Unexpected extreme weather events have become the new normal, and businesses are seeking opportunities to better predict and assess external risk.
This week we added the HIRLAM operational weather prediction model to Planet OS Datahub. HIRLAM, the High Resolution Limited Area Model is used in Scandinavia, Baltics and some other EU countries for weather forecasting.
Compared to global models like GFS, HIRLAM is able to better assess the local details — the landscape, coastline, etc. — that results in better overall forecasts. The HIRLAM program was founded by a consortium of Nordic countries in 1985, and the model is famous for representing wintertime extremes better than other weather models.
HIRLAM provides short range, one to three days forecasts, so I recommend using both, the local and the global GFS model. Should you need data only for Scandinavia and the Baltic States you might want to add the MET Norway HARMONIE Forecast that provides even higher resolution data than HIRLAM.
You can access all three datasets using Planet OS Datahub API. I have added the direct links to the datasets below:
- HARMONIE http://data.planetos.com/datasets/metno_harmonie_metcoop
- GFS Forecast
I’ve published a Jupyter Notebook demo in GitHub comparing the three different forecast models. First, I highlight the differences in their spatial coverages by plotting each model in an orthographic project. I then demonstrate their different spatial resolutions by plotting the forecasted air temperature as a discrete grid, and finally plot the same value as a time series to highlight their temporal coverage and deviation.