China's Environmental Disaster Reduction Satellite 2 (HJ-2) recently launched charge-coupled device (CCD) sensors, designed to monitor environmental and ecological changes. This study marks the first application of HJ-2A/B CCD imagery for quantifying chlorophyll-a (Chl-a) in lakes with diverse optical properties and trophic statuses, highlighting its potential for comprehensive water quality monitoring. Chl-a, a key indicator of algae biomass and nitrogen levels, was evaluated using empirical algorithms (EMs), semi-analytical algorithms (e.g. quasi-analytical algorithms (QAAs) and data-driven machine learning (ML) algorithms. Results showed that EMs struggled with lake-specific characteristics, while QAA demonstrated reliability (R² > 0.75, RPD > 2). ML algorithms, leveraging their data-driven adaptability, outperformed both the EM and the QAA, with CatBoost (CB) achieving the highest accuracy (R² = 0.97, RMSE = 6.69 μg/L, MAE = 4.78 μg/L, RPD = 5.16). CB-generated spatial Chl-a distribution maps highlighted HJ-2A/B CCD's significant potential for practical water quality monitoring across lakes with varying optical and trophic conditions. This study not only validates HJ-2A/B CCD's utility in Chl-a quantification but also underscores the superiority of ML approaches in handling complex, data-driven challenges. The findings provide a robust foundation for future large-scale and long-term applications of HJ-2A/B CCD imagery, offering valuable insights for environmental managers and researchers.
Zhou et al. (Mon,) studied this question.