The Influence of Snow and Ice Albedo towards Improved Lake Ice Simulations

Lake ice models are a vital tool for studying the response of ice-covered lakes to changing climates throughout the world. The Canadian Lake Ice Model (CLIMo) is a one-dimensional freshwater ice cover model that simulates Arctic and sub-Arctic lake ice cover well. Modelling ice cover in temperate regions has presented challenges due to the differences in ice composition between northern and temperate region lake ice. This study presents a comparison of measured and modelled ice regimes, with a focus on refining CLIMo for temperate regions. The study sites include two temperate region lakes (MacDonald Lake and Clear Lake, Central Ontario) and two High Arctic lakes (Resolute Lake and Small Lake, Nunavut) where climate and ice cover information have been recorded over three seasons. The ice cover simulations were validated with a combination of time lapse imagery, field measurements of snow depth, snow density, ice thickness and albedo data, and historical ice records from the Canadian Ice Database (for Resolute Lake). Simulations of High Arctic lake ice cover show good agreement with previous studies for ice-on and ice-off dates (MAE 6 to 8 days). Unadjusted simulations for the temperate region lakes show good ice-on timing, but an under-representation of ice thickness, and earlier complete ice-off timing (~3 to 5 weeks). Field measurements were used to adjust the albedo values used in CLIMo, which resulted in improvements to both simulated ice thickness (~3 cm MAE compared to manual measurements), and ice-off timing, within 0 to 7 days (2 days MAE) of observations. These findings suggest regionally specific measurements of albedo can improve the accuracy of lake ice simulations, which further our knowledge of the response of temperate and High Arctic lake ice regimes to climate conditions.

Lake Ice-Water Classification of RADARSAT-2 Images by Integrating IRGS Segmentation with Pixel-Based Random Forest Labeling

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.

Decadal Variability of Great Lakes Ice Cover in Response to AMO and PDO, 1963–2017

In this study, decadal variability of ice cover in the Great Lakes is investigated using historical airborne and satellite measurements from 1963 to 2017. It was found that Great Lakes ice cover has 1) a linear relationship with the Atlantic multidecadal oscillation (AMO), similar to the relationship of lake ice cover with the North Atlantic Oscillation (NAO), but with stronger impact than NAO; 2) a quadratic relationship with the Pacific decadal oscillation (PDO), which is similar to the relationship of lake ice cover to Niño-3.4, but with opposite curvature; and 3) decadal variability with a positive (warming) trend in AMO contributes to the decreasing trend in lake ice cover. Composite analyses show that during the positive (negative) phase of AMO, the Great Lakes experience a warm (cold) anomaly in surface air temperature (SAT) and lake surface temperature (LST), leading to less (more) ice cover. During the positive (negative) phase of PDO, the Great Lakes experience a cold (warm) anomaly in SAT and LST, leading to more (less) ice cover. Based on these statistical relationships, the original multiple variable regression model established using the indices of NAO and Niño-3.4 only was improved by adding both AMO and PDO, as well as their interference (interacting or competing) mechanism. With the AMO and PDO added, the correlation between the model and observation increases to 0.69, compared to 0.48 using NAO and Niño-3.4 only. When November lake surface temperature was further added to the regression model, the prediction skill of the coming winter ice cover increased even more.

LAKE ICE MONITORING WITH WEBCAMS

Continuous monitoring of climate indicators is important for understanding the dynamics and trends of the climate system. Lake ice has been identified as one such indicator, and has been included in the list of Essential Climate Variables (ECVs). Currently there are two main ways to survey lake ice cover and its change over time, in-situ measurements and satellite remote sensing. The challenge with both of them is to ensure sufficient spatial and temporal resolution. Here, we investigate the possibility to monitor lake ice with video streams acquired by publicly available webcams. Main advantages of webcams are their high temporal frequency and dense spatial sampling. By contrast, they have low spectral resolution and limited image quality. Moreover, the uncontrolled radiometry and low, oblique viewpoints result in heavily varying appearance of water, ice and snow. We present a workflow for pixel-wise semantic segmentation of images into these classes, based on state-of-the-art encoder-decoder Convolutional Neural Networks (CNNs). The proposed segmentation pipeline is evaluated on two sequences featuring different ground sampling distances. The experiment suggests that (networks of) webcams have great potential for lake ice monitoring. The overall per-pixel accuracies for both tested data sets exceed 95 %. Furthermore, per-image discrimination between ice-on and ice-off conditions, derived by accumulating per-pixel results, is 100 % correct for our test data, making it possible to precisely recover freezing and thawing dates.