Following more than two decades of research and developments made possible through various proof-of-concept hyperspectral remote sensing missions, it has been anticipated that hyperspectral imaging ...would enhance the accuracy of remotely sensed in-water products. This study investigates such expected improvements and demonstrates the utility of hyperspectral radiometric measurements for the retrieval of near-surface phytoplankton properties11Codes are accessible via https://github.com/STREAM-RS/STREAM-RS, i.e., phytoplankton absorption spectra (aph) and biomass evaluated through examining the concentration of chlorophyll-a (Chla). Using hyperspectral data (409–800 nm at ~5 nm resolution) and a class of neural networks known as Mixture Density Networks (MDN) (Pahlevan et al., 2020), we show that the median error in aph retrievals is reduced two-to-three times (N = 722) compared to that from heritage ocean color algorithms. The median error associated with our aph retrieval across all the visible bands varies between 20 and 30%. Similarly, Chla retrievals exhibit significant improvements (i.e., more than two times; N = 1902), with respect to existing algorithms that rely on select spectral bands. Using an independent matchup dataset acquired near-concurrently with the acquisition of the Hyperspectral Imager for the Coastal Ocean (HICO) images, the models are found to perform well, but at reduced levels due to uncertainties in the atmospheric correction. The mapped spatial distribution of Chla maps and aph spectra for selected HICO swaths further solidify MDNs as promising machine-learning models that have the potential to generate highly accurate aquatic remote sensing products in inland and coastal waters. For aph retrieval to improve further, two immediate research avenues are recommended: a) the network architecture requires additional optimization to enable a simultaneous retrieval of multiple in-water parameters (e.g., aph, Chla, absorption by colored dissolved organic matter), and b) the training dataset should be extended to enhance model generalizability. This feasibility analysis using MDNs provides strong evidence that high-quality, global hyperspectral data will open new pathways toward a better understanding of biodiversity in aquatic ecosystems.
•Machine-learning models for aph and chlorophyll-a retrievals are devised.•Performance of the model is independently evaluated with in situ data.•Models provide significant improvements over the state-of-the-art methods.•Model is implemented to proof-of-concept HICO images.•Chlorophyll-a maps and aph spectra are analyzed.
Constructing multi-source satellite-derived water quality (WQ) products in inland and nearshore coastal waters from the past, present, and future missions is a long-standing challenge. Despite ...inherent differences in sensors’ spectral capability, spatial sampling, and radiometric performance, research efforts focused on formulating, implementing, and validating universal WQ algorithms continue to evolve. This research extends a recently developed machine-learning (ML) model, i.e., Mixture Density Networks (MDNs) (Pahlevan et al., 2020; Smith et al., 2021), to the inverse problem of simultaneously retrieving WQ indicators, including chlorophyll-a (Chla), Total Suspended Solids (TSS), and the absorption by Colored Dissolved Organic Matter at 440 nm (acdom(440)), across a wide array of aquatic ecosystems. We use a database of in situ measurements to train and optimize MDN models developed for the relevant spectral measurements (400–800 nm) of the Operational Land Imager (OLI), MultiSpectral Instrument (MSI), and Ocean and Land Color Instrument (OLCI) aboard the Landsat-8, Sentinel-2, and Sentinel-3 missions, respectively. Our two performance assessment approaches, namely hold-out and leave-one-out, suggest significant, albeit varying degrees of improvements with respect to second-best algorithms, depending on the sensor and WQ indicator (e.g., 68%, 75%, 117% improvements based on the hold-out method for Chla, TSS, and acdom(440), respectively from MSI-like spectra). Using these two assessment methods, we provide theoretical upper and lower bounds on model performance when evaluating similar and/or out-of-sample datasets. To evaluate multi-mission product consistency across broad spatial scales, map products are demonstrated for three near-concurrent OLI, MSI, and OLCI acquisitions. Overall, estimated TSS and acdom(440) from these three missions are consistent within the uncertainty of the model, but Chla maps from MSI and OLCI achieve greater accuracy than those from OLI. By applying two different atmospheric correction processors to OLI and MSI images, we also conduct matchup analyses to quantify the sensitivity of the MDN model and best-practice algorithms to uncertainties in reflectance products. Our model is less or equally sensitive to these uncertainties compared to other algorithms. Recognizing their uncertainties, MDN models can be applied as a global algorithm to enable harmonized retrievals of Chla, TSS, and acdom(440) in various aquatic ecosystems from multi-source satellite imagery. Local and/or regional ML models tuned with an apt data distribution (e.g., a subset of our dataset) should nevertheless be expected to outperform our global model.
•A machine-learning model for retrieval of water quality indicators was developed.•Model was trained/tested using co-located in situ water quality and radiometric data.•It outperforms best-practice algorithms for a wide range of water quality conditions.•Performance is demonstrated for near-simultaneous images of OLI, MSI, and OLCI.•Model sensitivity to uncertainties in Rrs is evaluated for OLI and MSI matchups.
The Landsat series has marked the history of Earth observation by performing the longest continuous imaging program from space. The recent Landsat-9 carrying Operational Land Imager 2 (OLI-2) ...captures a higher dynamic range than sensors aboard Landsat-8 or Sentinel-2 (14-bit vs. 12-bit) that can potentially push forward the frontiers of aquatic remote sensing. This potential stems from the enhanced radiometric resolution of OLI-2, providing higher sensitivity over water bodies that are usually low-reflective. This study performs an initial assessment on retrieving water quality parameters from Landsat-9 imagery based on both physics-based and machine learning modeling. The concentration of chlorophyll-a (Chl-a) and total suspended matter (TSM) are retrieved based on physics-based inversion in four Italian lakes encompassing oligo to eutrophic conditions. A neural network-based regression model is also employed to derive Chl-a concentration in San Francisco Bay. We perform a consistency analysis between the constituents derived from Landsat-9 and near-simultaneous Sentinel-2 imagery. The Chl-a and TSM retrievals are validated using in situ matchups. The results indicate relatively high consistency among the water quality products derived from Landsat-9 and Sentinel-2. However, the Landsat-9 constituent maps show less grainy noise, and the matchup validation indicates relatively higher accuracies obtained from Landsat-9 (e.g., TSM R2 of 0.89) compared to Sentinel-2 (R2 = 0.71). The improved constituent retrieval from Landsat-9 can be attributed to the higher signal-to-noise (SNR) enabled by the wider dynamic range of OLI-2. We performed an image-based SNR estimation that confirms this assumption.
Optical types of inland and coastal waters Spyrakos, Evangelos; O’Donnell, Ruth; Hunter, Peter D. ...
Limnology and oceanography,
March 2018, Volume:
63, Issue:
2
Journal Article
Peer reviewed
Open access
Inland and coastal waterbodies are critical components of the global biosphere. Timely monitoring is necessary to enhance our understanding of their functions, the drivers impacting on these ...functions and to deliver more effective management. The ability to observe waterbodies from space has led to Earth observation (EO) becoming established as an important source of information on water quality and ecosystem condition. However, progress toward a globally valid EO approach is still largely hampered by inconsistences over temporally and spatially variable in-water optical conditions. In this study, a comprehensive dataset from more than 250 aquatic systems, representing a wide range of conditions, was analyzed in order to develop a typology of optical water types (OWTs) for inland and coastal waters. We introduce a novel approach for clustering in situ hyperspectral water reflectance measurements (n = 4045) from multiple sources based on a functional data analysis. The resulting classification algorithm identified 13 spectrally distinct clusters of measurements in inland waters, and a further nine clusters from the marine environment. The distinction and characterization of OWTs was supported by the availability of a wide range of coincident data on biogeochemical and inherent optical properties from inland waters. Phylogenetic trees based on the shapes of cluster means were constructed to identify similarities among the derived clusters with respect to spectral diversity. This typification provides a valuable framework for a globally applicable EO scheme and the design of future EO missions.
This paper presents an application of a physic-based method that relies on spectral inversion procedures to simultaneously estimate concentrations of water constituents, water column heights (cH) and ...benthic substrate types in Lake Trasimeno (Italy) from airborne imaging spectrometry. Complex waters of this lake are challenging due to the coexistence of optically-deep turbid waters and of optically-shallow waters, mostly characterised by dense submerged aquatic vegetation (SAV) beds. Airborne data acquired on 12 May 2009 by Multispectral Infrared and Visible Imaging Spectrometer (MIVIS) were converted into remote sensing reflectance Rrs(λ) with the atmospheric correction code ATCOR. A spectral inversion procedure implementing a bio-optical model (namely BOMBER), parameterised with in situ data, was firstly run to retrieve concentrations of suspended particulate matter (SPM), chlorophyll-a (chl-a) and coloured dissolved organic matter (i.e. aCDOM(440)) in the optically-deep waters. The areas where the retrieved optimisation error was higher than 10% were instead assumed as optically-shallow. In these areas two maps depicting the linear unmixing of three substrate types (i.e., silty-clay, Chara ssp. and other hydrophyte) and the water column heights were produced. The MIVIS-derived products were validated with field data providing a reliable estimation of SPM, chl-a, aCDOM(440) and cH (determination coefficients always R2>0.7). SPM concentrations were also similar to a 5.4-km long transect of flow-through turbidity data, and the SAV map was comparable to in situ observations. Generally, the colonisation patterns of SAV were reflecting the spatial distribution of SPM concentrations. In particular, the positive role of Chara on keeping SPM concentrations low was observed. Future research should extend this application to remote sensing data acquired in other seasons to trace the dynamics of SAV and its effect on spatial water clarity.
•MIVIS enabled discriminating optically deep turbid waters and vegetated substrates•The accuracy of physic-based retrieval of water quality parameters provided R2>0.7•The positive role of Chara spp. on keeping low SPM concentrations was assessed
Slow and long-term variations of sea surface temperature anomalies have been interpreted as a red-noise response of the ocean surface mixed layer to fast and random atmospheric perturbations. How ...fast the atmospheric noise is damped depends on the mixed layer depth. In this work we apply this theory to determine the relevant spatial and temporal scales of surface layer thermal inertia in lakes. We fit a first order auto-regressive model to the satellite-derived Lake Surface Water Temperature (LSWT) anomalies in Lake Garda, Italy. The fit provides a time scale, from which we determine the mixed layer depth. The obtained result shows a clear spatial pattern resembling the morphological features of the lake, with larger values (7.18± 0.3 m) in the deeper northwestern basin, and smaller values (3.18 ± 0.24 m) in the southern shallower basin. Such variations are confirmed by in-situ measurements in three monitoring points in the lake and connect to the first Empirical Orthogonal Function of satellite-derived LSWT and chlorophyll-a concentration. Evidence from our case study open a new perspective for interpreting lake-atmosphere interactions and confirm that remotely sensed variables, typically associated with properties of the surface layers, also carry information on the relevant spatial and temporal scales of mixed-layer processes.
Over the past half century, the demand for sand and gravel has led to extensive quarrying activities, creating many pit lakes (PLs) which now dot floodplains and urbanized regions globally. Despite ...the potential importance of these environments, systematic data on their location, morphology and water quality remain limited. In this study, we present an extensive assessment of the physical and optical properties in a large sample of PLs located in the Po River basin (Italy) from 1990 to 2021, utilizing a combined approach of remote sensing (Landsat constellation and Sentinel-2) and traditional limnological techniques. Specifically, we focused on the concentration of Suspended Particulate Matter (SPM) and the dominant wavelength (λdom, i.e., water colour). This study aims to contribute to the analysis of PLs at a basin scale as an opportunity for environmental rehabilitation and river floodplain management. ACOLITE v.2022, a neural network particularly suitable for the analysis of turbid waters and small inland water bodies, was used to atmospherically correct satellite images and to obtain SPM concentration maps and the λdom. The results show a very strong correlation between SPM concentrations obtained in situ and those obtained from satellite images, both for data derived from Landsat (R2 = 0.85) and Sentinel-2 images (R2 = 0.82). A strong correlation also emerged from the comparison of spectral signatures obtained in situ via WISP-3 and those derived from ACOLITE, especially in the visible spectrum (443–705 nm, SA = 10.8°). In general, it appeared that PLs with the highest mean SPM concentrations and the highest mean λdom are located along the main Po River, and more generally near rivers. The results also show that active PLs exhibit a poor water quality status, especially those of small sizes (<5 ha) and directly connected to a river. Seasonal comparison shows the same trend for both SPM concentration and λdom: higher values in winter gradually decreasing until spring–summer, then increasing again. Finally, it emerged that the end of quarrying activity led to a reduction in SPM concentration from a minimum of 43% to a maximum of 72%. In this context, the combined use of Landsat and Sentinel-2 imagery allowed for the evaluation of the temporal evolution of the physical and optical properties of the PLs in a vast area such as the Po River basin (74,000 km2). In particular, the Sentinel-2 images consistently proved to be a reliable resource for capturing episodic and recurring quarrying events and portraying the ever-changing dynamics of these ecosystems.
The Water Framework Directive requires European states to monitor the ecological quality of their lakes. Detailed information on the composition and abundance of biological groups such as aquatic ...plants (macrophytes) and phytoplankton (including chlorophyll
a
) must be expressed as an ecological quality ratio (EQR), ranging from 1 (close to reference status) to 0 (bad status). Effort is often focused on gathering this detailed information on selected lakes at the expense of more synoptic approaches that could capture a more holistic assessment of a catchment’s water quality. This could be rectified if remote sensing can provide predictions of ecological quality for unmonitored lakes. We found that data from Sentinel-2 satellites, based on regression model outputs of observed vs estimated results, successfully predicted the macrophyte EQR (
R
2
= 0.77) and the maximum lake depth that macrophytes colonised to (
R
2
= 0.80) but average chlorophyll
a
was less well predicted (
R
2
= 0.66). Predictions for a test catchment indicated that results were within one ecological assessment class width of measured values for macrophytes. This approach can potentially estimate status for unmonitored lakes in Ireland, be integrated with results on monitored lakes and used to direct resources where needed at national and catchment scales.
Satellite multi-sensor data were used to investigate the evolution in time and space of Lake Trasimeno, a shallow and turbid lake in central Italy. Large-swath MERIS and MODIS sensors were proposed ...for regular broad scale monitoring of water quality, having compared the retrieved chlorophyll-a (Chl-a) concentration, Secchi disk (SD) depth and surface water temperature with the 2005-2008 time-series of the in situ data. Although, in a shorter time span, also the MERIS-derived total suspended matter (TSM) matched the in situ data. MERIS-derived water quality products confirmed the meso-eutrophic conditions of Lake Trasimeno (average Chl-a = 8.5 mg/m³) and the low levels of transparency (average SD = 1 m). A negative correlation found between water levels and Chl-a suggest the importance of maintaining water levels as close as possible to the hydrometric zero. A spatial analysis of TSM also reveals how small tributaries may affect the load of suspended solids in the southern part of the lake. Higher spatial resolution satellite images were exploited both to describe land use/cover transformation from 1978 to 2008 and to assess the recent changes in macrophyte colonisation patterns. Land cover change detection analysis results showed a decrease in cultivated areas starting from the early Nineties and the subsequent increase in unproductive terrain (bare land and pastures) and natural woods as well as the changing fragmentation of agricultural areas through time. A reduction in macrophyte beds from 2003 to 2008 was also observed. We expect the results of this study to support local water authorities in redrawing the management plan of Lake Trasimeno.
In this study we evaluate the capabilities of three satellite sensors for assessing water composition and bottom depth in Lake Garda, Italy. A consistent physics-based processing chain was applied to ...Moderate Resolution Imaging Spectroradiometer (MODIS), Landsat-8 Operational Land Imager (OLI) and RapidEye. Images gathered on 10 June 2014 were corrected for the atmospheric effects with the 6SV code. The computed remote sensing reflectance (Rrs) from MODIS and OLI were converted into water quality parameters by adopting a spectral inversion procedure based on a bio-optical model calibrated with optical properties of the lake. The same spectral inversion procedure was applied to RapidEye and to OLI data to map bottom depth. In situ measurements of Rrs and of concentrations of water quality parameters collected in five locations were used to evaluate the models. The bottom depth maps from OLI and RapidEye showed similar gradients up to 7 m (r = 0.72). The results indicate that: (1) the spatial and radiometric resolutions of OLI enabled mapping water constituents and bottom properties; (2) MODIS was appropriate for assessing water quality in the pelagic areas at a coarser spatial resolution; and (3) RapidEye had the capability to retrieve bottom depth at high spatial resolution. Future work should evaluate the performance of the three sensors in different bio-optical conditions.