This study develops a 73-year dataset of water balance components from 1950 to 2022 for the Laurentian Great Lakes Basins. This is carried out using the Large Lakes Statistical Water Balance Model (L2SWBM), which provides a Bayesian statistical framework that assimilates binational input datasets sourced from the United States and Canada. The L2SWBM infers feasible water balance component estimates through this Bayesian framework by constraining the output with a standard water balance equation. The result is value-added time series, including expressions of uncertainty, that ultimately close the water balance across the interconnected Great Lakes system. Therefore, the L2SWBM facilitates the understanding of discrepancies in datasets and hydroclimate parameters. This enhanced reliability stemming from coordinated data, with an understanding and quantification of uncertainty, could significantly boost confidence in decision support tools for water resources practitioners and policymakers. This joint effort advances scientific understanding and strengthens strategies and policies designed to bolster resilience in Great Lakes communities and its ecosystem in the face of a shifting climate.

Environment and Climate Change Canada (ECCC) is in the process of deploying a continental-scale hydrological prediction system known as the National Surface and River Prediction System (NSRPS). Operating at a resolution of 30 arc seconds, NSRPS currently generates Ensemble Streamflow Predictions (ESPs) for over four million grid points. Issued once per day, the ensemble is composed of twenty members and provides 16-day forecasts. NSRPS differs from other hydrological forecasting systems in operational use in Canada by being continental in scope and by relying on an Earth System Modelling (ESM) approach for prediction. In order to assess the value of forecasts issued by NSRPS, a comparison is performed with a similar ESP system available over all of Canada: the Global Flood Awareness System (GloFAS) from the European Centre for Medium-Range Weather Forecasts (ECMWF). The evaluation focusses on the Great Lakes and St. Lawrence watershed as well as the Nelson and Churchill watersheds, each over one million km² in size. 393 streamflow stations are identified where NSRPS and GloFAS agree on the watershed delineation. The comparison is limited to the Spring, Summer and Fall of 2022 due to NSRPS forecast data availability. For most stations, NSRPS performs better than GloFAS in terms of Continuous Ranked Probability Score (CRPS), but the median of the potential CRPS across the 393 stations is very similar for days 3-16. Both systems suffer from a lack of spread, particularly for short lead times, but the problem is slightly more acute for GloFAS. Bayesian Model Averaging (BMA) is explored in order to obtain calibrated probabilistic forecasts that perform better than both NSRPS and GloFAS.

Operational flood forecasting in Canada is a provincial responsibility that is carried out by several entities across the country. However, the increasing costs and impacts of floods require better and nationally coordinated flood prediction systems. A more coherent flood forecasting framework for Canada can enable implementing advanced prediction capabilities across the different entities with responsibility for flood forecasting. Recently, the Canadian meteorological and hydrological services were tasked to develop a national flow guidance system. Alongside this initiative, the Global Water Futures program has been advancing cold regions process understanding, hydrological modeling, and forecasting. A community of practice was established for industry, academia, and decision-makers to share viewpoints on hydrological challenges. Taken together, these initiatives are paving the way towards a national flood forecasting framework. In this article, forecasting challenges are identified (with a focus on cold regions), and recommendations are made to promote the creation of this framework. These include the need for cooperation, well-defined governance, and better knowledge mobilization. Opportunities and challenges posed by the increasing data availability globally are also highlighted. Advances in each of these areas are positioning Canada as a major contributor to the international operational flood forecasting landscape. This article highlights a route towards the deployment of capacities across large geographical domains.

This study investigates the benefit of the dynamically zoned target release (DZTR) reservoir model to improve upon the natural lake model for storage and outflow simulations of regulated reservoirs. Simulations were performed with the GEM-Hydro hydrologic model over the Saskatchewan River Basin to study different implementation scenarios for DZTR, applicable for varying degrees of available observed data. Results show that DZTR brings significant improvements upon the natural lake model. Outflow simulations are always better with DZTR than with the natural lake model, while storage performances are sensitive to the implementation methodology considered. Even in the absence of observed lake level data, storage simulations with DZTR are more realistic than the natural lake model. Automatic calibration of the 76 DZTR model parameters did not significantly improve results upon the general implementation methodology. Given the performance variability across reservoirs and the subjective steps involved in the DZTR implementation, it seems risky to implement DZTR without any flow observations downstream of a reservoir.

As part of the National Hydrological Services Transformation Initiative, Environment and Climate Change Canada (ECCC) designed and implemented the National Surface and River Prediction System (NSRPS) in order to provide surface and river flow analysis and forecast products across Canada. Within NSRPS, the Canadian Land Data Assimilation System (CaLDAS) produces snow analyses that are used to initialise the land surface model, which in turn is used to force the river routing component. Originally, CaLDAS was designed to improve atmospheric forecasts with less focus on hydrological processes. When snow data assimilation occurs, the related increments remove/add water from/to the system, which can sometimes be problematic for streamflow forecasting, in particular during the snowmelt period. In this study, a new snow analysis method introduces multiple innovations that respond to the need for higher quality snow analyses for hydrological purposes, including the use of IMS snow cover extent data instead of in situ snow depth observations. The results show that the new snow assimilation methodology brings an overall improvement to snow analyses and substantially enhances water conservation, which is reflected in the generally improved streamflow simulations. This work represents a first step towards a new snow data assimilation process in CaLDAS, with the final objective of producing a reliable snow analysis to initialise and improve NWP as well as environmental predictions, including flood and drought forecasts.

This dataset provides gridded model simulations in NetCDF format over the Lake Erie using the GEM-Hydro model done within the Great Lakes Runoff Inter-comparison Project for Lake Erie (GRIP-E). The data are produced with SPS (GEM-Surf + SVS, the surface component of GEM-Hydro) open-loop runs with the SVS calibrated parameters obtained during GRIP-E project. For more information on the model and on calibration methodology, see GEM-Hydro section in Mai et al. 2020 (in prep.). The original model outputs had all variables accumulated for each day. During post-processing all variables have been de-accumulated by subtracting the accumulation of the previous hour from the accumulation of the current hour. Two variables (ALAT and O1) are also only valid over the land tile of each grid cell. Two additional variables (ALAT_full and O1_full) valid now over the whole grid cell have been added for convenience of the users. Domain boundaries (WGS84 system): - lon_min = -85.5, lon_max = -77.94 - lat_min = 40.3, lat_max = 44.26 Resolution of model variables provided: - spatial: ~10km x 10km - temporal: hourly Simulation period: - 01 Jan 2011 - 31 Dec 2014 - 01 Jan 2010 - 31 Dec 2010 (warm-up) Meteorological input data: - RDRS-v1; see Mai et al. 2020 (in prep) Variables available: float PR_0(time, rlat, rlon) ; PR_0:units = "m" ; PR_0:long_name = "Quantity of precipitation (valid over whole grid cell)" ; float AHFL_0(time, rlat, rlon) ; AHFL_0:units = "mm" ; AHFL_0:long_name = "Surface evaporation (valid over whole grid cell)" ; float TRAF_60268832(time, rlat, rlon) ; TRAF_60268832:units = "mm" ; TRAF_60268832:long_name = "Surface runoff (valid over whole grid cell)" ; float ALAT_0(time, rlat, rlon) ; ALAT_0:units = "mm" ; ALAT_0:long_name = "Accumulation of total soil lateral flow (valid over land tile of grid cell)" ; float ALAT_0_full(time, rlat, rlon) ; ALAT_0_full:units = "mm" ; ALAT_0_full:long_name = "Accumulation of total soil lateral flow (valid over whole grid cell)" ; float O1_0(time, rlat, rlon) ; O1_0:units = "mm" ; O1_0:long_name = "Accumulation of base drainage (valid over land tile of grid cell)" ; float O1_0_full(time, rlat, rlon) ; O1_0_full:units = "mm" ; O1_0_full:long_name = "Accumulation of base drainage (valid over whole grid cell)" ; float WT_59868832(time, rlat, rlon) ; WT_59868832:units = "1" ; WT_59868832:long_name = "Fraction of grid cell covered with land" ; =============================================================== These data and model runs have been performed under the Great Lakes Runoff Inter-comparison Project for Lake Erie (GRIP-E) led by Juliane Mai and Bryan Tolson (both University of Waterloo) and funded under the Integrated Modelling Program for Canada (IMPC) within the Global Water Futures program.

Spaceborne L-Band observations have been included in the Canadian Land Data Assimilation System (CaLDAS). The impact of these measurements on land surface analyses, hydrological forecasts, and numerical weather prediction has been investigated. Results indicate that the assimilation of L-Band brightness temperature leads to substantially better soil moisture analyses and hydrological forecasts, compared with control experiments in which only screen-level observations are used. The impact on numerical weather prediction is more mixed, with reduced errors at the surface for most regions of the world, but larger errors in the atmospheric boundary layer.

Abstract In this time of a changing climate, it is important to know whether lake levels will rise, potentially causing flooding, or river flows will dry up during abnormally dry weather. The Great Lakes region is the largest freshwater lake system in the world. Moreover, agriculture, industry, commerce, and shipping are active in this densely populated region. Environment and Climate Change Canada (ECCC) recently implemented the Water Cycle Prediction System (WCPS) over the Great Lakes and St. Lawrence River watershed (WCPS-GLS version 1.0) following a decade of research and development. WCPS, a network of linked models, simulates the complete water cycle, following water as it moves from the atmosphere to the surface, through the river network and into lakes, and back to the atmosphere. Information concerning the water cycle is passed between the models. WCPS is the first short-to-medium-range prediction system of the complete water cycle to be run on an operational basis anywhere. It currently produces two forecasts per day for the next three days. WCPS generally provides reliable results throughout the length of the forecast. The transmission of errors between the component models is reduced by data assimilation. Interactions between the environmental compartments are active. This ongoing intercommunication is valuable for extreme events such as rapid ice freeze-up and flooding or drought caused by abnormal amounts of precipitation. Products include precipitation; evaporation; lake water levels, temperatures, and currents; ice cover; and river flows. These products are of interest to a wide variety of governmental, commercial, and industrial groups, as well as the public.

Between January 2013 and December 2014, water levels on Lake Superior and Lake Michigan-Huron, the two largest lakes on Earth by surface area, rose at the highest rate ever recorded for a 2 year period beginning in January and ending in December of the following year. This historic event coincided with below-average air temperatures and extensive winter ice cover across the Great Lakes. It also brought an end to a 15 year period of persistently below-average water levels on Lakes Superior and Michigan-Huron that included several months of record-low water levels. To differentiate hydrological drivers behind the recent water level rise, we developed a Bayesian Markov chain Monte Carlo (MCMC) routine for inferring historical estimates of the major components of each lake's water budget. Our results indicate that, in 2013, the water level rise on Lake Superior was driven by increased spring runoff and over-lake precipitation. In 2014, reduced over-lake evaporation played a more significant role in Lake Superior's water level rise. The water level rise on Lake Michigan-Huron in 2013 was also due to above-average spring runoff and persistent over-lake precipitation, while in 2014, it was due to a rare combination of below-average evaporation, above-average runoff and precipitation, and very high inflow rates from Lake Superior through the St. Marys River. We expect, in future research, to apply our new framework across the other Laurentian Great Lakes, and to Earth's other large freshwater basins as well.

Long-range atmospheric transport and deposition are important sources of mercury (Hg) to Arctic aquatic and terrestrial ecosystems. We review here recent progress made in the study of the transport, transformation, deposition and reemission of atmospheric Hg in the Canadian Arctic, focusing on field measurements (see Dastoor et al., this issue for a review of modeling studies on the same topics). Redox processes control the speciation of atmospheric Hg, and thus impart an important influence on Hg deposition, particularly during atmospheric mercury depletion events (AMDEs). Bromine radicals were identified as the primary oxidant of atmospheric Hg during AMDEs. Since the start of monitoring at Alert (NU) in 1995, the timing of peak AMDE occurrence has shifted to earlier times in the spring (from May to April) in recent years, and while AMDE frequency and GEM concentrations are correlated with local meteorological conditions, the reasons for this timing-shift are not understood. Mercury is subject to various post-depositional processes in snowpacks and a large portion of deposited oxidized Hg can be reemitted following photoreduction; how much Hg is deposited and reemitted depends on geographical location, meteorological, vegetative and sea-ice conditions, as well as snow chemistry. Halide anions in the snow can stabilize Hg, therefore it is expected that a smaller fraction of deposited Hg will be reemitted from coastal snowpacks. Atmospheric gaseous Hg concentrations have decreased in some parts of the Arctic (e.g., Alert) from 2000 to 2009 but at a rate that was less than that at lower latitudes. Despite numerous recent advances, a number of knowledge gaps remain, including uncertainties in the identification of oxidized Hg species in the air (and how this relates to dry vs. wet deposition), physical–chemical processes in air, snow and water—especially over sea ice—and the relationship between these processes and climate change.

We assembled and applied five models (one of which included three different configurations) to the Lake Michigan basin to improve our understanding of how differences in model skill at simulating total runoff to Lake Michigan relate to model structure, calibration protocol, model complexity, and assimilation (i.e. replacement of simulated discharge with discharge observations into historical simulations), and evaluate historical changes in runoff to Lake Michigan. We found that the performance among these models when simulating total runoff to the lake varied relatively little, despite variability in model structure, spatial representation, input data, and calibration protocol. Relatively simple empirical, assimilative models, including the National Oceanic and Atmospheric Administration (NOAA) Great Lakes Environmental Research Laboratory (GLERL) area ratio-based model (ARM) and the United States Geological Survey (USGS) Analysis of Flows in Networks of CHannels (AFINCH) model, represent efficient and effective approaches to propagating discharge observations into basin-wide (including gaged and ungaged areas) runoff estimates, and may offer an opportunity to improve predictive models for simulating runoff to the Great Lakes. Additionally, the intercomparison revealed that the median of the simulations from non-assimilative models agrees well with assimilative models, suggesting that using a combination of different methodologies may be an appropriate approach for estimating runoff into the Great Lakes. We then applied one assimilative model (ARM) to the Lake Michigan basin and found that there was persistent reduction in the amount of precipitation that becomes runoff following 1998, corresponding to a period of persistent low Lake Michigan water levels. The study was conducted as a first phase of the Great Lakes Runoff Intercomparison Project, a regional binational collaboration that aims to systematically and rigorously assess a variety of models currently used (or that could readily be adapted) to simulate basin-scale runoff to the North American Laurentian Great Lakes. (C) 2014 Elsevier B.V. All rights reserved.

High levels of mercury in marine mammals threaten the health of Arctic inhabitants. Whether the Arctic Ocean (AO) is a sink or a source of atmospheric mercury is unknown. Given the paucity of observations in the Arctic, models are useful in addressing this question. GEOS-Chem and GRAHM, two complex numerical mercury models, present contrasting pictures of atmospheric mercury input to AO at 45 and 108 Mg yr(-1), respectively, and ocean evasion at 90 and 33 Mg yr(-1), respectively. We provide a comprehensive evaluation of GRAHM simulated atmospheric mercury input to AO using mercury observations in air, precipitation and snowpacks, and an analysis of the discrepancy between the two modeling estimates using observations. We discover two peaks in high-latitude summertime concentrations of atmospheric mercury. We show that the first is caused mainly by snowmelt revolatilization and the second by AO evasion of mercury. Riverine mercury export to AO is estimated at 50 Mg yr(-1) based on measured DOC export and at 15.5-31 Mg yr(-1) based on simulated mercury in meltwater. The range of simulated mercury fluxes to and from AO reflects uncertainties in modeling mercury in the Arctic; comprehensive observations in all compartments of the Arctic ecosystem are needed to close the gap.

Introduction Conclusions References

A portion of the highly toxic methylmercury that bioaccumulates in aquatic life is created from mercury entering bodies of water with snowpack meltwater. To determine the importance of meltwater as a source of aquatic mercury, it is necessary to understand the environmental processes that govern the behavior of snowpack-related mercury. In this study we investigate relationships among 5 types of snowpack-related mercury observations and 20 model environmental variables. The observation types are the 24-h fractional loss of mercury from surface snow, and the concentrations of mercury in surface snow, seasonal snowpacks, the snowpack meltwater's ionic pulse, and long-term snowpack-related records. The model environmental variables include those related to atmospheric mercury, insolation, wind, atmospheric stability, snowpack physical characteristics, atmospheric pressure, and solid precipitation. Bivariate and multiple linear regressions were performed twice for each mercury observation type: once with all observations, and once excluding observations from locations where the snowpack's burden of oxidizing and stabilizing halogens is known or presumed to affect snowpack mercury. Since no observations from long-term snowpack-related records were considered affected by halogens, this group of observations was included with the sets of uninfluenced observations and was not discussed with the complete, original sets of observations. When all observations are included, only 37% of their variability can be explained, on average, with significance confidence levels averaging 81%; a separate regression model predicts each mercury observation type. Without the influence of halogens, the regression models are able to explain an average of 79% of the observations' variability with significance confidence levels averaging 97%. The snowpack-related mercury observations are most strongly controlled by the dry and wet depositions of oxidized mercury, and by precipitation. Mercury deposited through wet processes is more strongly retained by snowpacks than mercury deposited through dry processes. Revolatilization of mercury deposited through wet processes may be inhibited through burial by fresh snowfalls and/or by its more central location, compared to that of mercury deposited through dry deposition, within snowpack snow grains. The two depositions of oxidized mercury together explain 84% of the variability in observed concentrations of mercury in surface snow, 52% of the variability of observed concentrations of mercury in seasonal snowpacks and their meltwater's ionic pulse, and only 20% of the variability of observed concentrations of mercury in long-term snowpack-related records; other environmental controls seemingly gain in relevance as time passes. The concentration of mercury in long-term records is apparently primarily affected by latitude; both the primary sources of anthropogenic mercury and the strong upper-level zonal winds are located in the midlatitudes.

Mercury in the Arctic is an important environmental and human health issue. The reliance of Northern Peoples on traditional foods, such as marine mammals, for subsistence means that they are particularly at risk from mercury exposure. The cycling of mercury in Arctic marine systems is reviewed here, with emphasis placed on the key sources, pathways and processes which regulate mercury levels in marine food webs and ultimately the exposure of human populations to this contaminant. While many knowledge gaps exist limiting our ability to make strong conclusions, it appears that the long-range transport of mercury from Asian emissions is an important source of atmospheric Hg to the Arctic and that mercury methylation resulting in monomethylmercury production (an organic form of mercury which is both toxic and bioaccumulated) in Arctic marine waters is the principal source of mercury incorporated into food webs. Mercury concentrations in biological organisms have increased since the onset of the industrial age and are controlled by a combination of abiotic factors (e.g., monomethylmercury supply), food web dynamics and structure, and animal behavior (e.g., habitat selection and feeding behavior). Finally, although some Northern Peoples have high mercury concentrations of mercury in their blood and hair, harvesting and consuming traditional foods have many nutritional, social, cultural and physical health benefits which must be considered in risk management and communication.