There is a growing number of applications leveraging remote sensing technologies to monitor and manage water quality in inland waters. The use of remote sensing technologies, especially from satellites, is extremely valuable for limnological applications due to their ability to provide routine, large-scale analyses of spatio-temporal dynamics of key ecological variables and their use for time-series analysis through retrospective imagery. Because of this, remote sensing has played an important role in advancing water quality mapping by allowing the retrieval of optical and physical water properties from local to global scales. Several key factors have contributed to this increase: i) development of space science and technology which has increased the availability of Earth Observation data in the past decades; ii) development of new sensors with improved spectral resolution; iii) establishment of global networks in the field of water quality remote sensing (e.g., GEO AquaWatch, World Water Quality Alliance Earth Observations Workstream, Global Lake Ecological Observatory Network Aquatic Remote Sensing Working Group, and others), as well as open-source global datasets (e.g., GLORIA); and iv) improvements in computational power and the development of cloud computing platforms. Concurrent with these factors has been an increasing acceptance of remotely sensed water quality data worldwide; however, freshwater ecosystems are optically complex, have high biogeochemical variability, and high biological diversity, making common remote sensing products traditionally made for terrestrial systems unsuitable for freshwater. In this session, we aim to highlight efforts focused on applying satellite remote sensing data, developing methods and tools for the estimation of water quality parameters, and supporting capacity building strategies to bridge knowledge gaps in remote sensing studies of inland water quality. This session will contribute to the continuous growth and integration of remote sensing technology in limnology and contribute to the standardization of satellite-based water quality products to further improve the use of remote sensing technology for reliable inland water quality monitoring.
Lead Organizer: Igor Ogashawara, Leibniz Institute of Freshwater Ecology and Inland Fisheries (IGB) (igoroga@gmail.com)
Co-organizers:
Megan Coffer, National Oceanic and Atmospheric Administration (NOAA) (megan.coffer@noaa.gov)
Harriet Wilson, University of Stirling (harriet.wilson@stir.ac.uk)
Daniel Andrade, National Institute for Space Research (INPE) (damaciel_maciel@hotmail.com)
Presentations
09:00 AM
Remote Sensing of Water Quality Parameters over Western Mississippi Sound by Using Sentinel-3 OLCI and Machine Learning (9716)
Primary Presenter: hafez ahmad, Mississippi State University (hafezahmad100@gmail.com)
Remote sensing has emerged as a crucial tool for monitoring water quality, offering a cost-effective and spatially comprehensive approach for monitoring water quality and assessing aquatic ecosystem health. The remote sensing sensor, Ocean and Land Color Instrument (OLCI), onboard the Sentinel 3A satellite provides significant opportunities for monitoring coastal waters. With a spatial resolution of 300 m across 21 spectral bands, it enables extraction of detailed information about water quality parameters. In this study, we investigated the feasibility of using OLCI products to monitor an optically complex coastal water body in the western Mississippi Sound (WMS), building on the potential of remote sensing to address water quality issues. We employed machine learning (ML) and deep learning models for developing algorithms for chlorophyll-a (Chl-a), colored dissolved organic matter (CDOM), turbidity, and surface dissolved oxygen using OLCI imagery and in situ measurements by an autonomous surface vessel in 2021, 2022, and 2023 from WMS. Our approach involved applying automatic model selection algorithms to determine the optimal combination and number of spectral bands for training models. Notably, for Chl-a estimation, the random forest (RF) model yielded the highest adjusted (Adj) R² of 0.96 with a root mean squared error (RMSE) of 0.31 μg/L. For CDOM, the RF, extreme gradient boosting (XGBoost), and decision tree models produced high Adj R² values of 0.99, 0.98, and 0.98, with RMSEs of 1.18, 1.16, and 1.18 μg/L, respectively. Similarly, for turbidity, RF, XGBoost, and K-Nearest Neighbors models emerged as top performers, demonstrating high accuracy. For dissolved oxygen estimation, RF and XGBoost models exhibited robust performance across various metrics, both achieving an Adj R² of 0.89, indicating an excellent fit between the estimated and actual values. This study demonstrates the efficacy of utilizing OLCI products coupled with ML techniques for robust monitoring of water quality parameters in optically complex coastal environments, significantly contributing to enhanced environmental monitoring, management, and conservation efforts.
09:15 AM
Mixture density networks are a better predictor of nutrients than chlorophyll in a well studied, optically complex estuary (9723)
Primary Presenter: Greg Silsbe, University of Maryland Center for Enviornmental Science (gsilsbe@umces.edu)
It has long been recognized that empirical algorithms that predict water quality from satellite remote sensing have poor accuracy in optically complex coastal, estuarine and inland waters. This is unfortunate as it these environments that are most susceptible to anthropogenic stressors and therefore stand to benefit the most from the promise of highly resolved monitoring from space. To fill this critical void, machine learning approaches have emerged as a superior method to predict water quality measurements in optically complex environments. In this study we begin by assembling a large dataset of multispectral Sentinel 3 data collocated in space and time with the data rich Chesapeake Bay water quality monitoring program. Following two common methods, we compare and contrast the segregation of data across Optical Water Types (OWTs via K-means clustering), and assess the efficacy of Mixture Density Networks (MDNs) to predict in-situ water quality data. Critically, we extend our analysis beyond traditional optically active constituents (e.g. chlorophyll, Secchi depth, total suspended solids) and include bulk total nitrogen and phosphorus concentrations. OWTs vary predictably along the estuarine salinity gradient and generally yield statistically different distributions of water quality parameters across spectrally dissimilar OWTs. Perhaps surprisingly, MDN predictions of total nitrogen and phosphorus not only have small errors (32% and 38% respectively), but they also exceed the measurement error of chlorophyll (72%).
09:30 AM
Unleashing the power of remote sensing data in aquatic research: Guidelines for optimal utilization (9313)
Primary Presenter: Igor Ogashawara, Leibniz Institute of Freshwater Ecology and Inland Fisheries (igoroga@gmail.com)
A growing number of lake studies use remote sensing technology due to its holistic perspective, its regional to global coverage, its potential to recover time series of data and its use for improving predictive models. Correctly used, remote sensing technologies allow to follow temporal developments of large numbers of water bodies to detect long-term trends and to identify immediately changes within the aquatic environment. The increasing availability of remote sensing products and their large potential for limnological studies have added another level of spatial information to traditional studies. However, far too little attention has been paid to the remote sensing methods and products used in many of these applied limnological studies. In this study, we argue that a profound understanding of remote sensing methods for inland waters is necessary for the correct application and interpretation of these type of data. Due to the optical complexity and biogeochemical variability of freshwater ecosystems, common remote sensing products made for land or oceans applications are not applicable in inland waters. For example, the necessity of a proper atmospheric correction tailored for inland waters which considers adjacency effects as well as the lack of understanding of the quality and assumptions of remote sensing products are recurring critical issues found also in recent limnological studies. Fortunately, given the growing interest in remote sensing studies of inland waters, capacity building strategies have been discussed to disseminate the knowledge and fill these gaps.
09:45 AM
Satellite-Based Retrieval of Water Quality Parameters in Inland Waters Using Mixture Density Network Models (8977)
Primary Presenter: Lorena Pinheiro-Silva, University of Missouri-Columbia (lsilva@umces.edu)
Retrieving water quality (WQ) parameters from remotely sensed imagery in inland and coastal waters is challenging as, unlike the open ocean, dissolved and mineral substances often do not covary with phytoplankton. In this study, we tested the accuracy of Mixture Density Networks (MDN) algorithms in estimating WQ parameters⸺Chlorophyll-a (Chla), Secchi, total suspended solids, total nitrogen (TN), and total phosphorus⸺from Landsat imagery in Missouri reservoirs, USA. Landsat 5-8 data were acquired from the LimnoSat database and paired with in situ measurements taken within 1 day of sampling from up to 150 water bodies (matchups, N ranged from 929 to 966). Each MDN was trained using 50% of the total samples, while the remaining data was used to evaluate the model's performance. Across the suite of WQ parameters, the best performing MDN model (41% error, low bias) was, perhaps unexpectedly, TN. In contrast, the MDN error for estimating Chla, by far the most studied WQ parameter from space, exceeded 100%. Model accuracy was modestly improved when MDNs were trained to estimate multiple parameters at the same time, rather than MDNs trained to estimate a single parameter. That satellite-based TN estimates have much lower error than Chla estimates is noteworthy not only from an applied nutrient management perspective, but also because only a few studies to date have attempted to estimate nutrient concentrations from space. The primary processing scheme used by LimnoSat is LaSRC; however, alternative atmospheric correction processors (e.g., ACOLITE) may improve MDN estimates of WQ.
10:00 AM
CYANOBACTERIA BLOOM CLASSIFICATION IN INLAND WATERS USING HYPERSPECTRAL DATA (9670)
Primary Presenter: Samantha Sharp, UC Davis (samanthalsharp@gmail.com)
Cyanobacteria Harmful Algal Blooms (CHABs) and their associated toxicity are a significant issue in inland waters. Remote sensing is a useful tool for monitoring CHABs in large lakes due to the extensive spatial coverage and frequent availability of satellite images. Advances in hyperspectral remote sensing have opened up a new frontier in CHAB monitoring. This study evaluates the feasibility of detecting cyanobacteria genera using hyperspectral data in hypereutrophic Clear Lake, CA, USA. In situ data were collected during 12 field events in 2021-2022 across all seasons to characterize the in situ cyanobacteria community composition, including coincident measurements of phytoplankton speciation and enumeration and spectral reflectance. Additionally, three field events were conducted concurrent with whole-lake hyperspectral image acquisitions by the DESIS sensor on the International Space Station. We apply the Spectral Mixture Analysis for Surveillance of HABs (SMASH) framework for Multiple Endmember Spectral Mixture Analysis (MESMA) to our DESIS scenes to demonstrate the potential to identify cyanobacteria genera from hyperspectral images. Both the original SMASH cyanobacteria spectral library and a Clear Lake-specific spectral library built from our field spectra measurements are employed. The results of this study will support the continued development of tools utilizing satellite-based hyperspectral images to identify the cyanobacteria genera present in a bloom, and thus assess the potential for cyanotoxin production.
10:15 AM
SS29 - The Pulse of Water Quality Remote Sensing in Inland Waters: State of the Art and Perspectives
Description
Time: 9:00 AM
Date: 30/3/2025
Room: W206A