Owing to its temporal resolution of 10-day and its polar orbit allowing several crossings over large lakes, the US National Aeronautics and Space Administration (NASA) and the French Centre National d’Etudes Spatiales (CNES) missions including Topex/Poseidon, Jason-1/2/3 demonstrated strong capabilities for the continuous and long-term monitoring (starting in 1992) of large and medium-sized water bodies. However, the presence of heterogeneous targets in the altimeter footprint, such as ice cover in boreal areas, remains a major issue to obtain estimates of water level over subarctic lakes of similar accuracy as over other inland water bodies using satellite altimetry (i.e., R ≥ 0.9 and RMSE ≤ 10 to 20 cm when compared to in-situ water stages). In this study, we aim to automatically identify the Jason-2 altimetry measurements corresponding to open water, ice and transition (water-ice) to improve the estimations of water level during freeze and thaw periods using only the point measurements of open water. Four Canadian lakes were selected to analyze active (waveform parameters) and passive (brightness temperature) microwave data acquired by the Jason-2 radar altimetry mission: Great Slave Lake, Lake Athabasca, Lake Winnipeg, and Lake of the Woods. To determine lake surface states, backscattering coefficient and peakiness at Ku-band derived from the radar altimeter waveform and brightness temperature at 18.7 and 37 GHz measured by the microwave radiometer contained in the geophysical data records (GDR) of Jason-2 were used in two different unsupervised classification techniques to define the thresholds of discrimination between open water and ice measurements. K-means technique provided better results than hierarchical clustering based upon silhouette criteria and the Calinski-Harabz index. Thresholds of discrimination between ice and water were validated with the Normalized Difference Snow Index (NDSI) snow cover products of the MODIS satellite. The use of open water threshold resulted in improved water level estimation compared to in situ water stages, especially in the presence of ice. For the four lakes, the Pearson coefficient (r) increased on average from about 0.8 without the use of the thresholds to more than 0.90. The unbiased RMSE were generally lower than 20 cm when the threshold of open water was used and more than 22 cm over smaller lakes, without using the thresholds.

ReSUMe La presence de cibles heterogenes, comme la glace, reste un defi majeur pour l'utilisation des donnees altimetriques au-dessus des plans d'eau continentaux. Les satellites Jason-2 et SARAL/Altika utilisent des algorithmes de retraitement concus pour traiter les formes d'onde non continentales afin d'obtenir des estimations ameliorees. Dans cette etude, nous analysons le potentiel des produits derives de ces algorithmes pour estimer le niveau d'eau de 20 plans d'eau couverts par la glace repartis a travers le Canada. Les estimations de niveaux d'eau des algorithmes de retraitement sont comparees aux mesuresin situpendant deux periodes: la periode entierement couverte par chacun des deux satellites dans l'etude (2008-2016 pour Jason-2, et 2013-2016 pour SARAL/Altika); ainsi que les periodes de gel-degel incluses dans les series chronologiques. Les algorithmes produisent des incertitudes tres variables, en fonction de la taille des cours d'eau et des conditions de la glace. Dans l'ensemble, l'algorithme ICE-1 utilise par Jason-2 fournit les meilleures estimations de niveau d'eau, avec des erreurs RMSE non biaisees <= 0.3 m et des R2 >= 0.8 pour 90% des plans d'eau. Tous les algorithmes de retraitement utilises par SARAL/Altika donnent des resultats tres comparables aux observationsin situ, demontrant les bonnes performances de la technologie SARAL. The presence of heterogeneous targets, such as ice, remains a major challenge for the use of altimetric data over inland water bodies. Jason-2 and SARAL/Altika satellites use retracking algorithms designed for oceanic waveforms to retrieve water levels from altimeter measurements. Nevertheless, the accuracy of these approaches need to be assessed over continental waters in northern latitudes, particularly during the presence of partial or almost complete ice cover. In this study, we compared water level estimates of different retracking algorithms within situmeasurements over 20 ice-covered lakes and rivers across Canada. This comparison was done for the following two periods: (1) the time series considered for each satellite (2008-2016 for Jason-2, and 2013-2016 for SARAL/Altika); and (2) the freeze-thaw periods included in each time series. The results showed that the algorithms produce variable uncertainties depending on the size of the water bodies and the ice conditions. ICE-1 (used with Jason-2 data) provided better water level accuracy for 90% of the water bodies studied (R-2 >= 0.8, unbiased RMSE <= 0.3 m). All the retracking algorithms used by SARAL/Altika provided results that are comparable toin situobservations, thus denoting the good performance of the SARAL technology.

Lake Saint-Pierre (LSP), the largest fluvial lake in the St-Lawrence River system, is a vital freshwater floodplain with rich biodiversity. However, expanding agricultural activities have disrupted the landscape, increasing turbidity and threatening the ecosystem. In response, the Quebec government launched the LSP strategic research cluster in 2018 to encourage sustainable land and water management. This study examines how different land-use types affected turbidity during the spring floods of 2019, 2020, and 2022, using Sentinel-2 satellite imagery. Land types included conventional and improved agriculture, cultivated and natural grasslands, and flooded forests. A new empirical model for flooded forests showed strong accuracy (adjusted R2 of 0.88, RMSE (root mean square error) of 10.91 FNU (Formazin Nephelometric Unit)). Combined with an existing model for open water, turbidity maps were created across all land types. Using a linear mixed model, we found that conventional and improved practices increased turbidity in the LSP floodplain by up to 600% compared with natural forests. Grasslands also contributed to higher turbidity, though to a lesser extent. The findings underscore that even improved practices cannot fully mitigate turbidity. Effective control requires integrated watershed management, including upstream inputs and targeted best practices to protect the LSP floodplain.

Ice cover on subarctic lakes is an important indicator of climate change at local- or regional scales. This study proposes a new approach for classifying altimetry data from Jason-2 and SARAL/Altika satellite missions to characterize surface states of subarctic lakes. It focuses on Great Slave Lake (Canada) during freeze-up and thaw periods. For the first time, parameters from altimetry waveforms were used in an unsupervised clustering to establish distinct clusters from waveforms observed from Jason-2 and SARAL/Altika during the freeze-up and the thaw period. Clusters are assigned to the different surface states (open water, pure ice and leads) based on a priori altimetry and radiometric information. The statistics of these clusters were then used to construct two trained models of supervised classification based upon KNN (K-nearest neighbour) and SVM (support vector machine). The SVM-based model yielded the best results (accuracy of 92% with Jason-2, and 98% with SARAL/Altika). It was used to classify all waveforms considered in the study from the nominal orbits of Jason-2 (2008–2016) and SARAL/Altika (2013–2016). Results were superimposed onto Moderate Resolution Imaging Spectroradiometer (MODIS) products for qualitative visual and semi-quantitative assessments.