Jump to content
Thunder?
Local
Radar
Hot?
  • entries
    30
  • comments
    25
  • views
    43,696

Weather Models Experiment


Zenarcher

1,869 views

[b]Introduction,[/b]

Over the course of September and October this year I carried out a experiment to see what weather models performed the best at different time ranges from 24 hours to 144 hours.

[b]How it works,[/b]

Firstly I would let all the models do their 12z runs so none over them had a 12 hour advantage over another. Once they had done their 12z run I would pick a location and each model predicted its thought on it. The models then get awarded points on how close they are.

[b]The results for forecasting 24 hours out. It was very tight most of them got the predictions correct but GME was more consistent gaining the highest average points.[/b]


1. GME 127.
2. ENS 119.
3. GFS 110.
4. ECM 107.
5. NOGAPS 107.
6. UKMO 106.
7. NAE 99.
8. GEM 94.
9. CMC 91.
10. JMA 80.
11. RHMC 78.

[b]The results for forecasting 48 hours out. The ECM always performed well and is 20 points clear of the GFS ENS.[/b]


1. ECM 98.
2. ENS 78.
3. GFS 72.
4. CMC 72.
5. NOGAPS 66.
5. JMA 61.
7. UKMO 58.
8. NAE 58.
9. GME 56.
10. GEM 56.
11. RHMC 34.

[b]The results for forecasting 72 hours out. This was another tight one with 6 weather models doing very well but overall we have a tie between ECM and GEM.[/b]


1. ECM 46.
2. GEM 46.
3. CMC 44.
4. NOGAPS 44.
5. UKMO 44.
6. GME 44.
7. GFS 40.
8. JMA 30.
9. ENS 28.
10. RHMC 26.

[b]The results for forecasting 96 hours out. You may have noticed there's only 8 now because RHMC and GME don't go over 96 hours. So overall JMA done very well here despite not doing to well in the previous results.[/b]


1. JMA 52.
2. UKMO 42.
3. GFS 34.
4. NOGAPS 34.
5. ECM 34.
6. GEM 26.
7. CMC 24.
8. ENS 24.

[b]The results for forecasting 120 hours out. I was surprised to see NOGAPS do well here but it always performed well.[/b]


1. NOGAPS 64.
2. GEM 45.
3. ECM 44.
4. ENS 42.
5. CMC 38.
6. JMA 36.
7. GFS 32.
8. UKMO 16.

[b]And finally for forecasting 144 hours out. The GFS done well here and followed by the ECM.[/b]


1. GFS 51.
2. ECM 44.
3. ENS 37.
4. UKMO 34.
5. NOGAPS 30.
6. CMC 20.
7. JMA 20.
8. GEM 19.

[b]So what weather models are the best? I have added up the points below and the ECM has won overall gaining the most points followed by a tie between GFS and NOGAPS. I think it may surprise people on how poorly the UKMO performed.[/b]


1. ECM 313.
2. GFS 293.
3. NOGAPS 293.
4. GFS ENS 284.
5. CMC 251.
6. UKMO 248.
7. GEM 242.
8. JMA 231.

I hope you found this interesting and I will carry out the same experiment again at the start of 2012.

3 Comments


Recommended Comments

Thanks guys to see the latest on the next experiment check here [url="http://forum.netweather.tv/blog/227/entry-4312-upcoming-weather-model-experiment/"]http://forum.netweather.tv/blog/227/entry-4312-upcoming-weather-model-experiment/[/url]
Link to comment
×
×
  • Create New...