Semi-analytical inversion modelling of Chlorophyll a variability in the U.S. Virgin Islands

Frontiers in Remote Sensing

Remote Sensing

M. E. Brandt¹  T. B. Smith¹  K. Adem Ali²  D. C. Flanagan²  J. D. Ortiz³ 

[1] Center for Marine and Environmental Studies, University of Virgin Islands, Charlotte Amalie, VI, United States;Department of Geology and Environmental Geosciences, College of Charleston, Charleston, SC, United States;Department of Geology, Kent State University, Kent, OH, United States;

关键词: chlorophyll-a;    remote sensing;    water quality;    US Virgin Islands;    coastal;   

DOI  :  10.3389/frsen.2023.1172819

 received in 2023-02-23, accepted in 2023-04-19,  发布年份 2023

来源: Frontiers

【 摘 要 】

Coral reef health in the U.S. Virgin Islands (USVI) is in decline due to land-based sources of pollution associated with watershed development and global climate change. Water quality is a good indicator of stress in these nearshore environments as it plays a key role in determining the health and distribution of coral reef communities. Conventional water quality assessment methods based on in situ measurements are both time consuming and costly, and they lack the spatial coverage and temporal resolution that can be achieved using satellite remote sensing techniques. Water quality parameters (WQPs) such as Chlorophyll a (Chl-a), can be studied remotely using models that account for the inherent optical properties (IOPs) of the water. In this study, empirical based standard ocean color algorithm (OC4) and two semi-analytical algorithms, the Garver–Siegel–Maritorena (GSM) and the Generalized Inherent Optical Properties (GIOP) model, were evaluated in retrieving Chl-a in the nearshore waters of the USVI. GSM and GIOP were also evaluated for modeling inherent optical properties such as absorption coefficient of phytoplankton (aph (443)). Analysis of the results from each model using a field database from six cruises during May/June and December between 2016 and 2018, showed that the OC4 performed poorly with R2 of 0.14 and RMSE = 0.15. Effects of suspended particulates and benthic reflectance most likely contributed to the poor performance of the algorithm. GSM is a slightly better estimator for aph (443) and Chl-a (R2 = 0.55, RMSE = 0.04; R2 = 0.60, RMSE = 0.09) than GIOP (R2 = 0.52, RMSE = 0.05; R2 = 0.17, RMSE = 0.15). Performance of the semi-analytical models are limited in estimating particulate back scattering (bbp (443)) also due to the benthic albedo effects in the shallow waters. The calibrated GSM model was applied to Landsat 8 OLI satellite imagery spanning 2016–2018 to develop a time series of the spatial changes in Chl-a concentrations in the coastal waters of the USVI. The Landsat GSM Chl-a model produced promising results of R2 = 0.45, RMSE = 0.07, in an environment where signal-to-noise ratio is significantly low.

【 授权许可】

Unknown   
Copyright © 2023 Ali, Flanagan, Brandt, Ortiz and Smith.