This study introduces the first use of Global Navigation Satellite System Reflectometry (GNSS-R) for monitoring lake ice phenology. This is demonstrated using Qinghai Lake, Tibetan Plateau, as a case study. Signal-to-Noise Ratio (SNR) values obtained from the Cyclone GNSS (CYGNSS) constellation over four ice seasons (2018 to 2022) were used to examine the impact of lake surface conditions on reflected GNSS signals during open water and ice cover seasons. A moving t-test algorithm was applied to time-varying SNR values allowing for the detection of lake ice at daily temporal resolution. Good agreement was achieved between ice phenology records derived from CYGNSS data and Moderate Resolution Imaging Spectroradiometer (MODIS) imagery. The CYGNSS timings for freeze-up, i.e., the period starting with the first appearance of ice on the lake (freeze-up start; FUS) until the lake becomes fully ice covered (freeze-up end; FUE), as well as those for breakup, i.e., the period beginning with the first pixel of open water (breakup start; BUS) and ending when the whole lake becomes ice-free (breakup end; BUE), were validated against the phenology dates derived from MODIS images. Mean absolute errors are 7, 5, 10, 4 and 5 days for FUS, FUE, BUS, BUE and ice cover duration, respectively. Observations revealed the sensitivity of GNSS reflected signals to surface melt prior to the appearance of open water conditions as determined from MODIS, which explains the larger difference of 10 days for BUS.

Remote retrieval of near-surface chlorophyll-a (Chla) concentration in small inland waters is challenging due to substantial optical interferences of various water constituents and uncertainties in the atmospheric correction (AC) process. Although various algorithms have been developed to estimate Chla from moderate-resolution terrestrial missions (∼10–60 m), the production of both accurate distribution maps and time series of Chla has proven challenging, limiting the use of remote analyses for lake monitoring. Here, we develop a support vector regression (SVR) model, which uses satellite-derived remote-sensing reflectance spectra (Rrsδ) from Sentinel-2 and Landsat-8 images as input for Chla retrieval in a representative eutrophic prairie lake, Buffalo Pound Lake (BPL), Saskatchewan, Canada. Validated against in situ Chla from seven ice-free seasons (N ∼ 200; 2014–2020), the SVR model outperformed both locally tuned, Rrsδ-fed empirical models (Normalized Difference Chlorophyll Index, 2- and 3-band, and OC3) and Mixture Density Networks (MDNs) by 15–65%, while exhibiting comparable performance to a locally trained MDN, with an error of ∼35%. Comparison of Chla retrieval models, AC processors (iCOR, ACOLITE), and radiometric products (Rayleigh-corrected, surface, and top-of-atmosphere reflectance) showed that the best Chla maps and optimal time series (up to 100 mg m−3) were produced using a coupled SVR-iCOR system.

Remote retrieval of near-surface chlorophyll-a (Chla) concentration in small inland waters is challenging due to substantial optical interferences of various water constituents and uncertainties in the atmospheric correction (AC) process. Although various algorithms have been developed to estimate Chla from moderate-resolution terrestrial missions (∼10–60 m), the production of both accurate distribution maps and time series of Chla has proven challenging, limiting the use of remote analyses for lake monitoring. Here, we develop a support vector regression (SVR) model, which uses satellite-derived remote-sensing reflectance spectra (Rrsδ) from Sentinel-2 and Landsat-8 images as input for Chla retrieval in a representative eutrophic prairie lake, Buffalo Pound Lake (BPL), Saskatchewan, Canada. Validated against in situ Chla from seven ice-free seasons (N ∼ 200; 2014–2020), the SVR model outperformed both locally tuned, Rrsδ-fed empirical models (Normalized Difference Chlorophyll Index, 2- and 3-band, and OC3) and Mixture Density Networks (MDNs) by 15–65%, while exhibiting comparable performance to a locally trained MDN, with an error of ∼35%. Comparison of Chla retrieval models, AC processors (iCOR, ACOLITE), and radiometric products (Rayleigh-corrected, surface, and top-of-atmosphere reflectance) showed that the best Chla maps and optimal time series (up to 100 mg m−3) were produced using a coupled SVR-iCOR system.

Lake Erie, the shallowest of the five North American Laurentian Great Lakes, exhibits degraded water quality associated with recurrent phytoplankton blooms. Optical remote sensing of these optically complex inland waters is challenging due to the uncertainties stemming from atmospheric correction (AC) procedures. In this study, the accuracy of remote sensing reflectance (Rrs) derived from three different AC algorithms applied to Ocean and Land Colour Instrument (OLCI) observations of western Lake Erie (WLE) is evaluated through comparison to a regional radiometric dataset. The effects of uncertainties in Rrs products on the retrieval of near-surface concentration of pigments, including chlorophyll-a (Chla) and phycocyanin (PC), from Mixture Density Networks (MDNs) are subsequently investigated. Results show that iCOR contained the fewest number of processed (unflagged) days per pixel, compared to ACOLITE and POLYMER, for parts of the lake. Limiting results to the matchup dataset in common between the three AC algorithms shows that iCOR and ACOLITE performed closely at 665 nm, while outperforming POLYMER, with the Median Symmetric Accuracy (MdSA) of ∼30 %, 28 %, and 53 %, respectively. MDN applied to iCOR- and ACOLITE-corrected data (MdSA < 37 %) outperformed MDN applied to POLYMER-corrected data in estimating Chla. Large uncertainties in satellite-derived Rrs propagated to uncertainties ∼100 % in PC estimates, although the model was able to recover concentrations along the 1:1 line. Despite the need for improvements in its cloud-masking scheme, we conclude that iCOR combined with MDNs produces adequate OLCI pigment products for studying and monitoring Chla across WLE.

Lake Erie, the shallowest of the five North American Laurentian Great Lakes, exhibits degraded water quality associated with recurrent phytoplankton blooms. Optical remote sensing of these optically complex inland waters is challenging due to the uncertainties stemming from atmospheric correction (AC) procedures. In this study, the accuracy of remote sensing reflectance (Rrs) derived from three different AC algorithms applied to Ocean and Land Colour Instrument (OLCI) observations of western Lake Erie (WLE) is evaluated through comparison to a regional radiometric dataset. The effects of uncertainties in Rrs products on the retrieval of near-surface concentration of pigments, including chlorophyll-a (Chla) and phycocyanin (PC), from Mixture Density Networks (MDNs) are subsequently investigated. Results show that iCOR contained the fewest number of processed (unflagged) days per pixel, compared to ACOLITE and POLYMER, for parts of the lake. Limiting results to the matchup dataset in common between the three AC algorithms shows that iCOR and ACOLITE performed closely at 665 nm, while outperforming POLYMER, with the Median Symmetric Accuracy (MdSA) of ∼30 %, 28 %, and 53 %, respectively. MDN applied to iCOR- and ACOLITE-corrected data (MdSA < 37 %) outperformed MDN applied to POLYMER-corrected data in estimating Chla. Large uncertainties in satellite-derived Rrs propagated to uncertainties ∼100 % in PC estimates, although the model was able to recover concentrations along the 1:1 line. Despite the need for improvements in its cloud-masking scheme, we conclude that iCOR combined with MDNs produces adequate OLCI pigment products for studying and monitoring Chla across WLE.

Instrumented buoys are used to monitor water quality, yet there remains a need to evaluate whether in vivo fluorometric measures of chlorophyll a (Chl a) produce accurate estimates of phytoplankton abundance. Here, 6 years (2014–2019) of in vitro measurements of Chl a by spectrophotometry were compared with coeval estimates from buoy-based fluorescence measurements in eutrophic Buffalo Pound Lake, Saskatchewan, Canada. Analysis revealed that fluorometric and in vitro estimates of Chl a differed both in terms of absolute concentration and patterns of relative change through time. Three models were developed to improve agreement between metrics of Chl a concentration, including two based on Chl a and phycocyanin (PC) fluorescence and one based on multiple linear regressions with measured environmental conditions. All models were examined in terms of two performance metrics; accuracy (lowest error) and reliability (% fit within confidence intervals). The model based on PC fluorescence was most accurate (error = 35%), whereas that using environmental factors was most reliable (89% within 3σ of mean). Models were also evaluated on their ability to produce spatial maps of Chl a using remotely sensed imagery. Here, newly developed models significantly improved system performance with a 30% decrease in Chl a errors and a twofold increase in the range of reconstructed Chl a values. Superiority of the PC model likely reflected high cyanobacterial abundance, as well as the excitation–emission wavelength configuration of fluorometers. Our findings suggest that a PC fluorometer, used alone or in combination with environmental measurements, performs better than a single-excitation-band Chl a fluorometer in estimating Chl a content in highly eutrophic waters.

With the increasing availability of SAR imagery in recent years, more research is being conducted using deep learning (DL) for the classification of ice and open water; however, ice and open water classification using conventional DL methods such as convolutional neural networks (CNNs) is not yet accurate enough to replace manual analysis for operational ice chart mapping. Understanding the uncertainties associated with CNN model predictions can help to quantify errors and, therefore, guide efforts on potential enhancements using more–advanced DL models and/or synergistic approaches. This paper evaluates an approach for estimating the aleatoric uncertainty [a measure used to identify the noise inherent in data] of CNN probabilities to map ice and open water with a custom loss function applied to RADARSAT–2 HH and HV observations. The images were acquired during the 2014 ice season of Lake Erie and Lake Ontario, two of the five Laurentian Great Lakes of North America. Operational image analysis charts from the Canadian Ice Service (CIS), which are based on visual interpretation of SAR imagery, are used to provide training and testing labels for the CNN model and to evaluate the accuracy of the model predictions. Bathymetry, as a variable that has an impact on the ice regime of lakes, was also incorporated during model training in supplementary experiments. Adding aleatoric loss and bathymetry information improved the accuracy of mapping water and ice. Results are evaluated quantitatively (accuracy metrics) and qualitatively (visual comparisons). Ice and open water scores were improved in some sections of the lakes by using aleatoric loss and including bathymetry. In Lake Erie, the ice score was improved by ∼2 on average in the shallow near–shore zone as a result of better mapping of dark ice (low backscatter) in the western basin. As for Lake Ontario, the open water score was improved by ∼6 on average in the deepest profundal off–shore zone.

Recent investigations using polarimetric decomposition and numerical models have helped to improve understanding of how radar signals interact with lake ice. However, further research is needed on how radar signals are impacted by varying lake ice properties. Radiative transfer models provide one method of improving this understanding. These are the first published experiments using the Snow Microwave Radiative Transfer (SMRT) model to investigate the response of different imaging SAR frequencies (L, C, and X-band) at HH and VV polarizations using various incidence angles (20°, 30°, and 40°) to changes in ice thickness, porosity, bubble radius, and ice-water interface roughness. This is also the first use of SMRT in combination with a thermodynamic lake ice model. Experiments were for a lake with tubular bubbles and one without tubular bubbles under difference scenarios. Analysis of the backscatter response to different properties indicate that increasing ice thickness and layer porosity have little impact on backscatter from lake ice. X-band backscatter shows increased response to surface ice layer bubble radius; however, this was limited for other frequencies except at shallower incidence angles (40°). All three frequencies display the largest response to increasing RMS height at the ice-water interface, which supports surface scattering at the ice-water interface as being the dominant scattering mechanism. These results demonstrate that SMRT is a valuable tool for understanding the response of SAR data to changes in freshwater lake ice properties and could be used in the development of inversion models.

Cyanobacterial harmful algal blooms are an increasing threat to coastal and inland waters. These blooms can be detected using optical radiometers due to the presence of phycocyanin (PC) pigments. The spectral resolution of best-available multispectral sensors limits their ability to diagnostically detect PC in the presence of other photosynthetic pigments. To assess the role of spectral resolution in the determination of PC, a large ( <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$N =905$ </tex-math></inline-formula> ) database of colocated <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">in situ</i> radiometric spectra and PC are employed. We first examine the performance of selected widely used machine-learning (ML) models against that of benchmark algorithms for hyperspectral remote sensing reflectance ( <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$R_{\mathrm {rs}}$ </tex-math></inline-formula> ) spectra resampled to the spectral configuration of the Hyperspectral Imager for the Coastal Ocean (HICO) with a full-width at half-maximum (FWHM) of < 6 nm. Results show that the multilayer perceptron (MLP) neural network applied to HICO spectral configurations (median errors < 65%) outperforms other ML models. This model is subsequently applied to <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$R_{\mathrm {rs}}$ </tex-math></inline-formula> spectra resampled to the band configuration of existing satellite instruments and of the one proposed for the next Landsat sensor. These results confirm that employing MLP models to estimate PC from hyperspectral data delivers tangible improvements compared with retrievals from multispectral data and benchmark algorithms (with median errors between <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$\sim 73$ </tex-math></inline-formula> % and 126%) and shows promise for developing a globally applicable cyanobacteria measurement approach.

Chlorophyll-a concentration (chla) is a useful indicator of harmful algal blooms in early warning systems that use remote sensing data as input. However, its retrieval is challenging in small waterbodies due to the lack of high spatial-resolution water-color sensors and the substantial optical interference of other water constituents. Here, we demonstrate the potential of support vector machines and Sentinel-2 images to retrieve chla in a shallow eutrophic lake (Buffalo Pound Lake, Saskatchewan, Canada). Following validation against in-situ chla measurements over three open water seasons (2017–2019), our proposed method based on Support Vector Regression (SVR) outperforms the most common semi-empirical models, i.e., locally-tuned indices (normalized difference chlorophyll index, 2band, and 3band), as well as the state-of-the-art global empirical model (Mixture Density Network). The superiority of SVR is shown in terms of overall and stratified accuracy, as well as spatial validity. We argue that for small waterbodies where numerous matched pairs of in-situ chla and reflectance are not available, SVR might retrieve chla more accurately than locally-tuned indices and global empirical models.

Chlorophyll-a concentration (chla) is a useful indicator of harmful algal blooms in early warning systems that use remote sensing data as input. However, its retrieval is challenging in small waterbodies due to the lack of high spatial-resolution water-color sensors and the substantial optical interference of other water constituents. Here, we demonstrate the potential of support vector machines and Sentinel-2 images to retrieve chla in a shallow eutrophic lake (Buffalo Pound Lake, Saskatchewan, Canada). Following validation against in-situ chla measurements over three open water seasons (2017–2019), our proposed method based on Support Vector Regression (SVR) outperforms the most common semi-empirical models, i.e., locally-tuned indices (normalized difference chlorophyll index, 2band, and 3band), as well as the state-of-the-art global empirical model (Mixture Density Network). The superiority of SVR is shown in terms of overall and stratified accuracy, as well as spatial validity. We argue that for small waterbodies where numerous matched pairs of in-situ chla and reflectance are not available, SVR might retrieve chla more accurately than locally-tuned indices and global empirical models.

Landsat 4–5 Thematic Mapper, Landsat 8 Operational Land Imager, and RapidEye-3 data sets were used to identify potential groundwater discharge zones, via icings, in the Central Mackenzie Valley (CMV) of the Northwest Territories. Given that this area is undergoing active shale oil exploration and climatic changes, identification of groundwater discharge zones is of great importance both for pinpointing potential contaminant transport pathways and for characterizing the hydrologic system. Following the work of Morse and Wolfe (2015), a series of image algorithms were applied to imagery for the entire CMV and for the Bogg Creek watershed (a sub watershed of the CMV) for selected years between 2004 and 2017. Icings were statistically examined for all of the selected years to determine whether a significant difference in their spatial occurrence existed. It was concluded that there was a significant difference in the spatial distribution of icings from year to year (α = 0.05), but that there were several places where icings were recurring. During the summer of 2018, these recurrent icings, which are expected to be spring sourced, were verified using a thermal camera aboard a helicopter, as well as in situ measurements of hydraulic gradient, groundwater geochemistry, and electroconductivity. Strong agreement was found between the mapped icings and summer field data, making them ideal field monitoring locations. Furthermore, identifying these discharge points remotely is expected to have drastically reduced the field efforts that would have been required to find them in situ. This work demonstrates the value of remote sensing methods for hydrogeological applications, particularly in remote northern locations.

Landsat 4–5 Thematic Mapper, Landsat 8 Operational Land Imager, and RapidEye-3 data sets were used to identify potential groundwater discharge zones, via icings, in the Central Mackenzie Valley (CMV) of the Northwest Territories. Given that this area is undergoing active shale oil exploration and climatic changes, identification of groundwater discharge zones is of great importance both for pinpointing potential contaminant transport pathways and for characterizing the hydrologic system. Following the work of Morse and Wolfe (2015), a series of image algorithms were applied to imagery for the entire CMV and for the Bogg Creek watershed (a sub watershed of the CMV) for selected years between 2004 and 2017. Icings were statistically examined for all of the selected years to determine whether a significant difference in their spatial occurrence existed. It was concluded that there was a significant difference in the spatial distribution of icings from year to year (α = 0.05), but that there were several places where icings were recurring. During the summer of 2018, these recurrent icings, which are expected to be spring sourced, were verified using a thermal camera aboard a helicopter, as well as in situ measurements of hydraulic gradient, groundwater geochemistry, and electroconductivity. Strong agreement was found between the mapped icings and summer field data, making them ideal field monitoring locations. Furthermore, identifying these discharge points remotely is expected to have drastically reduced the field efforts that would have been required to find them in situ. This work demonstrates the value of remote sensing methods for hydrogeological applications, particularly in remote northern locations.

The topic of satellite remote sensing of lake ice has gained considerable attention in recent years. Optical satellite data from the Moderate Resolution Imaging Spectroradiometer (MODIS) allow for the monitoring of lake ice cover (an Essential Climate Variable or ECV), and dates associated with ice phenology (freeze-up, break-up, and ice cover duration) over large areas in an era where ground-based observational networks have nearly vanished in many northern countries. Ice phenology dates as well as dates of maximum and minimum ice cover extent (for lakes that do not form a complete ice cover in winter or do not totally lose their ice cover in summer) are useful for assessing long-term trends and variability in climate, particularly due to their sensitivity to changes in near-surface air temperature. Existing knowledge-driven (threshold-based) retrieval algorithms for lake ice cover mapping that use top-of-atmosphere (TOA) reflectance products do not perform well under lower solar illumination conditions (i.e. large solar zenith angles), resulting in low TOA reflectance. This research assessed the capability of four machine learning classifiers (i.e. multinomial logistic regression, MLR; support vector machine, SVM; random forest, RF; gradient boosting trees, GBT) for mapping lake ice cover, water and cloud cover during both break-up and freeze-up periods using the MODIS/Terra L1B TOA (MOD02) product. The classifiers were trained and validated using samples collected from 17 large lakes across the Northern Hemisphere (Europe and North America); lakes that represent different characteristics with regards to area, latitude, freezing frequency, and ice duration. Following an accuracy assessment using random k-fold cross-validation (k = 100), all machine learning classifiers using a 7-band combination (visible, near-infrared and shortwave-infrared) were found to be able to produce overall classification accuracies above 94%. Both RF and GBT provided overall and class-specific accuracies above 98% and a more visually accurate depiction of lake ice, water and cloud cover. The two tree-based classifiers offered the most robust spatial transferability over the 17 lakes and performed consistently well across ice seasons. However, only RF was relatively insensitive to the choice of the hyperparameters compared to the other three classifiers. The results demonstrate the potential of RF for mapping lake ice cover globally from MODIS TOA reflectance data. • This study assessed the capability of ML classifiers for lake ice mapping from MOIDS. • RF and GBT produced the best performance in terms of classification accuracies. • RF and GBT offered the most robust spatial and temporal transferability. • RF was insensitive to the choice of the hyperparameters compared to other classifiers. • The results show the potential of RF for mapping lake ice cover globally from MODIS.

Abstract. Northwestern Alaska has been highly affected by changing climatic patterns with new temperature and precipitation maxima over the recent years. In particular, the Baldwin and northern Seward peninsulas are characterized by an abundance of thermokarst lakes that are highly dynamic and prone to lake drainage, like many other regions at the southern margins of continuous permafrost. We used Sentinel-1 synthetic aperture radar (SAR) and Planet CubeSat optical remote sensing data to analyze recently observed widespread lake drainage. We then used synoptic weather data, climate model outputs and lake-ice growth simulations to analyze potential drivers and future pathways of lake drainage in this region. Following the warmest and wettest winter on record in 2017/2018, 192 lakes were identified to have completely or partially drained in early summer 2018, which exceeded the average drainage rate by a factor of ~ 10 and doubled the rates of the previous extreme lake drainage years of 2005 and 2006. The combination of abundant rain- and snowfall and extremely warm mean annual air temperatures (MAAT), close to 0 °C, may have led to the destabilization of permafrost around the lake margins. Rapid snow melt and high amounts of excess meltwater further promoted rapid lateral breaching at lake shores and consequently sudden drainage of some of the largest lakes of the study region that likely persisted for millenia. We hypothesize that permafrost destabilization and lake drainage will accelerate and become the dominant drivers of landscape change in this region. Recent MAAT are already within the range of predictions by UAF SNAP ensemble climate predictions in scenario RCP6.0 for 2100. With MAAT in 2019 exceeding 0 °C at the nearby Kotzebue, Alaska climate station for the first time since continuous recording started in 1949, permafrost aggradation in drained lake basins will become less likely after drainage, strongly decreasing the potential for freeze-locking carbon sequestered in lake sediments, signifying a prominent regime shift in ice-rich permafrost lowland regions.

Abstract The freeze/thaw state of permafrost landscapes is an essential variable for monitoring ecological, hydrological and climate processes. Ground surface state can be obtained from satellite data through time series analysis of C-band backscatter from scatterometer and Synthetic Aperture Radar (SAR) observations. Scatterometer data has been used in a variety of studies concerning freeze/thaw retrieval of the land surface. Coarse spatial resolution scatterometer data has great potential for application in this field due to its high temporal resolution (approx. daily observations). In this study, we investigate the influence of sub-grid cell (12.5 km) surface water (ice free and ice covered) on freeze/thaw retrieval based on ASCAT data using a threshold algorithm. We found discrepancies related to the surface water fraction in the detected timing of thawing and freezing of up to 2 days earlier thawing for spring and 3.5 days earlier freezing for autumn for open water fractions of 40% resulting in an overestimation of the frozen season. Results of this study led to the creation of a method for correction of water fraction impact on freeze/thaw data. Additionally, this study demonstrates the applicability of a new approach to freeze/thaw retrieval which has not so far been tested for SAR, specifically Sentinel-1.

Lake ice thickness is a sensitive indicator of climate change largely through its dependency on near-surface air temperature and on-ice snow mass (depth and density). Monitoring of the seasonal variations and trends in ice thickness is also important for the operation of winter ice roads that northern communities rely on for the movement of goods as well as for cultural and leisure activities (e.g., snowmobiling). Therefore, consistent measurements of ice thickness over lakes is important; however, field measurements tend to be sparse in both space and time in many northern countries. Here, we present an application of L-band frequency Global Navigation Satellite System (GNSS) Interferometric Reflectometry (GNSS-IR) for the estimation of lake ice thickness. The proof of concept is demonstrated through the analysis of Signal-to-Noise Ratio (SNR) time series extracted from Global Positioning System (GPS) constellation L1 band raw data acquired between 8 and 22 March (2017 and 2019) at 14 lake ice sites located in the Northwest Territories, Canada. Dominant frequencies are extracted using Least Squares Harmonic Estimation (LS-HE) for the retrieval of ice thickness. Estimates compare favorably with in-situ measurements (mean absolute error = 0.05 m, mean bias error = −0.01 m, and root mean square error = 0.07 m). These results point to the potential of GPS/GNSS-IR as a complementary tool to traditional field measurements for obtaining consistent ice thickness estimates at many lake locations, given the relatively low cost of GNSS antennas/receivers.

Changes to ice cover on lakes throughout the northern landscape has been established as an indicator of climate change and variability, expected to have implications for both human and environmental systems. Monitoring lake ice cover is also required to enable more reliable weather forecasting across lake-rich northern latitudes. Currently, the Canadian Ice Service (CIS) monitors lakes using synthetic aperture radar (SAR) and optical imagery through visual interpretation, with total lake ice cover reported weekly as a fraction out of ten. An automated method of classification would allow for more detailed records to be delivered operationally. In this research, we present an automatic ice-mapping approach which integrates unsupervised segmentation from the Iterative Region Growing using Semantics (IRGS) algorithm with supervised random forest (RF) labeling. IRGS first locally segments homogeneous regions in an image, then merges similar regions into classes across the entire scene. Recently, these output regions were manually labeled by the user to generate ice maps, or were labeled using a Support Vector Machine (SVM) classifier. Here, three labeling methods (Manual, SVM, and RF) are applied after IRGS segmentation to perform ice-water classification on 36 RADARSAT-2 scenes of Great Bear Lake (Canada). SVM and RF classifiers are also tested without integration with IRGS. An accuracy assessment has been performed on the results, comparing outcomes with author-generated reference data, as well as the reported ice fraction from CIS. The IRGS-RF average classification accuracy for this dataset is 95.8%, demonstrating the potential of this automated method to provide detailed and reliable lake ice cover information operationally.

Lake ice is a dominant component of Canada’s landscape and can act as an indicator for how freshwater aquatic ecosystems are changing with warming climates. While lake ice monitoring through government networks has decreased in the last three decades, the increased availability of remote sensing images can help to provide consistent spatial and temporal coverage for areas with annual ice cover. Synthetic aperture radar (SAR) data are commonly used for lake ice monitoring, due to the acquisition of images in any condition (time of day or weather). Using Sentinel-1 A/B images, a high-density time series of SAR images was developed for Lake Hazen in Nunavut, Canada, from 2015–2018. These images were used to test two different methods of monitoring lake ice phenology: one method using the first difference between SAR images and another that applies the Otsu segmentation method. Ice phenology dates determined from the two methods were compared with visual interpretation of the Sentinel-1 images. Mean errors for the pixel comparison of the first difference method ranged 3–10 days for ice-on and ice-off, while average error values for the Otsu method ranged 2–10 days. Mean errors for comparisons of different sections of the lake ranged 0–15 days for the first difference method and 2–17 days for the Otsu method. This research demonstrates the value of temporally consistent image acquisition for improving the accuracy of lake ice monitoring.

Abstract. Northwestern Alaska has been highly affected by changing climatic patterns with new temperature and precipitation maxima over the recent years. In particular, the Baldwin and northern Seward peninsulas are characterized by an abundance of thermokarst lakes that are highly dynamic and prone to lake drainage like many other regions at the southern margins of continuous permafrost. We used Sentinel-1 synthetic aperture radar (SAR) and Planet CubeSat optical remote sensing data to analyze recently observed widespread lake drainage. We then used synoptic weather data, climate model outputs and lake ice growth simulations to analyze potential drivers and future pathways of lake drainage in this region. Following the warmest and wettest winter on record in 2017/2018, 192 lakes were identified as having completely or partially drained by early summer 2018, which exceeded the average drainage rate by a factor of ∼ 10 and doubled the rates of the previous extreme lake drainage years of 2005 and 2006. The combination of abundant rain- and snowfall and extremely warm mean annual air temperatures (MAATs), close to 0 ∘C, may have led to the destabilization of permafrost around the lake margins. Rapid snow melt and high amounts of excess meltwater further promoted rapid lateral breaching at lake shores and consequently sudden drainage of some of the largest lakes of the study region that have likely persisted for millennia. We hypothesize that permafrost destabilization and lake drainage will accelerate and become the dominant drivers of landscape change in this region. Recent MAATs are already within the range of the predictions by the University of Alaska Fairbanks' Scenarios Network for Alaska and Arctic Planning (UAF SNAP) ensemble climate predictions in scenario RCP6.0 for 2100. With MAAT in 2019 just below 0 ∘C at the nearby Kotzebue, Alaska, climate station, permafrost aggradation in drained lake basins will become less likely after drainage, strongly decreasing the potential for freeze-locking carbon sequestered in lake sediments, signifying a prominent regime shift in ice-rich permafrost lowland regions.

Algorithms for the generation of a bedfast/floating lake ice product from Sentinel-1A/B synthetic aperture radar (SAR) data were implemented, cross-compared, and validated for various permafrost regions (Alaska, Canada and Russia). The algorithms consisted of: 1) thresholding; 2) Iteration Region Growing with Semantics (IRGS); and 3) K-means. The thresholding algorithm (92.4%) was found to perform slightly better on average than the IRGS algorithm (90.1%), and to outperform K-means (85.3%). The thresholding algorithm was therefore selected for implementation of a processing chain to generate a novel bedfast/floating lake ice product. Using a time series of Sentinel-1 SAR data, the new map product shows the day of year (DOY) when the ice becomes bedfast or remains afloat for individual lake sections.

This work describes a pilot study in southern Ontario, Canada evaluating the use of the ‘Headwall Nano-Hyperspec’ hyperspectral imager onboard a Remotely Piloted Aircraft System (RPAS). Hyperspectral imagers are extremely useful for monitoring vegetation health and water quality, among other environmental parameters. However, guidelines on the use of this specific instrument for these applications are not yet available. As such, recommended operational settings and calibration procedures are presented here, based on nearly 50 flight campaigns over water bodies and vineyards. Using these procedures, spectral reflectance was successfully captured using an RPAS.

Canada has vast water resources that span an enormous range in geography, climate, and ecosystems [1]. Water supply and water quality are the two critical issues relevant to water resources, not only in Canada but globally in a warming climate. The water microsatellite mission described here aims to better prepare end users to respond to the emerging spectrum of water futures issues by revolutionizing remote sensing of water quality and quantity parameters, and permitting unprecedented interconnection and data gathering from Canadian environmental monitoring networks.

A winter time series of ground-based (X- and Ku-bands) scatterometer and spaceborne synthetic aperture radar (SAR) (C-band) fully polarimetric observations coincident with in situ snow and ice measurements are used to identify the dominant scattering mechanism in bubbled freshwater lake ice in the Hudson Bay Lowlands near Churchill, Manitoba. Scatterometer observations identify two physical sources of backscatter from the ice cover: the snow–ice and ice–water interfaces. Backscatter time series at all frequencies show increases from the ice–water interface prior to the inclusion of tubular bubbles in the ice column based on in situ observations, indicating scattering mechanisms independent of double-bounce scatter. The co-polarized phase difference of interactions at the ice–water interface from both scatterometer and SAR observations is centered at 0° during the time series, also indicating a scattering regime other than double bounce. A Yamaguchi three-component decomposition of the RADARSAT-2 C-band time series is presented, which suggests the dominant scattering mechanism to be single-bounce off the ice–water interface with appreciable surface roughness or preferentially oriented facets, regardless of the presence, absence, or density of tubular bubble inclusions. This paper builds on newly established evidence of single-bounce scattering mechanism for freshwater lake ice and is the first to present a winter time series of ground-based and spaceborne fully polarimetric active microwave observations with polarimetric decompositions for bubbled freshwater lake ice.

Lake ice is a significant component of the cryosphere due to its large spatial coverage in high-latitude regions during the winter months. The Laurentian Great Lakes are the world’s largest supply of freshwater and their ice cover has a major impact on regional weather and climate, ship navigation, and public safety. Ice experts at the Canadian Ice Service (CIS) have been manually producing operational Great Lakes image analysis charts based on visual interpretation of the synthetic aperture radar (SAR) images. In that regard, we have investigated the performance of the semi-automated segmentation algorithm “glocal” Iterative Region Growing with Semantics (IRGS) for lake ice classification using dual polarized RADARSAT-2 imagery acquired over Lake Erie. Analysis of various case studies indicated that the “glocal” IRGS algorithm could provide a reliable ice-water classification using dual polarized images with a high overall accuracy of 90.4%. However, lake ice types that are based on stage of development were not effectively identified due to the ambiguous relation between backscatter and ice types. The slight improvement of using dual-pol as opposed to single-pol images for ice-water discrimination was also demonstrated.