Algal biomass mapping of eutrophic lakes using a machine learning approach with MODIS images

Machine learning for cyanobacteria inversion via remote sensing and AlgaeTorch in the Třeboň fishponds, Czech Republic

Cyanobacteria blooms in fishponds, driven by climate change and anthropogenic activities, have become a critical concern for aquatic ecosystems worldwide. The diversity in fishpond sizes and fish densities further complicates their monitoring. This study addresses the challenge of accurately predicting cyanobacteria concentrations in turbid waters via remote sensing, hindered by optical complexities and diminished light signals. A comprehensive dataset of 740 sampling points was compiled, encompassing water quality metrics (cyanobacteria levels, total chlorophyll, turbidity, total cell count) and spectral data obtained through AlgaeTorch, alongside Sentinel-2 reflectance data from three Třeboň fishponds (UNESCO Man and Biosphere Reserve) in the Czech Republic over 2022-2023. Partial Least Squares Regression (PLSR) and three machine learning algorithms, Random Forest (RF), Support Vector Machine (SVM), and Extreme Gradient Boosting (XGBoost), were developed based on seasonal and annual data volumes. The SVM algorithm demonstrated commendable performance on the one-year data validation dataset from the Svět fishpond for the prediction of cyanobacteria, reflected by the key performance indicators: R2 = 0.88, RMSE = 15.07 μg Chl-a/L, and RPD = 2.82. Meanwhile, SVM displayed steady results in the unified one-year validation dataset from Naděje, Svět, and Vizír fishponds, with metrics showing R2 = 0.56, RMSE = 39.03 μg Chl-a/L, RPD = 1.50. Thus, Sentinel data proved viable for seasonal cyanobacteria monitoring across different fishponds. Overall, this study presents a novel approach for enhancing the precision of cyanobacteria predictions and long-term ecological monitoring in fishponds, contributing significantly to the water quality management strategies in the Třeboň region.

Satellite mapping reveals phytoplankton biomass's spatio-temporal dynamics and responses to environmental factors in a eutrophic inland lake

Chlorophyll a (Chla) concentration can be used as an indicator of algal biomass, and the accumulation of algal biomass in water column is essential for the emergence of surface blooms. By using Moderate Resolution Imaging Spectrometer (MODIS) data, a machine learning algorithm was previously developed to assess algal biomass within the euphotic depth (Beu). Here, a long-term Beu dataset of Lake Taihu from 2003 to 2020 was generated to examine its spatio-temporal dynamics, sensitivity to environmental factors, and variations in comparison to the surface algal bloom area. During this period, the daily Beu (total Beu within the whole lake) exhibited temporal fluctuations between 40 and 90 t Chla, with an annual average of 63.32 ± 5.23 t Chla. Notably, it reached its highest levels in 2007 (72.34 t Chla) and 2017 (73.57 t Chla). Moreover, it demonstrated a clear increasing trend of 0.197 t Chla/y from 2003 to 2007, followed by a slight decrease of 0.247 t Chla/y after 2017. Seasonal variation showed a bimodal annual cycle, characterized by a minor peak in March ∼ April and a major peak in July ∼ September. Spatially, the average pixel-based Beu (total Beu of a unit water column) ranged from 21.17 to 49.85 mg Chla, with high values predominantly distributed in the northwest region and low values in the central region. The sensitivity of Beu to environmental factors varies depending on regions and time scales. Temperature has a significant impact on monthly variation (65.73%), while the level of nutrient concentrations influences annual variation (55.06%). Wind speed, temperature, and hydrodynamic conditions collectively influence the spatial distribution of Beu throughout the entire lake. Algal bloom biomass can capture trend changes in two mutant years as well as bimodal phenological changes compared to surface algal bloom area. This study can provide a basis for scientific evaluation of water environment and a reference for monitoring algal biomass in other similar eutrophic lakes.

Estimation of water quality variables based on machine learning model and cluster analysis-based empirical model using multi-source remote sensing data in inland reservoirs, South China

Reservoirs play important roles in the drinking water supply for urban residents, agricultural water provision, and the maintenance of ecosystem health. Satellite optical remote sensing of water quality variables in medium and micro-sized inland waters under oligotrophic and mesotrophic status is challenging in terms of the spatio-temporal resolution, weather conditions and frequent nutrient status changes in reservoirs, etc., especially when quantifying non-optically active components (non-OACs). This study was based on the surface reflectance products of unmanned aerial vehicle (UAV) multispectral images, Sentinel-2B Multispectral instrument (MSI) images and Landsat 7 Enhanced Thematic Mapper Plus (ETM+) by utilizing fuzzy C-means (FCM) clustering algorithm was combined with band combination (BC) model to construct the FCM-BC empirical model, and used mixed density network (MDN), extreme gradient boosting (XGBoost), deep neural network (DNN) and support vector regression (SVR) machine learning (ML) models to invert 12 kinds of optically active components (OACs) and non-OACs. Compared with the unclustered BC (UC) model, the mean coefficient of determination (MR) of the FCM-BC models was improved by at least 46.9%. MDN model showed best accuracy (R2 in the range of 0.60-0.98) and stability (R2 decreased by up to 13.2%). The accuracy of UAV was relatively higher in both empirical methods and machine learning methods. Additionally, the spatio-temporal distribution maps of four water quality variables were mapped based on the MDN model and UAV images, all platforms showed good consistency. An inversion strategy of water quality variables in various monitoring frequencies and weather conditions were proposed finally. The purpose of introducing the UAV platform was to cooperate with the satellite to improve the monitoring response ability of OACs and non-OACs in small and micro-sized oligotrophic and mesotrophic water bodies.

Distribution pattern of dissolved organic matter in pore water of sediments from three typical areas of western Lake Taihu and its environmental implications

The migration, transformation, and accumulation of dissolved organic matter (DOM) in pore water of sediment cores play a pivotal role in lacustrine carbon cycling. In order to understand the dynamics of DOM in the sediments of large shallow eutrophic lakes, we examined the vertical profiles of DOM and the benthic fluxes of dissolved organic carbon (DOC) in sediment cores located in algae accumulated, dredged, and central areas of eutrophic Lake Taihu, China. Optical properties showed the significant influence of terrestrial inputs on the DOM components of pore water in the algae accumulated area but an abundant accumulation of autochthonous DOM in the central area. The benthic fluxes of DOC ranging from -458.2 to -139.4 mg·m-2·d-1 in the algae accumulated area displayed an opposite diffusion direction to the other two areas. The flux ranges of 9.5-31.2 mg·m-2·d-1 in the dredged area and 14.6-48.0 mg·m-2·d-1 in the central area were relatively smaller than those in the previously reported lake ecosystems with low trophic levels. Dredging engineering disturbed the pre-dredging distribution patterns of DOM in sediment cores. The deposition, accumulation, and transformation of massive algae scums in eutrophic lakes probably promoted the humification degree of sediments.

MODIS Terra and Aqua images bring non-negligible effects to phytoplankton blooms derived from satellites in eutrophic lakes

Phytoplankton-induced lake eutrophication has drawn ongoing interest on a global scale. One of the most popular remote sensing satellite data for observing long-term dynamic changes in phytoplankton is Moderate-resolution Imaging Spectroradiometer (MODIS). However, it is worth noting that MODIS provides two images with different transit times: Terra (local time, about 10:30 am) and Aqua (local time, about 1:30 pm), which may result in a considerable bias in monitoring phytoplankton bloom areas due to the rapid migration of phytoplankton under wind or hydrodynamic conditions. To analyze this quantitatively, we selected MODIS Terra and Aqua images to generate datasets of phytoplankton bloom areas in Lake Taihu from 2003 to 2022. The results showed that Terra more frequently detected larger ranges of phytoplankton blooms than Aqua, whether on daily, monthly, or annual scales. In addition, long-term trend changes, seasonal characteristics, and abrupt years also varied with different transit times. Terra detected mutation years earlier, while Aqua displayed more pronounced seasonal characteristics. There were also differences in sensitivity to climate factors, with Terra being more responsive to temperature and wind speed on monthly and annual scales, while Aqua was more sensitive to nutrient and meteorological factors. These conclusions have also been further confirmed in Lake Chaohu, Lake Dianchi, and Lake Hulun. In conclusion, our findings strongly advocate for a linear relationship to fit Terra to Aqua results to mitigate long-term monitoring errors of phytoplankton blooms in inland lakes (R2 = 0.70, RMSE = 101.56). It is advised to utilize satellite data with transit times between 10 am and 1 pm to track phytoplankton bloom changes and to consider the diverse applications resulting from the transit times of Terra and Aqua.