Overview
This document focuses on temperature data from Canadian gridded anomalies (CANGRD) dataset. The dataset provides historical gridded temperature and precipitation anomalies, interpolated from adjusted and/or homogenized climate station data at a 50 km resolution across Canada. The anomalies are computed as the departure from a mean reference period (1961-1990) and are used to produce the Climate Trends and Variations Bulletin (CTVB).
Provider's contact information
Environment and Climate Change Canada (Contact the Climate Services Support Desk)
Electronic Mail Address at CRD: f.ccds.info-info.dscc.f@ec.gc.ca
Licensing
Open Government Licence - Canada.
The end-user licence for Environment and Climate Change Canada's data servers specifies the conditions of use of this data.
Variable name and units:
Maximum daily temperature anomaly in degree Celsius (°C);
Minimum daily temperature anomaly in degree Celsius (°C);
Mean daily temperature anomaly in degree Celsius (°C)
Spatial coverage and resolution:
Canada, 50 km spatial resolution, polar stereographic grid.
Temporal coverage and resolution:
Temperature anomalies: 1948 - 2020
Data is available at monthly, seasonal, and annual time steps.
Information about observations (number, homogeneity)
CANGRD is based on AHCCD for historical climate observations from 1948 for all of Canada until 2020 for temperature (based on the 3rd generation temperature AHCCD). AHCCD has a small number of stations in North Canada, and the distance between stations is large.
Methodology
The monthly, seasonal, annual mean daily maximum temperature, and annual mean daily minimum temperature anomalies are computed at each station by subtracting the relevant baseline average (defined as average over 1961-1990 reference period) from the relevant monthly, seasonal, and annual values. The anomalies are interpolated to the evenly spaced (50 km) grid using the Gandin’s Optimal Interpolation. The grid box values of mean temperature departures are the average of those for daily minimum and maximum.
Information about the technical and scientific quality
A positive aspect of this dataset is the use of adjusted and/or homogenized data from the AHCCD (homogenized temperature data). This station-based datasets have undergone rigorous quality control, and have been adjusted for identified inhomogeneities caused by station relocation, changes in instrumentation and in observing practices.
The dataset uses classical methods for the interpolation. In general, results from the interpolation of anomalies are better than those from actual values because anomalies vary less in space. Another positive aspect of interpolating anomalies instead of actual values is related to temperature inversions in Yukon. Gandin Optimal Interpolation is a classical interpolation method that is also used in other reanalyses. This method does not take into consideration the elevation, which is indicated when interpolating temperature. Generally, interpolation of absolute temperature values in Yukon region is challenging because of the presence of semi-permanent temperature inversions in that region. Inversions make the temperature to be colder in valleys than at elevation, which is opposite to normal situations; consequently, the temperature lapse rate cannot be used uniformly over the domain. However, interpolation of anomalies is less affected, and it offers a good option for such complex regions. The correlation coefficients for anomalies in maximum temperature at the nearest grid and station are generally larger than 0.85 in the northern part of the country. However, it is expected that the errors increase with the distance from the station.
Data is available in .grd format on the CRD webpage. This format is not a standard format. Less advanced users should use the CCCS webpage to download GeoTIFF format. CCCS webpage offers the possibility to download the data in NetCDF format as well, but this is also not in the standard form (uses bands instead of time coordinate; this should be corrected by the end of 2021).
The dataset is accompanied by a technical report.
Limitations and strengths for application in North Canada
The data was constructed to describe large-scale climate change over Canada and for national-scale assessments. Interpolation errors are expected to rise with the increase in the distance between stations. Because there is a small number of stations available in North Canada (most of them in coastal and valley locations), interpolation errors in inland regions and in Yukon high elevation regions can be significant. Consequently, the result should be interpreted as a mean change over a large region (e.g., the mean change over the North or a territory), and it is not recommended to be used for local applications.
Because it provides anomalies, not actual temperature values, it cannot be used to compute other climate indices.
Some users can find the data in .grd difficult to use because of its custom projection and format.
References to documents describing the methodology or/and the dataset
Environment and Climate Change Canada (2018). Canadian Gridded Temperature Anomalies CANGRD. Accessed August 16 2018.
Vincent, L.A., M.M. Hartwell, and X.L. Wang, 2020. A third generation of homogenized temperature for trend analysis and monitoring changes in Canada’s climate, Atmosphere-Ocean., 58:3, 173-191, doi:10.1080/07055900.2020.1765728.
Link to download the data and format of data
https://climate-change.canada.ca/climate-data/#/historical-gridded-data (GeoTIFF and NetCDF on CCCS/ECCC webpage)
GeoTIFF data available on the MSC Datamart
http://crd-data-donnees-rdc.ec.gc.ca/CDAS/products/EC_data/CANGRD/ (gridded data .grd on CRD/ECCC webpage)
Publications including dataset evaluation or comparison with other data in northern Canada
Rapaić, M., R. Brown, M. Markovic, and D. Chaumont, 2015: An Evaluation of Temperature and Precipitation Surface-Based and Reanalysis Datasets for the Canadian Arctic 1950–2010. Atmos. Ocean, 53(3), 283-303, 10.1080/07055900.2015.1045825. (The paper compares CANGRD to several other gridded datasets and other coarse-resolution reanalysis)