Statistically downscaled multi-model ensembles of minimum temperature are available at a 10km spatial resolution for 1951-2100. Statistically downscaled ensembles are based on output from twenty-four Coupled Model Intercomparison Project Phase 5 (CMIP5) global climate models (GCM). Daily minimum temperature from GCM outputs were downscaled using the Bias Correction/Constructed Analogues with Quantile mapping version 2 (BCCAQv2). A historical gridded minimum temperature dataset of Canada (ANUSPLIN) was used as the downscaling target. The 5th, 25th, 50th, 75th and 95th percentiles of the monthly, seasonal and annual ensembles of downscaled minimum temperature (°C) are available for the historical time period, 1951-2005, and for emission scenarios, RCP2.6, RCP4.5 and RCP8.5, for 2006-2100. Note: Projections among climate models can vary because of differences in their underlying representation of earth system processes. Thus, the use of a multi-model ensemble approach has been demonstrated in recent scientific literature to likely provide better projected climate change information.

Each pixel value corresponds to the quality control, cloud cover and snow fraction value for each pixel in the Best-Quality Max-NDVI product.

The “Prairie Agricultural Landscapes (PAL)” datasets identify areas of the agricultural portions of the Canadian Prairies with similar land and water resources, land use and farming practices. They are represented by vector polygons. Based on selected attributes from the Soil Landscapes of Canada (SLC) and the 1996 Census of Agriculture, the Prairies were classified into 13 (thirteen) classes of Land Practices Group and five (5) Major Land Practices Groups. Typical attributes used to define the Land Practice Groups include: land in pasture, land in summerfallow, crop mixture, farm size and the level of chemical and fertilizer inputs. The five (5) Major Groups were devised to help better understand the relationships between the groups.

The national wetland layer contains wetland data compiled from the best available data from each region, classified by wetland type. Wetlands are mapped as polygons in geographic layers, which are integrated into a master geodatabase at the national scale.Information from each contributing dataset was classified based on the Canadian Wetland Classification System, which contains five main wetland classes (Bog, Fen, Marsh, Swamp, and Shallow Water) that represent the types of wetlands encountered in Canada. An additional category, “partially classified” was used to preserve boundary information for wetlands that could not be classified into the main categories with existing information.

These datasets show the areas where major crops can be expected within the agricultural regions of Canada. Results are provided as rasters with numerical values for each pixel indicating the level of spatial density calculated for a specific crop type in that location. Regions with higher spatial density for a certain crop have higher likelihood to have the same crop based on the previous years mapped crop inventories.

Statistically downscaled multi-model ensembles of maximum temperature are available at a 10km spatial resolution for 1951-2100. Statistically downscaled ensembles are based on output from twenty-four Coupled Model Intercomparison Project Phase 5 (CMIP5) global climate models (GCM). Daily maximum temperature from GCM outputs were downscaled using the Bias Correction/Constructed Analogues with Quantile mapping version 2 (BCCAQv2). A historical gridded maximum temperature dataset of Canada (ANUSPLIN) was used as the downscaling target. The 5th, 25th, 50th, 75th and 95th percentiles of the monthly, seasonal and annual ensembles of downscaled maximum temperature (°C) are available for the historical time period, 1951-2005, and for emission scenarios, RCP2.6, RCP4.5 and RCP8.5, for 2006-2100. Note: Projections among climate models can vary because of differences in their underlying representation of earth system processes. Thus, the use of a multi-model ensemble approach has been demonstrated in recent scientific literature to likely provide better projected climate change information.

The Agri-Environmental Indicator (AEI) dataset series provides information that was created using indicators that assess the environmental impact of agricultural activities. These agri-environmental indicators integrate information on soils, climate and land surface features with statistics on land use and crop and livestock management practices. The datasets provide valuable, location-specific information on the overall environmental risks and conditions in agriculture across Canada and how these change over time. This dataset series collects AEI data that is related to geographic features and can be represented on a map. Other types of AEI data are not included. The datasets can be organized into the following major groups: • Farm land management • Soil health • Water quality • Air quality • Food and beverage industry (not included) Farm land management datasets: • Soil cover • Wildlife habitat • Farm land management (not included) Soil health datasets: • Soil erosion • Soil organic matter • Trace elements • Soil salinity Water quality datasets: • Nitrogen • Phosphorus • Coliforms • Pesticides Air quality datasets: • Greenhouse Gases • Ammonia • Particulate Matter

The Grain Elevators in Canada dataset maps the list of grain elevators in Canada as provided by the Canadian Grain Commission (CGC). The elevators have been located as much as possible to an actual location rather than generalizing to the station name centroid. Additionally car spot information from CN, CP and the grain companies has been added where this has been published. This dataset attempts to provide a temporal and geographical extent of the grain elevators in Canada.

Description: This dataset consists of three simulations from the Northeastern Pacific Canadian Ocean Ecosystem Model (NEP36-CanOE) which is a configuration of the Nucleus for European Modelling of the Ocean (NEMO) V3.6. The historical simulation is an estimate of the 1986-2005 mean climate. The future simulations project the 2046-2065 mean climate for representative concentration pathways (RCP) 4.5 (moderate mitigation scenario) and 8.5 (no mitigation scenario). Each simulation is forced by a climatology of atmospheric forcing fields calculated over these 20 year periods and the winds are augmented with high frequency variability, which introduces a small amount of interannual variability. Model outputs are averaged over 3 successive years of simulation (the last 3, following an equilibration period); standard deviation among the 3 years is available upon request. For each simulation, the dataset includes the air-sea carbon dioxide flux, monthly 3D fields for potential temperature, salinity, potential density, total alkalinity, dissolved inorganic carbon, nitrate, oxygen, pH, total chlorophyll, aragonite saturation state, total primary production, and monthly maximum and minimum values for oxygen, pH, and potential temperature. The data includes 50 vertical levels at a 1/36 degree spatial resolution and a mask is provided that indicates regions where these data should be used cautiously or not at all. For a more detailed description please refer to Holdsworth et al. 2021. Methods: This study uses a multi-stage downscaling approach to dynamically downscale global climate projections at a 1/36° (1.5 − 2.25 km) resolution. We chose to use the second-generation Canadian Earth System model (CanESM2) because high-resolution downscaled projections of the atmosphere over the region of interest are available from the Canadian Regional Climate Model version 4 (CanRCM4). We used anomalies from CanESM2 with a resolution of about 1° at the open boundaries, and the regional atmospheric model, CanRCM4 (Scinocca et al., 2016) for the surface boundary conditions. CanRCM4 is an atmosphere only model with a 0.22° resolution and was used to downscale climate projections from CanESM2 over North America and its adjacent oceans. The model used is computationally expensive. This is due to the relatively high number of points in the domain (715 × 1,021 × 50) and the relatively complex biogeochemical model (19 tracers). Therefore, rather than carrying out interannual simulations for the historical and future periods, we implemented a new method that uses atmospheric climatologies with augmented winds to force the ocean. We show that augmenting the winds with hourly anomalies allows for a more realistic representation of the surface freshwater distribution than using the climatologies alone. Model Output Section 2.1 describes the ocean model that is used to estimate the historical climate and project the ocean state under future climate scenarios. The time periods are somewhat arbitrary; 1986–2005 was chosen because the Coupled Model Intercomparison Project Phase 5 (CMIP5) historical simulations end in 2005 as no community-accepted estimates of emissions were available beyond that date (Taylor et al., 2009); 2046–2065 was chosen to be far enough in the future that changes in 20 year mean fields are unambiguously due to changing GHG forcing (as opposed to model internal variability) (e.g., Christian, 2014), but near enough to be considered relevant for management purposes. While it is true that 30 years rather than 20 is the canonical value for averaging over natural variability, in practice the difference between a 20 and a 30 year mean is small (e.g., if we average successive periods of an unforced control run, the variance among 20 year means will be only slightly larger than for 30 year means). Also, there is concern that longer averaging periods are inappropriate in a non-stationary climate (Livezey et al., 2007; Arguez and Vose, 2011). We chose 20 year periods because they are adequate to give a mean annual cycle with little influence from natural variability, while minimizing aliasing of the secular trend into the means. As the midpoints of the two time periods are separated by 60 years, the contribution of natural variability to the differences between the historical and future simulations is negligible e.g., (Hawkins and Sutton, 2009; Frölicher et al., 2016). Section 2.2 describes how climatologies derived from observations were used for the initialization and open boundary conditions for the historical simulations and pseudo-climatologies were used for the future scenarios. The limited availability of observations means that the years used for these climatologies differs somewhat from the historical and future periods. Section 2.3 details the atmospheric forcing fields and the method that we developed to generate winds with realistic high-frequency variability while preserving the daily climatological means from the CanRCM4 data. Section 2.4 shows the equilibration of key modeled variables to the forcing conditions Uncertainties: The historical climatologies were evaluated using observational climatologies generated from stations with a long time series of data over the time period including CTDs, nutrient profiles, lighthouse, satellite SST and buoy data. The model is able to represent the historical conditions with an acceptable bias. The resolution of this model is insufficient to represent the narrow straits and channels of this region so the dataset includes a cautionary mask to exclude these regions. These climate projections are downscaled from a single global climate model (CanESM2/CanRCM4) because the cost of ensembles is presently prohibitive. Our experimental design uses climatological forcing for each time period so the differences between them are almost entirely due to anthropogenic forcing with little effect of natural variability. We caution that our experimental design does not permit analysis of the effects of natural climate variability. We recommend using both of the scenarios (RCP4.5 and RCP8.4) to estimate the scenario uncertainty in these projections.

30-year Average precipitation represents the average amount (mm) of precipitation received in a month across a 30 year period (1961-1991, 1971-2000, 1981-2010, 1991-2020). These values are calculated across Canada in 10x10 km cells.