ABSTRACT This paper describes the development of two improved Canadian homogenized monthly precipitation datasets, the CanHomP mlyV2 station dataset, which includes the entire data record of 425 long-term stations across Canada (since 1840 or later), and its gridded version CanGridP mlyV2, which covers the entire Canadian land mass for the 1949–2023 period and southern Canada for 1916–2023. The latter is subsequently used to provide updated estimates of Canada's historical precipitation trends with an assessment of trend representativeness. The V2 datasets benefit from the use of improved station data and metadata and an improved data homogenization procedure, which together result in better spatial consistency of trends than seen in unhomogenized data. Estimates of precipitation trends based on CanGridP mlyV2 also exhibit temperature scaling rates that are more in line with physical expectations than the previous versions of gridded precipitation datasets, which exhibited unphysically high temperature scaling rates. Precipitation is estimated to have increased in most areas from southern Nunavut to the Arctic Archipelago and from Labrador to northeastern Quebec in all seasons. It has also increased in a zonal band around 62°N in summer, and in most areas in British Columbia and along the St. Lawrence River in spring and autumn. The most outstanding variation of trends in seasonal precipitation is seen in a broad band across southern Canada, where winter precipitation has decreased significantly without extensively significant changes in the other seasons. The best estimate of increase in the period 1949–2023 is 9.7% for Canada as a whole, 18.9% for Canada's North, and 7.5% for Canada's South. The estimated rate of change in Canada's annual precipitation expressed as a function of surface air temperature change is 4.9% per 1˚C of warming for the period 1949–2023. Over the century-long period 1916–2023, annual precipitation in Canada's South is estimated to have increased 10.7%.

ABSTRACT The Eurasian continent, specifically Europe, West Asia, East Asia, and the Russian Far East, experienced anomalously high temperatures during the summer of 2022. The physical drivers of this unusually warm summer are investigated using the dominant empirical orthogonal function (EOF) modes of summer 2-m air temperature over Eurasia for the period 1960–2022. The results indicate that the exceptionally warm conditions were primarily linked to three climatic modes. The first mode, explaining 36.8% of the total variance, represents a background warming trend favouring elevated temperatures. The second mode, accounting for 12.3% of the variance, is closely related to the interannual variability of La Niña conditions and a pronounced sea surface temperature gradient in the North Atlantic. These factors trigger mid–high-latitude wave trains and modulate the Hadley and Walker circulations, collectively fostering persistent anticyclonic anomalies. The third mode, explaining 11.1% of the variance, corresponds to the Pacific Decadal Oscillation (PDO), which triggers a circumglobal teleconnection pattern that guides wave energy to modulate Eurasian circulation. Diagnoses based on Rossby wave source analysis and wave activity flux further support the dynamical roles of these modes.

ABSTRACT Characterised by rapid development, flash droughts challenge timely responses and severely affect agricultural productivity, ecological stability, and socioeconomic resilience, making them a critical research focus. Existing studies reveal substantial spatiotemporal disparities in flash droughts identified by different definitions. Although definitional effectiveness varies regionally, systematic evaluation of their regional applicability remains limited. Flash drought definition involves two key components: drought indicator selection and the objective identification method. This study developed four definitions by combining two indicators – the soil moisture (SM) and evapotranspiration stress ratio (ESR) – with two methods: based on the change rate of percentiles (CROP) and based on percentiles of variation (POV). We evaluated definitional impacts on drought identification and validated results against documented flash droughts (2022–2024). The results show that drought indicators exert greater influence than identification methods. SM and ESR show distinct spatial patterns in identifying flash drought hotspots across China. POV criteria are stricter than CROP, requiring further threshold investigation. SM better identifies “high evapotranspiration demand, fast evapotranspiration rates, and relatively low initial SM” events, particularly in the humid region (south of Qinling-Huaihe Line), where flash droughts often manifest as “high-temperature-driven”, as evident in the Yangtze River Basin during summer 2022 and in central Sichuan in 2023. In contrast, ESR outperforms for detecting “high onset-stage evapotranspiration demand but limited evapotranspiration due to already very low initial SM” events, which are common in the transition region (north of the Qinling-Huaihe Line and south of Inner Mongolia), where flash droughts are typically “water-deficit-driven”, as observed in the Huanghuai region in summer 2024.

ABSTRACT Changes in lake ice cover resulting from systematic perturbations to individual meteorological forcing variables are examined here by way of numerical experimentation with a 1-dimensional thermodynamic lake model. Examination of a simplified vertical energy budget suggests that wind speed, air temperature, precipitation, and incoming shortwave radiation are key variables governing the creation and evolution of ice. Synthetic 30-year meteorological forcing datasets over a small boreal lake are generated by replicating 1 year of detailed observations with added Gaussian noise or by a scaling factor to each of these forcing variables in turn, and the impact on lake ice phenology, quality, and maximum thickness analysed. For the wind speed experiments, changes in phenology were nonlinear and asymmetric. For the largest wind speed reductions ice-on was delayed but for increasing mean wind speed perturbations, the ice-on date was essentially unchanged. For large wind speed perturbations of either sign the ice-off date was early, but smaller changes in mean wind speed, of either sign, had no effect. Thus, any significant change in mean wind speed would lead to a reduction in ice cover duration. Ice-on dates were only weakly affected by perturbations to any of the other forcing variables considered, including air temperature. Thus, observational studies that link increasing air temperatures to delays in ice-on should also consider the impacts of wind speed if data are available. Changes in mean air temperature led to changes in ice thickness and duration. Increasing precipitation was found to increase ice thickness as well as the fraction of white ice, while changing mean insolation had a significant impact on ice-off.

ABSTRACT Speculation surrounds the possibility that increasing world temperatures could trigger a decrease in the heat supplied to the Subpolar North Atlantic (SPNA) by the Gulf Stream (GS) and its extension, the North Atlantic Current (NAC). Here we provide evidence for just such a drop in heat and salt delivery in 2014, leading to the development of a cold fresh anomaly (CFA) in the SPNA and reciprocal warm salty anomaly (WSA) near North America in 2015. These coincident dipoles of heat and salt anomalies first appeared in February–March 2014 when cooling and freshening in the NAC near 45oN, 42oW was reciprocated by warming and increased salinity near 42oN, 50oW. The negative anomalies subsequently migrated north-eastwards along the path of the NAC to constitute the CFA centered on 50oN, 30oW by July 2015 while the reciprocal positive anomalies spread south-westwards to constitute the WSA centered on 40oN, 70oW. Sea surface temperature and height anomaly data revealed a rectangular three-fold structure for the CFA representing the path of the cooled and freshened NAC across the Mid-Atlantic Ridge. Surface current velocity data for early 2014 showed that northwards transport in the NAC was temporarily reversed near its exit from the Mann Eddy, possibly due to an influx of Labrador Sea Water from the Deep Western Boundary Current. We conclude that the unprecedented cooling/freshening of the SPNA and warming/salinification of the North American coastal region in 2014–2015 were reciprocal consequences of an interruption of northwards mass transport in the NAC near its exit from the Mann Eddy.

ABSTRACT Limited long-term snowfall observations make it difficult to document how snowfall is changing across Canada. Proxy snowfall measures derived from more plentiful temperature and precipitation may therefore be helpful. We consider simple partitioning of daily precipitation into rainfall and snowfall based on whether temperature is above or below either 0°C or a station specific threshold. Using daily mean temperature and the fixed 0°C threshold resulted in more accurate estimates of annual and seasonal snow-day number and water equivalent snowfall amount than using daily maximum or daily minimum temperature. Using station-specific thresholds further improved estimation accuracy. Trends estimated from these proxy snowfall indices well match those estimated from observed snowfall data for periods and locations when both are available. The median annual proxy snowfall amount in Canada derived from homogenized daily precipitation and temperature data decreased 2.5% per decade over 1949–2023 south of 60°N and increased 0.5% per decade north of 60°N. Seasonally, annual proxy snowfall amount has changed most rapidly in winter, declining 2.6% per decade in southern Canada and increasing 3.6% per decade in northern Canada. This simple approach improves prospects for the continuation of long-term snowfall monitoring in Canada by exploiting long-term daily precipitation and temperature data.

ABSTRACT Marine fog varies on annual, decadal, and climate change scales, with implications on transportation and the global radiative budget. Using reanalysis and airport meteorological data from 1953 to 2019, this study investigates these long-term variations along the Canadian Atlantic coast and its underlying drivers. A shift in dominant drivers is observed in the early 1990s: prior to that, sea-level pressure moderately correlated with annual fog at Sable Island (R = 0.58, p < 0.001), whereas sea-surface temperature (SST) became the primary influence afterward, with a significant negative correlation (R = −0.55, p = 0.003). This change coincides with a rapid warming of SST along the Scotian Shelf, which reduced the air–sea temperature contrast necessary for fog formation. Annual fog frequency also declined significantly over time, with trends of −25 to −45 h per decade across the six coastal stations studied. These trends were most pronounced in the foggiest period of the year: spring and summer. In addition to ocean warming, a weakening of near-surface temperature inversions and long-term rise in boundary layer height (BLH) suggest reduced atmospheric stability as a key mechanism limiting fog formation. These stability indicators co-vary with fog on interannual timescales and reinforce the role of stratification in supporting marine fog. This study highlights the evolving role of fog drivers in a changing climate and offers a physical basis to improve future fog projections.

ABSTRACT This paper describes the development of an updated Canadian homogenized monthly mean wind speed dataset, CanHomW mlyV2, for the period 1953–2023 and characterizes observed changes in surface wind speed across Canada. Hourly data from 154 stations in Canada were first quality controlled and adjusted for any non-standard anemometer heights. Then, monthly mean wind speed series were derived and subject to a semi-automated comprehensive data homogenization procedure to identify and diminish non-climatic changes. The procedure uses a combination of station metadata and multiple statistical tests with and without using reference series. The results of the automated procedure were reviewed manually. All of the 154 data series were identified to have one or more non-climatic changes, which were diminished by quantile matching adjustments. Station relocation and/or joining (i.e. joining of different stations’ data records into one data series), and instrument changes/problems were found to be the main causes of non-climatic changes. The homogenized dataset shows weakening winds in a large part of southern Canada (spanning from the southern Prairies to Labrador) and strengthening winds in most other regions, particularly in the area that spans south-central British Columbia to the Rocky Mountains. The weakening winds in the southern Prairies are also seen consistently in the three modern reanalysis datasets (ERA5, OCADA, 20CRv3), while the four datasets show inconsistent trends in most of the other regions. The Canadian wind trends show notable seasonality, as do the agreement/disagreement among the four datasets.

ABSTRACT A numerical ocean model with biogeochemistry has been developed for a domain that spans oceans around Canada: extending to 26°N in the Atlantic and 44°N in the Pacific and including the whole Arctic Ocean. The spatial resolution is ∼0.25° in longitude/latitude with 75 vertical levels. A series of simulations was conducted to assess the best choices for biogeochemical model parameters across the diverse regions, using a variety of validation data sets including satellite ocean colour (surface chlorophyll and particulate organic carbon, integrated primary production), surface underway pCO2, and depth profiles of oxygen and nitrate concentration from ships and Argo floats. In addition to parameter values, processes examined include interactive sediments, fluvial nutrients, light attenuation by fluvial coloured dissolved organic matter (CDOM), and iron limitation. The results indicate that the optimal parameter set is one that limits phytoplankton losses to grazing and other processes so as to ensure strong biological drawdown of dissolved inorganic carbon and nutrients in spring and summer; both insufficient and excessive drawdown were observed. Sensitivity to other processes such as interactive sediments, fluvial nutrients or CDOM attenuation was weak in most regions. Fluvial nutrients can cause localized reduction of pCO2 by as much as 60 μatm, and attenuation by CDOM or sequestration of nutrients in the sediment can reduce primary production and zooplankton biomass in regions with large river inputs or broad continental shelves. Iron limitation has a non-negligible effect on the model solution even in regions generally considered iron-replete; a model that successfully spans iron-limited and non-iron-limited domains will require complete and accurate specification of iron sources and sinks.

ABSTRACT The Last Ice Area (LIA) of the Arctic Ocean, spanning the northern coasts of the Canadian Arctic Archipelago and Greenland, is expected to host the last perennial sea ice in the Arctic this century. Knowledge about the long-term evolution of the ocean structure in the LIA is limited due to sparse observations, restricting our ability to identify and understand past change and the impacts of recent climate warming. This study uses a 63-year ocean circulation hindcast simulation from 1958–2021, evaluated with available ocean and sea ice observations, to understand interdecadal changes in the region. Despite this region being resilient to the 20th century sea ice loss observed in most other regions of the Arctic, the model shows that the LIA's sea ice began to experience losses starting in the 2010s, aligned with satellite observations. During the study period, the modelled ocean structure in the LIA was increasingly modified by Atlantic Water (AW), with overall warming influenced by low-frequency variability. Modelled AW metrics, including AW core temperature, core depth, and layer thickness initially decrease from the start of the model run to a minimum in 1970, then increase to reach maximum AW influence on the water column in the 2000s, before subsequently dipping in the 2010s. Most of these modelled ocean properties are broadly consistent with available observations, such as in the AW core, which warmed by 0.3–0.5 degrees to 0.8 ∘ C in the model and 0.7 ∘ C in the observations. The model suggests a slight cooling and thinning of the AW layer after 2010; however, observations suggest a continued warming of the core through 2023 to 0.78 ∘ C. While the model's representation of stratification and water masses in the upper water column needs improvement, and the model presents an overly vertically diffuse AW layer, the warming of the AW layer and the variability of its position in the water column suggests a trend towards the Atlantification of the LIA, a process observed elsewhere in the Arctic. Given the rapid sea ice loss of the past decade and changing ocean properties such as stratification and heat content, further ocean changes are likely. Improved and timely monitoring and modelling is imperative to understanding current changes and potential impacts in the region.