The Third Generation of Homogenized Temperature dataset was prepared for use in climate trend analysis in Canada. In this version, the list of stations was revised to include observations from a larger number of surface monitoring stations, in particular those collected at Reference Climate Stations and at some Canadian Aviation Weather Services stations, which were used to extend past climate observations into recent times. The data were quality controlled. The daily minimum temperature was adjusted from 1961 to recent years, for the change in observing time in 1961 at principal stations (Vincent et al. 2009). Parallel daily data were used to detect non-climatic shifts when observations from nearby sites were merged into a single record to create long-term series (Vincent et al. 2018a). Series of annual and seasonal mean temperatures were tested for homogeneity (Wang et al. 2007; Vincent et al. 2002; Vincent et al. 1998). Daily temperatures were adjusted using a Quantile-Matching procedure to remove inhomogeneities if needed (Wang et al. 2010). Homogenized temperature datasets have been used in the analysis of climate trends (Vincent et al. 2015) and trends in climate indices in Canada (Vincent et al. 2018b). The procedures used to produce the Third Generation are described in Vincent et al. (2020).

The Adjusted Precipitation data consist of monthly, seasonal and annual totals of daily adjusted rain, snow and total precipitation (millimetres) for 464 locations in Canada. Adjusted precipitation data incorporate adjustments (derived from comparison of instruments) to the original station data to account for discontinuities from non-climatic factors, such as instrument changes or station relocation. The time periods of the data vary by location, with the oldest data available from the early 1880s at some stations to the most recent update in 2017. Observations at co-located sites were sometimes joined in order to create longer time series. Data availability over most of the Canadian Arctic is restricted to the mid-1940s to present.

Monthly, seasonal and annual trends of mean wind speed change (kilometres per hour) based on homogenized station data (AHCCD) are available. Trends are calculated using the Theil-Sen method using the station’s full period of available data. The availability of surface wind speed trends will vary by station; if more than 5 consecutive years are missing data or more than 10% of the data within the time series is missing, a trend was not calculated.

Monthly, seasonal and annual trends of total precipitation change (millimetres) based on adjusted station data (AHCCD) are available. Trends are calculated using the Theil-Sen method using the station’s full period of available data. The availability of precipitation trends will vary by station; if more than 5 consecutive years are missing data or more than 10% of the data within the time series is missing, a trend was not calculated.

The dataset includes timeseries of horizontal current speed and direction, vertical current speed, water depth, and temperature at instrument depth from Acoustic Doppler Current Profiler (ADCP) moorings. Data were collected as part of a multiyear effort lead by Fisheries and Oceans Canada (DFO) to support sustainable aquaculture regulation in the Coast of Bays, an area of the south coast of Newfoundland. This dataset is the third of a series aiming to provide an oceanographic knowledge baseline of the Coast of Bays, Newfoundland. It consists of 73 ADCP timeseries varying in length from about 26 days to 235 days collected between 2009 and 2014. Analyses from this dataset were presented during a Canadian Science Advisory Secretariat (CSAS) meeting which took place in St John’s in March 2015 (http://www.dfo-mpo.gc.ca/csas-sccs/schedule-horraire/2015/03_25-26b-eng.html) and from which a Science Advisory Report (http://www.dfo-mpo.gc.ca/csas-sccs/Publications/SAR-AS/2016/2016_039-eng.html), Proceedings (http://www.dfo-mpo.gc.ca/csas-sccs/Publications/Pro-Cr/2017/2017_043-eng.html) and several research documents were published.

The AHCCD collection includes two types of climate datasets: homogenized and adjusted. The homogenized datasets include monthly or daily surface air temperature, precipitation, and wind speed data for hundreds of climatological locations across Canada. These datasets incorporate a homogenization process, which typically uses statistical methods to detect and remove discontinuities in observed climate records caused by non-climatic influences, such as changes in measurement instruments or monitoring station relocations. In some cases, non-statistical adjustment methods are also applied to correct measured values for known biases caused by instruments or observation techniques (for example, under-catch by certain rain gauges or wind effects). Sometimes, data from multiple nearest stations are combined to extend the length of the record. Together, these methods produce consistent and more reliable Canadian homogenized station data series suitable for detecting and analyzing climate trends and variability. The adjusted datasets are available for precipitation variables only (i.e., rainfall, snowfall, and total precipitation). These datasets do not undergo a homogenization process; instead, they apply corrections for known measurement biases. AHCCD datasets are updated as new adjustment or homogenization methods are developed and/or to include the most recent observations. The homogenized datasets are also used to develop Canadian gridded homogenized datasets (https://catalogue.ec.gc.ca/geonetwork/srv/eng/catalog.search#/metadata/10ff2865-ff61-4f35-af71-3902da51f23a) to provide more continuous spatial representations of climate variables, including in areas where direct observations are unavailable. While homogenized data are the most suitable for studying long-term climate trends, both homogenization and adjustment procedures can introduce their own uncertainties or biases, and the results may change as methods are updated based on new research. It is important for users to carefully read the dataset documentation and understand how the data were processed to ensure their suitability for their research needs and other applications.

The first version of this gridded dataset (on a 10-km EASE grid) of monthly total precipitation amounts was produced using a kriging-based gridding scheme to grid the Canadian homogenized monthly precipitation dataset CanHoPmlyV1 (Wang et al. 2023; available at https://open.canada.ca/data/dataset/1dd0c28e-2266-42e2-8985-2f47659e9d02). More specifically, ordinary kriging was used to grid the 1961-1990 climate normal values and the relative anomalies, separately; the resulting gridded datasets were used to produce this gridded dataset of monthly total precipitation amounts (Wang et al. 2023, Abbasnezhadi and Wang, 2024). As detailed in Wang et al. (2023), CanHomPmlyV1 is based on the quality-controlled version 2020 of the Adjusted Daily Rainfall and Snowfall (AdjDlyRS) dataset (Wang et al. 2017, available at https://open.canada.ca/data/en/dataset/d8616c52-a812-44ad-8754-7bcc0d8de305), and on daily total precipitation data from automated gauges (including Belfort, Fisher & Porter, Nipher, Geonor, and Pluvio), with some records from neighbouring stations being joined to form long-term data series. Version 1 of ANUSPLIN surfaces of the adjusted monthly precipitation (MacDonald et al. 2021) was used to infill temporal data gaps in the 425 data series. A comprehensive semi-automatic data homogenization procedure was used to homogenize the data series. The aforementioned ANUSPLIN data and the Twentieth Century Reanalysis 20CRv3 ensemble-mean series of monthly precipitation (Slivinski et al., 2019) were used as reference in the homogeneity tests (Wang et al., 2023). The homogenized dataset CanHoPmlyV1, and its gridded version CanGridP mlyV1, which was called CanKrig mlyPv1 in Wang et al. (2023), provide more realistic estimates of precipitation trends (Wang et al. 2023). Although the latter is much better than the pre-existing CanGRD precipitation relative anomalies data, both gridded datasets contain biases due to changes in data availability over time and space (i.e., inhomogeneous sampling), which are non-negligible in the early period. Such biases are being assessed and corrected to produce a sampling bias-corrected gridded dataset CanGridP mlyV2 (upcoming). References: Abbasnezhadi, K. and X. L. Wang, 2024: Comparison of gridding methods for precipitation over Canada and assessment of station/data density effects on gridding results. Atmos.-Ocean, 62:4, 320-346, https://doi.org/10.1080/07055900.2024.2394829. Wang, X.L, Y. Feng, V. Y. S. Cheng, H. Xu, 2023: Observed precipitation trends inferred from Canada’s homogenized monthly precipitation dataset, J. Clim., 36, 7957-7971. DOI: 10.1175/JCLI-D-23-0193.1. Wang, X. L., H. Xu, B. Qian, Y. Feng, E. Mekis, 2017: The adjusted daily rainfall and snowfall data for Canada. Atmos.-Ocean, 55:3, 155-168, DOI:10.1080/07055900.2017.1342163. MacDonald, H., D. W. McKenney, X. L. Wang, J. Pedlar, P. Papadopol, K. Lawrence, M. F. Hutchinson, 2021: Spatial Models of adjusted precipitation for Canada at varying time scales. J. Appl. Meteor. And Climatol., 60, 291-304. DOI: 10.1175/JAMC-D-20-0041.1. Slivinski, L. and coauthors, 2019: Towards a more reliable historical reanalysis: Improvements for version 3 of the Twentieth Century Reanalysis system. Q. J. R. Meteor. Soc., 2876-2908, https://doi.org/10.1002/qj.3598.

644 datasets of hourly meteorological data for all of Canada from various periods (1998 to 2020). The values of the records for solar irradiance are primarily based on satellite-derived solar estimates. This dataset has been updated with the most recent changes made in March 2023. The solar values in these files are based on 0.1° x 0.1° (11 km x 11 km grid) for all of Canada. Refer to Data Resources below for additional information on the CWEEDS file format and revision history.

Version 2 of the Canadian homogenized wind speed (CanHomW V2) datasets contain homogenized monthly and daily mean wind speed data series for the period 1953-2023 for 154 long-term stations in Canada. As detailed in Wang et al. (2025), the monthly dataset (CanHomW mlyV2) was produced using a modified version of the comprehensive semi-automatic changepoint detection procedure developed by Wang et al. (2023), which includes homogeneity testing with and without using a reference series. To produce the homogenized daily wind speed dataset (CanHomW dlyV2), daily wind speed data series were tested to find the most probable day of an artificial change detected in the corresponding monthly data series whenever the exact date of change in the daily series is undocumented. The Twentieth Century Reanalysis 20CRv3 ensemble-mean series of monthly wind speeds (Slivinski et al., 2019), 49 monthly mean geostrophic wind speeds derived from surface pressure data, and up to four nearest stations’ data series were used as reference in the homogeneity tests. However, no reference was used to adjust the data series to diminish the detected inhomogeneities. This is because the density of long-term wind observing stations in Canada is too low to find a representative reference series to use for estimating reliable adjustments to homogenize the data series. A modified version of the quantile matching (QM) adjustment method with no reference (Wang et al. 2010) was used to homogenize the data series. The QM method adjusts the whole distribution of the data in one segment to match another, rather than just adjust the mean. The modifications here include an approach to prevent having unphysical negative values in the homogenized wind speed series (see Wang et al. 2025 for more details). Despite the higher uncertainty arising from applying adjustments estimated without using a reference series, compared to when applying adjustments estimated using a representative reference series (if such a reference were available), this is the best (i.e., most usable for trend assessment) Canadian homogenized wind speeds dataset that has been produced to date. For transparency and traceability, the raw (original/unhomogenized) data series along with the station joining information are included in the CanHomW datasets. Trends in the raw and homogenized data are analyzed and discussed in Wang et al. (2025). References: Wang, X.L., Y. Feng, V. Isaac, F. W. Zwiers, L. A., Vincent, and M. Hartwell, 2025: Observed Surface Wind Speed Trends Inferred from Homogenized in Situ Data and Reanalysis Datasets. Atmosphere-Ocean iFirst article, 2025, 1-17, https://doi.org/10.1080/07055900.2025.2570920. Wang, X.L, Y. Feng, V. Y. S. Cheng, H. Xu, 2023: Observed precipitation trends inferred from Canada’s homogenized monthly precipitation dataset, J. Clim., 36, 7957-7971. DOI: 10.1175/JCLI-D-23-0193.1. Wang, X. L., Chen, H., Wu, Y., Feng, Y., & Pu, Q. (2010). New Techniques for the Detection and Adjustment of Shifts in Daily Precipitation Data Series. Journal of Applied Meteorology and Climatology, 49(12), 2416–2436. https://doi.org/10.1175/2010JAMC2376.1. Slivinski, L. and coauthors, 2019: Towards a more reliable historical reanalysis: Improvements for version 3 of the Twentieth Century Reanalysis system. Q. J. R. Meteor. Soc., 2876-2908, https://doi.org/10.1002/qj.3598.