Publications

AI-ML Applications for CyanoHAB Detection and Multi-sensor Data Fusion

Satellite-based long-term monitoring of Cyanobacterial harmful algal blooms (CyanoHABs) is highly needed as these blooms threaten water quality, aquatic ecosystems, and public health. MERIS and OLCI sensors have been widely used to monitor CyanoHAB due to the availability of the 620 nm, 681nm, and 708nm bands required for chlorophyll fluorescence and phycocyanin confirmation. However, due to the sudden decommissioning of MERIS, we have a 4-year mission gap (2012 to 2015). To address that gap, we developed CyanNet, a deep learning model that reconstructs the Cyanobacteria Index (CI) from MODIS-Terra data, bridging the critical satellite data gap between the MERIS and Sentinel-3 OLCI missions and constructing a long-term satellite-based cyanoHAB timeseries from 2000 to the present. The model successfully reproduced bloom magnitude and spatial–temporal patterns and demonstrated strong performance across multiple lakes without regional retraining. By enabling continuous, decades-long satellite records, this work demonstrates how artificial intelligence can enhance environmental monitoring and support more informed water-quality management under changing environmental conditions. All relevant publications are available in the profile page under publications.


Related Publications


CyanoHAB Characterization and Long-term Trends in the U.S.


Quantifying Harmful Algal Bloom Severity from Satellite Observations

Cyanobacterial harmful algal blooms (cyanoHABs) threaten ecosystems, drinking water, and public health. Yet, most satellite approaches traditionally focus on detecting blooms and their concentrations at a given pixel rather than measuring the lake-wide severity of blooms over the bloom season. Our study introduced a new framework to quantify seasonal and annual bloom magnitude using MERIS satellite observations by combining spatial and temporal patterns of peak cyanobacteria biomass. Applied across lakes in the United States, the method enables consistent ranking of water bodies based on bloom intensity over time. Although the method was developed using Envisat's MERIS and Sentinel-3 OLCI data, the approach can be extended to any satellite observation that quantifies cyanobacteria biomass. Our approach provides a scalable tool to support long-term monitoring and inform water quality management and decision-making.


National Trends in Harmful Algal Bloom Severity Across U.S. Lakes

Using nine years of satellite observations, this study assessed changes in cyanobacterial bloom magnitude across 1,881 large U.S. lakes during 2016-2020 compared to 2008-2011. We observed that bloom severity declined in many lakes and increased in few, with most showing no significant change, highlighting strong regional climate and land–water interactions rather than uniform nationwide worsening trends.


Related Publications


Bio-optical Algorithms for mapping Chlorophyll and Phycocyanin in Optically Complex Waters

We developed a quasi-analytical algorithm to accurately estimate phytoplankton and detrital absorption coefficients from remote sensing reflectance measurements in highly turbid waters. Validation with filter-pad measurements from in situ water samples showed strong agreement, highlighting improved accuracy when accounting for detrital matter. Using the same bio-optical inversion method, we developed a novel remote sensing approach (QAA-PC) to monitor cyanobacterial blooms by estimating phycocyanin, a characteristic photo-pigment associated with cyanobacteria. Our method separates phytoplankton absorption coefficients from total phytoplankton absorption coefficients at 620 nm to quantify phycocyanin. The model performed well, particularly under high-biomass conditions common in nutrient-rich waters dominated by cyanobacteria. By improving the detection of chlorophyll and phycocyanin pigments in optically complex environments, this approach advances satellite-based monitoring of harmful algal blooms. Thus,  it supports a more accurate assessment of water quality in productive inland waters.


Related Publications


Regional and Global Datasets

Our hyperspectral remote sensing measurements, along with absorption coefficients of phytoplankton pigments, detrital matter, and colored dissolved organic matter (CDOM), and chlorophyll a and phycocyanin concentrations, have been widely used in peer-reviewed, high-impact publications. Now the dataset is available as part of the global dataset - GLORIA.


Related Publications



Other Relevant Works