The recently developed Quality Water Index Polynomial (QWIP) was applied to evaluate three different atmospheric correction approaches for satellite-derived Rrs(λ) from the Sentinel-3 Ocean and Land Colour Instrument (OLCI) sensor in LIS, finding Polymer to be the preferred approach. Our results suggest that integrative, continuous indices such as AVW can be effective indicators to assess nearshore biogeochemical variability and evaluate the quality of both in situ and satellite bio-optical datasets, as needed for improved ecosystem and water resource management in LIS and similar regions.

Impacts of Hydrology and Extreme Events on Dissolved Organic Carbon Dynamics in a Heavily Urbanized Estuary and Its Major Tributaries: A View From Space

“…Hyperspectral remote sensing reflectance, R rs , was measured using an above‐water handheld Spectra Vista Corporation HR512i (SVC) (350−1050 nm range, resolution ≤1.5 nm) (Turner et al., 2022). Measurements were collected under ideal conditions (clear or fully overcast skies) with a consistent viewing geometry following the standard IOCCG protocols (Zibordi et al., 2019).…”

“…Less turbid, saline water from the Mid‐Atlantic Bight enters the Sound through the estuary mouth at its eastern boundary, driven primarily by semi‐diurnal tidal flushing and subtidal exchange (Gay et al., 2004; Whitney et al., 2016). As a result, optical and biogeochemical conditions, including sediment concentrations, primary production, nutrient fluxes, and organic matter quality, are highly variable (Aurin et al., 2010; Byrd et al., 2020; Goebel et al., 2006; Sherman, Tzortziou, Turner, Goes, & Grunert, 2023; Turner et al., 2022; Varekamp et al., 2014). LIS water quality conditions can be further complicated by non‐fluvial urban coastal processes, including export of wastewater from point sources, benthic resuspension of sediments and nutrients, wetland tidal exchanges, industrial discharge, and polluted stormwater runoff from urban landscapes (Georgas et al., 2014; Signell et al., 2000).…”

“…Care was taken to avoid shading or reflection by boat structures or perturbations caused by bubbles, floating debris, or foam on the water surface. The mean R rs was calculated from successive scans of the water surface, sky, and a calibrated Lambertian reference plaque using the three‐component (3C) reflectance model (Groetsch et al., 2017) to correct for residual skylight and sun glint effects (Pitarch et al., 2020; Turner et al., 2022). The full workflow for the SVC data processing is available online (https://github.com/tzortziou-lab).…”

“…Surface water samples were also collected for concurrent biogeochemical measurements (Figure 1c). Water samples were collected using Niskin bottles attached to a CTD‐rosette into clean 4 L Nalgene bottles pre‐rinsed with sampled water before collection (Turner et al., 2022). Samples were filtered on the day of collection using 0.7 μm Whatman GF/F glass fiber filters followed by pre‐washed 0.22 μm polycarbonate membrane filters.…”

Nano scattering particle: an approach to improve quality of the commercial led

“…The Mie-scattering function [23][24][25][26] of spherical particles is applied to investigate the scattering factor of the TiO2 in the phosphor layer. The calculated scattering factors include 𝜇 𝑠𝑐 (𝜆)scattering coefficients (SC) and 𝛿 𝑠𝑐reduced scattering coefficients (RSC).…”