GLORIA - A globally representative hyperspectral in situ dataset for optical sensing of water quality.

The development of algorithms for remote sensing of water quality (RSWQ) requires a large amount of in situ data to account for the bio-geo-optical diversity of inland and coastal waters. The GLObal Reflectance community dataset for Imaging and optical sensing of Aquatic environments (GLORIA) includes 7,572 curated hyperspectral remote sensing reflectance measurements at 1 nm intervals within the 350 to 900 nm wavelength range. In addition, at least one co-located water quality measurement of chlorophyll a, total suspended solids, absorption by dissolved substances, and Secchi depth, is provided. The data were contributed by researchers affiliated with 59 institutions worldwide and come from 450 different water bodies, making GLORIA the de-facto state of knowledge of in situ coastal and inland aquatic optical diversity. Each measurement is documented with comprehensive methodological details, allowing users to evaluate fitness-for-purpose, and providing a reference for practitioners planning similar measurements. We provide open and free access to this dataset with the goal of enabling scientific and technological advancement towards operational regional and global RSWQ monitoring.

Non-photochemical quenching estimates from in situ spectroradiometer measurements: implications on remote sensing of sun-induced chlorophyll fluorescence in lakes.

Quantum yield of fluorescence (ϕF) is key to interpret remote measurements of sun-induced fluorescence (SIF), and whether the SIF signal is governed by photochemical quenching (PQ) or non-photochemical quenching (NPQ). Disentangling PQ from NPQ allows using SIF estimates in various applications in aquatic optics. However, obtaining ϕF is challenging due to its high temporal and physiological variability, and the combined measurements needed to enclose all relevant optical paths. In inland waters, this type of data is scarce and information on diurnal and seasonal ϕF dynamics are almost unknown. Using an autonomous hyperspectral Thetis profiler in Lake Geneva, we demonstrate how to estimate ϕF using an ensemble of in-situ measurements acquired between 2018 to 2021. We use vertical and temporal changes in retrieved ϕF to determine NPQ and PQ conditions. We observed NPQ in 36% of the total daytime profiles used in the ϕF analysis. While downwelling irradiance is a significant contributor to ϕF, its role cannot be easily interpreted. Other factors such as phytoplankton photoregulation and assemblages also likely play significant roles in quenching mechanisms. We conclude that an adapted approach exploiting in-situ data is suitable to determine diurnal and seasonal NPQ occurrence, and helps develop future remote sensing algorithms.

The value of human data annotation for machine learning based anomaly detection in environmental systems.

Anomaly detection is the process of identifying unexpected data samples in datasets. Automated anomaly detection is either performed using supervised machine learning models, which require a labelled dataset for their calibration, or unsupervised models, which do not require labels. While academic research has produced a vast array of tools and machine learning models for automated anomaly detection, the research community focused on environmental systems still lacks a comparative analysis that is simultaneously comprehensive, objective, and systematic. This knowledge gap is addressed for the first time in this study, where 15 different supervised and unsupervised anomaly detection models are evaluated on 5 different environmental datasets from engineered and natural aquatic systems. To this end, anomaly detection performance, labelling efforts, as well as the impact of model and algorithm tuning are taken into account. As a result, our analysis reveals the relative strengths and weaknesses of the different approaches in an objective manner without bias for any particular paradigm in machine learning. Most importantly, our results show that expert-based data annotation is extremely valuable for anomaly detection based on machine learning.

The Imprint of Primary Production on High-Frequency Profiles of Lake Optical Properties.

Water inherent optical properties (IOPs) contain integrative information on the optical constituents of surface waters. In lakes, IOP measurements have not been traditionally collected. This study describes how high-frequency IOP profiles can be used to document short-term physical and biogeochemical processes that ultimately influence the long-term trajectory of lake ecosystems. Between October 2018 and May 2020, we collected 1373 high-resolution hyperspectral IOP profiles in the uppermost 50 m of the large mesotrophic Lake Geneva (Switzerland-France), using an autonomous profiler. A data set of this size and content does not exist for any other lake. Results showed seasonal variations in the IOPs, following the expected dynamic of phytoplankton. We found systematic diel patterns in the IOPs. Phases of these diel cycles were consistent year-round, and amplitudes correlated to the diurnal variations of dissolved oxygen, clarifying the link between IOPs and phytoplankton metabolism. Diel amplitudes were largest in spring and summer under low wind condition. Wind-driven changes in thermal stratification impacted the dynamic of the IOPs, illustrating the potential of high-frequency profiles of water optical properties to increase our understanding of carbon cycling in lake ecosystems.

Long-term impacts of nutrient control, climate change, and invasive clams on phytoplankton and cyanobacteria biomass in a large temperate river.

Recent studies suggest that climate change, with warmer water temperatures and lower and longer low flows, may enhance harmful planktic cyanobacterial growth in lakes and large rivers. Concomitantly, controlling nutrient loadings has proven effective in reducing phytoplankton biomass especially in North America and Western Europe. In addition, the impact of invasive benthic filter-feeder species such as Corbicula on phytoplankton has largely been overlooked in large rivers, leading to even more uncertainty in predicting future trajectories in river water quality. To investigate how nutrient control, climate change and invasion of benthic filter-feeders may affect phytoplankton biomass and composition, we assembled a large database on the entire water course of the River Loire (France) over three decades (1991-2019). We focus on cyanobacteria to provide an in-depth analysis of the 30-year trend and insights on future possible trajectories. Since 1991, total phytoplankton and cyanobacteria biomasses have decreased 10-fold despite warmer water temperature (+0.23 °C·decade-1) and lower summer flow (-0.25 L·s-1·km-2·decade-1). In the long-term, the contribution of planktic cyanobacteria to total biomass was on average 2.8%. The main factors driving total phytoplankton and cyanobacteria biomasses were total phosphorus (4-fold decrease), the abundance of Corbicula clams (from absence before 1998 to 250-1250 individuals·m-2 after 2010), the duration of summer low flows and the intensity of summer heatwaves. The River Loire constitutes an example in Europe of how nutrient control can be an efficient mitigation strategy, counteracting already visible effects of climate change on the thermal regime and flow pattern of the river. This may hold true under future conditions, but further work is needed to account for the climate trajectory, land and water use scenarios, the risk of enhanced benthic biofilm and macrophyte proliferation, together with the spread of invasive filter-feeding bivalves.

Integrating Inland and Coastal Water Quality Data for Actionable Knowledge.

Water quality measures for inland and coastal waters are available as discrete samples from professional and volunteer water quality monitoring programs and higher-frequency, near-continuous data from automated in situ sensors. Water quality parameters also are estimated from model outputs and remote sensing. The integration of these data, via data assimilation, can result in a more holistic characterization of these highly dynamic ecosystems, and consequently improve water resource management. It is becoming common to see combinations of these data applied to answer relevant scientific questions. Yet, methods for scaling water quality data across regions and beyond, to provide actionable knowledge for stakeholders, have emerged only recently, particularly with the availability of satellite data now providing global coverage at high spatial resolution. In this paper, data sources and existing data integration frameworks are reviewed to give an overview of the present status and identify the gaps in existing frameworks. We propose an integration framework to provide information to user communities through the the Group on Earth Observations (GEO) AquaWatch Initiative. This aims to develop and build the global capacity and utility of water quality data, products, and information to support equitable and inclusive access for water resource management, policy and decision making.

Nutrient inputs and hydrology together determine biogeochemical status of the Loire River (France): Current situation and possible future scenarios.

The Grafs-Seneque/Riverstrahler model was implemented for the first time on the Loire River for the 2002-2014 period, to explore eutrophication after improvement of wastewater treatments. The model reproduced the interannual levels and seasonal trends of the major water quality variables. Although eutrophication has been impressively reduced in the drainage network, a eutrophication risk still exists at the coast, as shown by the N-ICEP indicator, pointing out an excess of nitrogen over silica and phosphorus. From maximum biomass exceeding 120 µgChla l-1 in the 1980's, we observed decreasing maximum values from 80 to 30 µgChla l-1 during the period studied. Several scenarios were explored. Regarding nutrient point sources, a low wastewater treatment scenario, similar to the situation in the 1980's, was elaborated, representing much greater pollution than the reference period (2002-2014). For diffuse sources, two agricultural scenarios were elaborated for reducing nitrogen, one with a strict application of the agricultural directives and another investigating the impact of radical structural changes in agriculture and the population's diet. Although reduced, a risk of eutrophication would remain, even with the most drastic scenario. In addition, a pristine scenario, with no human activity within the basin, was devised to assess water quality in a natural state. The impact of a change in hydrology on the Loire biogeochemical functioning was also explored according to the effect of climate change by the end of the 21st century. The EROS hydrological model was used to force Riverstrahler, considering the most pessimistic SRES A2 scenario run with the ARPEGE model. Nutrient fluxes all decreased due to a >50% reduction in the average annual discharge, overall reducing the risk of coastal eutrophication, but worsening the water quality status of the river network. The Riverstrahler model could be useful to help water managers contend with future threats in the Loire River, at the scale of its basin and at smaller nested scales.