Get invited to our slack community and get access to opportunities and data science insights

Rendering GRIB2 Data, 54 Hour Weather Forecast

Temperature forecast, -20°C=green, 0°C=yellow, +20°C=red

Maximum hourly wind gust, logarithmic scale, 0m/s=blue, 40m/s=yellow

An example of timeseries data rendered to a video, using WhereOS. The data is provided by Finnish Meteorological Institute: “HIRLAM (High Resolution Limited Area Model) is an operational synoptic and mesoscale weather prediction model.” See article: FMI Hirlam Weather Model on Amazon AWS S3

To create this visualization, WhereOS is using a GRIB2 driver to read the data. To manipulate the data and create the video with WhereOS you just need a few lines of SQL:

select rendervideo(frame) from (
  select time, collectandrendergeometry(poly, interpolatehsb("lime","red",linearscale(array(253.15, 293.15),array(0,1),val)), 1.0, "blue", 0.0, 0.0, 58.205317, 16.0, 4, 800,500, concat("Forecast for ",from_unixtime(time))) as frame from (
    select geokey_to_rectangle(geokey) as poly, avg(value) as val, time from (
      select geokeylatlon(lat,lon,11) as geokey, value, time from (
        select latlontopos(lat,lon,5) as pos, lat,lon,value,unix_timestamp(basedate)+3600*time as time from weather_data_orc
      ) where pos[0] >= 14 and pos[0] <= 20 and pos[1] >= 7 and pos[1] <= 11
    ) group by geokey, time
  ) group by time order by time

One idea would be to combine this with routing, i.e. a model that shows you detailed forecast for your route: when you can encounter snowfall, icy conditions etc, based on your destination and time of departure.

If you have any other ideas for timeseries datasets, or you are weather data enthusiast and want to do something cool about it, write to me

The data is licenced by the Creative Commons Attribution 4.0 International license (CC BY 4.0)

Related Post

Leave a Comment