Be Careful When Reading Modeled Rain Forecasts

I see this statement (or something like it) winter after winter…“the models were showing X rain this weekend and we only got X much…they suck!” .  More than 50% of the time I think it’s misinterpretation of the model output.

Take a look at the 00z WRF-GFS output this evening for total rain ending Sunday morning.  This is the 36km resolution model:


Definitely a big soaker on the way!  Looks like Somewhere between 2.50-5.00″ at PDX right?  Wow, that could cause some flooding. Salem maybe 2.50″ as well since it’s right on the line between pink and black.  Likely over 5″ in the Southwest Washington Cascades.  Now take a look at the 12km version of the same model:


What?  It says 2.50″ at PDX, less at Hillsboro, and a bit more at Troutdale.  Salem is about the same or a little less than 2.50″.  Now take a look at the GFS model output on our broadcast software:MarkRain_GFS_ECMWF_Compare

That is extreme!  Looks like PDX could easily see 6-7″ here, Maybe 4″ out in Hillsboro and a crazy 8″ at Troutdale and Battle Ground right?  Nope.  Here is what I estimate from looking at each of these 4 models this evening:


Why so low?  Two things are going on:

  1.  In a coarser resolution model, the space between data points is larger.  That means in areas where terrain changes quickly (PDX to Larch Mountain WA anyone???), the amounts are going to be unrealistically high near the rising terrain and often shifted a bit to the west of where it really is.  I’ve seen that on our RPM model many times.  So you generally want to find the highest resolution output you can.  That also applies to going to a place like IGES and looking only at output for PDX.  It’s always too wet in a strong orographic flow (like what is coming this weekend).
  2. Specifically in the case of the ECMWF and GFS we get on our system at FOX12, WSI (our vendor for wx graphics) only gives us grids spaced 100 kilometers apart!  Yikes.  That means there’s only a point to contour from every 60 miles…really bad.  There may be one grid point in the Cascades and another over the western valley for example.  The contouring program is just averaging between those two points…pure interpolation.  Thus the inaccurate extreme amounts seen over and east of I-5.  Our RPM that goes out to 72 hours is 12 km resolution so that’s much better, like the 12 km WRF-GFS run above.

So how did I get those numbers I used on-air?  I generally look for the driest part of the northern Willamette Valley , even if it appears to be way over at Hillsboro or Forest Grove on a map.  That works well in these orographic events.  Keep in mind this applies to meteograms produced by picking a specific point as well.  Here’s the PDX 00z GFS meteogram:


But Hillsboro is far more reasonable…this is what I would expect in most of the metro area:


Of course almost nothing I’ve written applies in showery weather patterns or precipitation coming from a different direction, but our big heavy rain/flood events are almost always under westerly flow.

By the way, yes, we’re going to get a soaking this weekend, but I don’t think 6″ or more in the valleys will be happening.  The Cascades will get a real pounding with 5-10″ possible.  Luckily rivers are low and the ground is dry…the first 3-5″ should soak in nicely.

Chief Meteorologist Mark Nelsen

33 Responses to Be Careful When Reading Modeled Rain Forecasts

  1. High Desert Mat (Redmond Or.) says:

    It was down to 26.3 this morning at my house. Average for Redmond on this date is 59/30. Looks to be another chilly morning tomorrow as I’m down to 44.6 already. Winter here we come.

    A special weather statement was issued yesterday for central Oregon stating that wet and windy conditions will occur Friday into Saturday with the chilliest air of the year coming Sunday and beyond.

    • Boydo3 N. Albany says:

      Still picking tomatoes here in the valley. But soon they will be exploding with water if the forecast holds..

  2. Sapo says:

    Interesting, ya looks like it’s going to be really rainy this weekend! Wonder what that means for trick or treaters…more candy for the people handing it out?

  3. LDT says:

    I still don’t know how I feel about their long range forecasts but accuweather has been consistently forecasting a very warm and somewhat dry period from Nov. 9-21st. I was curious if anyone had any insight on this. I felt relieved to see some snow on the mountains this morning, Mt. Hood’s glaciers/snowfields were looking anemic.

  4. () says:

    So in other words we should get a quarter of an inch out of this storm.

  5. Brian Schmit says:

    10/26/2015 Oregon (All) Temperature Summary

    High:70 at AGNESS2( 247 ft) & CW3932 Central P(1290 ft)
    Low: 54 at BROOKS(187 ft) & WYETH(102 ft) & EW1135 Yachats1(32 ft)

    High:34 at Timberline Lodge(5880 ft) & HOWARD Mt Howard(8150 ft) & Mt Hood Meadows(7300 ft)
    Low: 16 at Burns Municipal (4144 ft )

    Largest Diurnal Change: 47 degrees
    Lorella (65/18) (4160 ft)

    Heaviest Rainfall:
    1.10″ at NORTH FORK(3060ft)
    1.04″ at WB6RDV-13 Breite(2404ft)
    1.03″ at SOUTH FORK(2257ft)
    1.02″ at K7ZQU-7 Santiam(4790ft)
    1.00″ at SADDLE MOUNTAIN(3110ft)
    0.64″ at Brookings Airpor(459ft)
    0.59″ at North Bend Munic(16ft)

  6. Gina says:

    Yikes not really looking foreword to 8 inches of rain lol My rain gage so far for Oct is 1.71 . 10 miles north of battle ground, near larger lake area,

  7. High Desert Mat (Redmond Or.) says:

    Mark, why do you have Portland at 64 today and Salem at 62? You should know by now the weather in Salem is always better. Lets stay in the real world shall we?

    • Josh "the snowman" Gladstone, OR says:

      Too bad all that beautiful Salem weather doesn’t make the women look better there though.

    • WEATHERDAN says:

      It must be close to Halloween because the trolls have come out to annoy.

    • Josh "the snowman" from Gladstone, OR says:

      I just want some good mountain snowpack to build. Could care less about the valley right now. We need these progressive patterns in the next 3 months. Too bad it most likely isn’t going to happen. Makes me sad really.

    • Boring Oregon says:

      Yeah. Kate Brown is pretty ugly.

  8. Thanks for the clear explanation, Mark. I often see a local website in my area (Vancouver weatherpage) giving estimates of precipitation that seem quite extreme. For instance, the past few days they’ve been forecasting 55-80 mm. on the 30th here. (that’s about 2.25-3.25 inches for those of you who don’t know metric)I usually reduce the amounts they say to about a third, and I frequently get much closer to the actual amount.

  9. Longview 400 ft says:

    Thanks Mark for the update this weekend.

    I have been warning my students of the coming down pours and to plan ahead. So far, the candy seems to embolden them to persevere through whatever comes their way, according to their comments.

    • Mark bergal says:

      Looks like a cool air mass skirts us next week but the outlook is also dry. My guess, an east wind event.

  10. W7ENK says:

    Well, this is interesting.

    2 weeks out, so take that for what it’s worth…
    …basically nothing, but it’s still pretty! 😀

  11. W7ENK says:

    So, can people still complain when you forecast 2 inches of rain for the Portland area, but we end up only getting a quarter inch or so?


  12. runrain says:

    Some kind of sunrise this morning. Puts it all in perspective.

  13. Joshua Downtown PDX says:

    Love the weather porn, Mark. Open the floodgates!

  14. timbers15 says:

    As always, you presented it in a modern fashion for weather geeks.

  15. Mark bergal says:

    Quite interesting

    • DEL X V says:

      Well done Mark! You beat the other TV Weather Persons by closely studying the model output, and letting us know your findings.

  16. WhiteEagle - Garden Home/SW Portland says:

    Very good points, Mark!! (First?)

%d bloggers like this: