66km Resolution Numerical Data of the Global Deterministic Prediction System (GDPS) model - GRIB2 format

The fields in the Global Deterministic Forecast System (GDPS) GRIB2 dataset are made available on a 601x301 latitude-longitude grid at a resolution of .6x.6 degrees, which corresponds to about 66 km resolution.

GDPS on 66km Lat-Lon grid

66 km global grid

Grid specifications

Table lists the values of various parameters of the 66km resolution lat-lon grid.

ni 601
nj 301
resolution 0.6°
coordinate of first grid point 90° S 180° W

There is also a formatted html file containing geographical coordinates for each grid point.

The fields in the high resolution polar-stereographic 30 km global dataset are made available on a 247x200 Polar-Stereographic grid covering North America and adjacent waters with a 30 km resolution at 60°N.

GDPS on 30 km Polar Stereographic grid

30 km global grid

Grid specifications

Table lists the values of various parameters of the high resolution polar-stereographic grid.

ni 247
nj 200
resolution at 60° N 30 km
coordinate of first grid point 32.549° N
134.614° W
(i,j) coordinate of North Pole (88.0, 200.0)
grid orientation
(with respect to j axis)
-111.0°

There is also a formatted html file (ASCII with an easy-to-remove html wrapper) containing geographical coordinates for each grid point.

Download

The data is available using the HTTP protocol and resides in a directory that is plainly accessible to a web browser. Visiting that directory with an interactive browser will yield a raw listing of links, each link being a downloadable GRIB2 file. In practice, we recommend writing your own script to automate the downloading of the desired data (using wget or equivalent). If you are unsure of how to proceed, you might like to take a look at our brief wget usage guide.

The data can be accessed at the following URLs:
http://dd.weather.gc.ca/model_gem_global/66km/grib2/lat_lon/HH/hhh/

or
http://dd.weather.gc.ca/model_gem_global/66km/grib2/polar_stereographic/HH/hhh/

where:

  • HH: model run start, in UTC [00,12]
  • hhh: forecast hour [000,006,012,...,144]

File name nomenclature

The files have the following nomenclature:
CMC_glb_Variable_LevelType_Level_projection_YYYYMMDDHH_Phhh.grib2

where:

  • CMC: constant string indicating that the data is from the Canadian Meteorological Centre
  • glb: constant string indicating that the data is from the GEM-GDPS model
  • Variable: Variable type included in this file. To consult a complete list, refer to the variables section.
  • LevelType: Level type. To consult a complete list, refer to the variables section.
  • Level: Level value. To consult a complete list, refer to the variables section.
  • Projection: projection used for the data. Can take the values [latlon, ps]
  • YYYYMMDD: Year, month and day of the beginning of the forecast.
  • HH: UTC run time [00,12]
  • Phhh: P is a constant character. hhh is the forecast hour [000,006,012,...,144]
  • grib2: constant string indicating the GRIB2 format is used

Example of file name:
CMC_glb_TMP_ISBL_925_latlon.6x.6_2010090800_P042.grib2

This file originates from the Canadian Meteorological Center (CMC) and contains the data of the GEM-GDPS. The data in the file start on September 8th 2010 at 00Z (2010090800). It contains the temperature component (TMP) at the isobaric level 925 mb (ISBL_0925) for the forecast hour 42 (P042) in GRIB2 format (.grib2).

Levels

Vertical coverage of three-dimensional fields is provided by up to 28 isobaric levels.

Isobaric levels (hPa): 1015, 1000, 985, 970, 950, 925, 900, 850, 800, 750, 700, 650, 600, 550, 500, 450, 400, 350, 300, 275, 250, 225, 200, 175, 150, 100, 50.

Some fields receive limited three-dimensional coverage of 4 levels in the vertical.

Isobaric levels (reduced coverage, hPa): 850, 700, 500, 250.

Additional levels that may be applicable to a given parameter are:

  • surface
  • fixed height above ground
  • fixed height below ground
  • thickness between two isobaric levels
  • nominal top of the atmosphere
  • entire atmospheric column

Verification - a note regarding the growth of errors in NWP models

Note that there is a substantial difference in accuracy between a one-day forecast and a ten-day forecast. NWP model errors inevitably increase with forecast time. This increase is inherent in the numerical weather prediction process.

Root mean square error for the temperature at 850 hPa for hours 24 to 240

The above graph shows the progression of Root Mean Square Errors (RMSE) of the CMC GDPS model 850 hPa temperature forecasts, from 24 to 240 hours lead-time, averaged over the year 2010 and calculated against the North American radiosonde observation network. Note how the errors increase nearly linearly with forecast time. This is typical of most model forecast fields.

RMSE is a measure of the average error of a field, without indicating the direction of the error (known as the error bias). It is weighted according to the square of the error and as such is sensitive to large errors.

The 850 hPa height (about 5 000 feet or 1500 m above sea level) is generally representative of the free atmosphere above the turbulent boundary layer close to the earth's surface.

These considerations do not call into question the utility of the forecasts or the progress in the field of NWP. In general, there has been a gain of about a day in forecast skill for each decade since numerical modeling began in the mid-20th century.

Data in GRIB2 format

NOTES:

  1. u (UGRD) and v (VGRD) components of the wind vector are to be resolved relative to the defined grid, in the direction of increasing i and j coordinates. Please refer to Code Table 7 of the GRIB standard for further details.
  2. Radiation fluxes parameters NLWRS_SFC, DSWRF_NTAT, and DLWRF_NTAT are encoded as accumulated values, as per value 4 of the Time Range Indicator (Code Table 5 of the GRIB format standard). This changes the unit from Watts per square metre to Joules per square meter.
  3. Specific Humidity (SPFH) is the model's native moisture variable. It should be used in all NWP applications. Dew point depression (DEPR) is a diagnostic parameter and is not meant for use in high-precision applications.