The Usefulness of Spaghetti…Charts

The past few days we’ve seen an unusally high level of uncertainty in the long range forecast.  Models are having major difficulties deciding whether strong upper level ridging will develop over us Sunday/Monday and whether a deep and cold trough will drop in over us beyond that next week.  This is the type of forecast nightmare we don’t want to see in the cold season.  This exact same issue would have us worrying about a major arctic blast vs. sunshine and 50 degrees.   There would be lots of unhappiness as models swing wildly from cold to warm over us with chances for snow too. 

However, it’s early October next week so it’s mainly a matter of whether we’ll be sunny and 85 or partly cloudy/showery and 60 for highs.  Not such a big deal this time of year.

One tool that has been very helpful is the “spaghetti chart”.   It displays one parameter from models out in time on a chart.  But instead of just one model, you see the many different versions (ensemble members).  When all those runs are similar, you have a high level of confidence. When they are all over the chart, like that pasta, reliability is low.  This 2nd situation continues this evening. 

Let’s take the 850mb temperature (5,000′ temp) over Portland (in celsius).  This is the 12z ECMWF model run:

Several obvious issues…look how quickly the solutions diverge Sunday and beyond.  By Tuesday, the 2nd of October, the range is anywhere from -1 to +24!  The “operational” (blue) or main run of this model is +10 at that time and headed downhill quickly.  Then look how the operational run with it’s deep cold trough bottoms out as the coldest of all 51 ensemble members next Thursday, the 4th.  That tells me it’s extremely unlikely that run will verify.  This is where these charts come in useful.  It’s generally safer to stick closer to the red line, that’s the average of all the runs.  That is why I didn’t go crazy warm/hot on the 7 Day forecast Monday and Tuesday.  Also didn’t go with such chilly temps next Wednesday.  This ensemble forecasting is where we are headed the next 10 years when we think of forecast improvement.

How about the fresh 00z GFS:

Some more clues.  About half the members are showing a flatter ridge and quicker cooling late Monday-Wednesday.  Even the operational model says 80 might be a little high both Sunday and Monday.  The trough must drop down out of Canada more quickly on those members.  Then, like the ECMWF, the operational GFS is the coldest or near-coldest run next Thursday and beyond.  Luckily, just about all of that is beyond our 7 Day forecast duties.

Chief Meteorologist Mark Nelsen

17 Responses to The Usefulness of Spaghetti…Charts

  1. David B. says:

    “Several obvious issues…look how quickly the solutions diverge Sunday and beyond. By Tuesday, the 2nd of October, the range is anywhere from -1 to +24!”

    Seriously, this underscores a problem with how weather forecasts are communicated. Unless you pay attention to the forecast discussion (which, face it, most people don’t), all you have is a set of forecast temperatures and conditions, with no information as to how confident the forecasters are in that forecast. Nothing to distinguish a forecast based on all ensembles tightly agreeing with each other from the case we have here.

    If there was some sort of accompanying value to show forecast confidence, even say a small number from 0 to 9 or 0 to 10, then people would know to put more faith in a forecast with a confidence of 9 than one with a confidence of say 2, and there would be less griping about how weather forecasters get paid to make incorrect predictions.

  2. This is just wierd…

  3. alohabb says:

    Geesh Mr Data….. They have medication you can take you know.

  4. alohabb says:

    Nice post. I went to the Fox 12 iphone app and the 7 day forecast showed a high next Wednesday of 49!!!

  5. chiefWright (Marquam) says:

    I’m worried the week or two of cool wet weather in mid-late Sept that signals the start of a true indian summer (my all-time favorite pacNW season) might not happen this year. Boo.

    By the way, be careful about that red average line, statistically speaking. If it’s living in an area where the spaghetti strands are densely packed (and the ECMWF 12Z is a good example), then it’s probably a good predictor.

    But be careful if there’s a couple of densely packed regions with the red line right between them!

    That means there’s two very distinct probable outcomes, and the probability of the average is quite low. It ends up being a coin toss for those other outcomes, but you can more confidently say that it won’t be the average.

    This isn’t all that rare; like when there’s a strong system on it’s way, just the timing of when it hits is very uncertain. Or when we’re teetering on the edge of offshore/onshore flow with a bucketload of cold air pooled up to the east.

    • W7ENK says:

      What do you mean “might not happen”? Many outlying areas (esp. Clark County and the far Western reaches of the Valley) had their first frost a couple weeks ago… we’re officially in it now!

  6. As much fun as a “neverending summer” for the first half of October would be, it would be equally nice to turn brisk for a few days to start setting the leaf colors – then warm back up in the middle of the month. Let’s say a dry trough similar to the 2nd week of September, only this time highs at DLS would be in the low to mid 60s and lows in the 30s for about 5 days. Then back to 40s for lows and 75-80 for highs. Wonder if that would be enough to trigger some vibrant color by October 15?

    By the way I wrote a Facebook note about October 2003. It had the dubious honor of being the 4th warmest October in PDX recorded history, also the warmest October of all the past 20 years. Yet for the first half of the month, there was absolutely NO WAY you could tell it would end up earning such a title:

  7. Mountain Man says:

    You need to talk to someone if you are depressed. Find a friend or some help with your issues. Granted, the end of the world as we know it will probably come sooner or later, but this is a weather blog where people talk about weather and climate or share something comical from time to time. If you’re going to post, please, for the sake of us all make it about something that relates to this blog and enjoy your passion for weather. I understand that things in life can bring you down, but please find a place that fits your issues and visit this blog when you have something positive to share or you wish to engage in the positive things others will share with you here.

  8. Mr Data says:

    I am very thankful however that the smoke is either gone or being pushed away from us.

    The sky is still somewhat murky from farmers burning away like crazy but it’s not as sick as several days ago.

  9. bgb41 says:

    9/26/2012 Oregon (All) Temperature Summary

    High:91 at AGNESS2( 247 ft)
    Low: 63 at RED MOUND(1753 ft)

    High:53 at NERRS MET SITE A(10 ft)
    Low: 22 at Horse Ridge (US (4160 ft )

    Largest Diurnal Change: 56 degrees
    Horse Ridge (US (78/22 ) (4160 ft )

%d bloggers like this: