Tracking River Health: From Lab to Satellite with Chlorophyll-a Monitoring

Greg Howard
12th December, 2025

Tracking River Health: From Lab to Satellite with Chlorophyll-a Monitoring

Satellite-derived chlorophyll-a fluorescence in the lower Hudson River shows substantially greater variability across time between survey dates than spatially along the river on any single day.

Image adapted from: Salls et al. / CC BY (Source)

Key Findings

  • This study focused on a 220-kilometer stretch of the lower Hudson River between 2021 and 2023 to improve satellite monitoring of algal blooms
  • Laboratory measurements of chlorophyll-a (chl-a) and field-measured chl-a fluorescence showed a weak initial correlation, but this improved when data were analyzed separately for each day
  • Machine learning models trained on data from individual days provided the most accurate estimates of chl-a fluorescence, highlighting the importance of accounting for daily changes in river conditions
Harmful algal blooms (HABs) pose a significant threat to both human and ecological health in freshwater systems worldwide[2]. These blooms can contaminate drinking water, disrupt aquatic ecosystems, and cause economic losses. The Hudson River, a vital resource for the northeastern United States, is particularly susceptible to these blooms, necessitating effective monitoring strategies. Traditionally, monitoring relies on collecting water samples and analyzing them in laboratories, a process that is time-consuming, expensive, and provides only a snapshot of conditions at specific locations and times. Satellite remote sensing offers a promising alternative, enabling broader and more frequent observation of water quality. However, rivers present unique challenges for satellite-based monitoring due to their dynamic flow, complex geometry, and varying water properties. Researchers at the U.S. Geological Survey[1] recently investigated methods to improve the accuracy of satellite-based chlorophyll-a (chl-a) monitoring in the Hudson River. Chl-a is a pigment found in all algae, and its concentration is often used as a proxy for algal biomass – a higher chl-a level generally indicates a larger algal population. The study focused on a 220-kilometer stretch of the lower Hudson River between 2021 and 2023. The team combined field measurements of chl-a collected during boat surveys with data from Sentinel-2 satellites, a series of European satellites designed for environmental monitoring. A key difficulty identified was the weak relationship between laboratory chl-a measurements and in-situ (on-site) chl-a fluorescence readings (r² = 0.25). Chl-a fluorescence is a measure of the light emitted by algae, which can be measured in real-time using specialized sensors. The initial poor correlation highlighted the inherent variability within the river environment. However, the researchers found that separating the data based on the time of day significantly improved the correlation (mean r² = 0.53), suggesting that daily patterns influence chl-a concentrations. To translate satellite observations into accurate estimates of chl-a fluorescence, the team developed machine learning models called random forests. These models were trained using the extensive dataset of in-situ fluorescence measurements collected during the boat surveys. They tested three different modeling approaches: models trained on data from individual days, models trained on all days except one (used as a “holdout” day to test performance on unseen data), and a single model trained on all available data. The results showed that models trained on individual days performed best, with the lowest error rates (mean absolute error of 0.16 relative fluorescence units). This was followed by the single pooled model (mean absolute error of 0.22 RFU). The holdout models, which were designed to simulate real-world prediction scenarios, exhibited the highest error rates (mean absolute error of 0.40 RFU). This difference in performance underscores the importance of accounting for temporal variability when modeling riverine systems. This study builds upon previous work demonstrating the potential of satellite remote sensing for monitoring cyanobacterial blooms in lakes[3]. While satellite technology can effectively detect bloom frequency and spatial extent, accurately quantifying toxin concentrations still relies on traditional grab samples[3]. The research from the Hudson River highlights that simply applying methods developed for lakes to rivers is not sufficient. The dynamic nature of rivers requires more sophisticated modeling approaches that can capture daily and other short-term variations in chl-a concentrations. Furthermore, the study echoes findings from research using advanced sensor platforms[4], which also emphasized the importance of high-density spatial data to understand variability in aquatic environments. The findings from emphasize that while remote sensing provides a valuable tool for monitoring water quality in rivers, the accuracy of these measurements can be significantly improved by incorporating temporal variability into the models. This is achieved by developing models that are specific to the time of day, or by using more complex models that can account for daily fluctuations in chl-a concentrations. The need to account for spatial heterogeneity when forecasting cyanobacterial blooms, as suggested by the intensive sampling events in Lake Erie[3], is also reinforced by the findings of.

EnvironmentEcologyOceanography

References

Main Study

1) From sample to sonde to Sentinel-2: insights from a multi-scale chlorophyll-a monitoring effort in the Hudson River, New York

Published 9th December, 2025

https://doi.org/10.1007/s10661-025-14844-3

Related Studies

2) Satellite monitoring of cyanobacterial harmful algal bloom frequency in recreational waters and drinking source waters.

https://doi.org/10.1016/j.ecolind.2017.04.046

3) The Lake Erie HABs Grab: A binational collaboration to characterize the western basin cyanobacterial harmful algal blooms at an unprecedented high-resolution spatial scale.

https://doi.org/10.1016/j.hal.2021.102080

4) High-speed limnology: using advanced sensors to investigate spatial variability in biogeochemistry and hydrology.

https://doi.org/10.1021/es504773x


Related Articles

An unhandled error has occurred. Reload 🗙