Add Average (or Median) Forecast Based on (ECMWF, GFS, NEMS, ICON-EU)



  • Hello,
    usually, when I check the weather forecast for my location, I use a compare feature to look at all 4 available models (ECMWF, GFS, NEMS, ICON-EU) at once to determine how close the models are to each other and therefore what is a probability the weather will be as forecasted (I do not calculate anything, I just look if models predict similar forecast values).

    It would be nice to have an option to view only one combined forecast based on these four models where the values (temperature, wind speed, clouds and so on...) would be an average of the four individual models or a median, as sometimes the forecasts can differ especially for the wind values.

    Best regards,
    ZvireZet.



  • @zvirezet
    0_1540796423362_cf4fd03f-4e9d-4bb6-801d-50c59b493a3c-εικόνα.png

    You got it !

    0_1540796541236_cebe260c-f277-4453-bc3a-1e15439d1e18-εικόνα.png



  • Hello @Gkikas-LGPZ ,

    that is exactly what I do, as I mentioned in my post above. But what I suggested is that it would be nice to have an option to see only one chart with AVERAGE (or MEDIAN) values synthetized from the four separate models.



  • @zvirezet
    And what will be your forecast if 3 models forecast no rain,
    and one model forecasts 20 mm of rain?
    Will you expect 20:4=5 mm of rain?

    Also, high resolution models, forecast wind and temperature better
    (especialy over rough terrain).
    Computing the average, is not a good practice in such a case.



  • @zvirezet

    Very bad idea.
    You cannot get average from apples and oranges.



  • @gkikas-lgpz said in Add Average (or Median) Forecast Based on (ECMWF, GFS, NEMS, ICON-EU):

    @zvirezet
    And what will be your forecast if 3 models forecast no rain,
    and one model forecasts 20 mm of rain?
    Will you expect 20:4=5 mm of rain?

    Also, high resolution models, forecast wind and temperature better
    (especialy over rough terrain).
    Computing the average, is not a good practice in such a case.

    @gkikas-lgpz That is why I suggested MEDIAN as well. (https://en.wikipedia.org/wiki/Median)



  • @siim said in Add Average (or Median) Forecast Based on (ECMWF, GFS, NEMS, ICON-EU):

    @zvirezet

    Very bad idea.
    You cannot get average from apples and oranges.

    @siim No one suggests mixing apples and oranges. If every model contains a value assigned to a specific weather term (temperature, wind speed, gust,...) for an exact hour (and that is exactly what your four models show you in the comparison in the second post of this topic), then you can easily compute average or median value.

    Where are the apples and oranges?

    Moreover, this should/could be only optional.



  • @zvirezet

    If it is so easy then tell me what is the average/median wind direction/speed in this forecast:
    0_1540813087814_front.PNG

    Is this median value useful compared to other models?



  • @siim I believe that your call to calculate the average and median from the selected data is purely rhetorical as you can calculate it by yourself or Excel can do it for you.

    By the way, if you look at the wind forecast for the hour you selected in the image, it will have the same meaning both as averaged (or median) value or in four separate forecasts - it will tell you the wind will blow... from some direction. No matter if you look at four separate charts or at only one averaged (or "medianed") chart. But you have to be familiar with it. Maybe there could be an indicatior that for the particular hour the average(or median) has a large deviation.

    On the other hand, if you look at 6 a.m. that day in the image, you can see the one averaged/"medianed" chart could fit four separate forecasts very well. But again, as I mentined, you have to be familiar with it and you should know what it actualy means.



  • Ofcourse it is very easy to calculate median/average when values are very similar.

    But these averages/medians are totally useless in situation where all models disagree.



  • Hello @siim, you are right, they are useless, as well as charts where all models disagree. As I mentioned - you have to be aware what you are looking at.


Log in to reply
 

Windyty, S.E. - all rights reserved. Powered by excellent NodeBB
NodeBB & contributors, OSM & contributors, HERE maps

Looks like your connection to Windy Community was lost, please wait while we try to reconnect.