|
|
The HIRLAM model at The Norwegian Meteorological Institute (DNMI)
The model is run for a limited area on a rotated spherical grid in two versions, one with horizontal resolution 0.5 degrees covering Europe, the North Atlantic, the Polar basin and the northern areas of Russia and America. The other version has horizontal resolution 0.1 degrees and covers the north western part of Europe and the Norwegian Sea. The vertical resolution is 31 layers in a hybrid coordinate system for both versions.
The 0.5 degree version is run four times a day from initial state 00, 06, 12 and 18 UTC, and the high resolution version is run from 00 UTC, both doing a 48 hours forecast. Boundary values are acquired from the ECMWF global forecasts on pressure levels. At the moment, we run the 2.4 version of HIRLAM, modified with a semi-Lagrangian advection. More details on the HIRLAM model at DNMI can be found here.
Observations
In addition to traditional observations from temps, pilots, synops, ships and drifting buoys, aireps and remotely sensed sea surface wind are utilized operationally in DNMI's numerical model.
Analysis
Analysis is the method of distributing the information in the observations to the model grid. In the operational analysis, a modified version of the succesive correction method is used to correct the first guess, a 6 hours forecast. The method was suggested by Bratseth in 1986, and implemented and made operational at DNMI in 1986.
Forecast
The weather elements show a large variability in space, in particular in mountainous areas as Norway. In order to improve the local description of weather elements in the forecasts, the development of high resolution models has high priority at DNMI. An example on a numerical forecast of 2m temperature from HIRLAM with 0.1 degrees horizontal resolution is shown. Red lines are above freezing level temperature, blue lines are below freezing level and black line is the freezing level line.
Verification
Numerical forecasts from DNMI's models are continuously
verified against observations, and results are analysed and distributed
internally on a seasonal basis. Verification serves as background for subjective
interpretation of numerical forecasts in the weather service, as a contribution
to quality assurance of
numerical products and as feedback for model improvement.