Jump to content

forecaster

Members
  • Content count

    245
  • Joined

  • Last visited

Community Reputation

266 Excellent

Profile Information

  • Gender
    Male
  • Location
    Wellington, NZ, about 120m ASL.

Contact Methods

Recent Profile Visitors

5,099 profile views
  1. forecaster

    Cyclone Gita

    There is a live thread here: https://www.stuff.co.nz/national/101542667/live-cyclone-gita-nears-new-zealand Some locations in the upper South Island received 50mm of rain in an hour yesterday. In just 18 hours, Kaikoura recorded one third of it's typical annual rainfall! Westport had easterly gales, gusting 60 knots. Westport is a pretty wild, windy place but is not accustomed to easterly gales - they tend to be very damaging. There's also lots of snow to 1000m or less in the lower South Island.
  2. forecaster

    Cyclone Gita

    Cyclone Gita last week moved directly over Tonga's main island, incredibly causing no fatalities. It's still classified as a TC at 33S, though will transition soon. It crosses New Zealand on Tuesday, bringing extreme weather. As is often the case with these former cyclones, it brings up cold air the time of year. Below is some GFS guidance showing the incredible 850hPa temperature contrasts: 23C in the warm core of Gita, and 0C just 500 miles south. A 23C 850hPa temperature gradient from the north to the south of the South Island!
  3. forecaster

    Model output discussion pm 13/01/2016

    A quick look at the ECMWF model for Monday evening suggests that there could be a period of snow to lowish levels ahead of the approaching occluded front. It may turn to rain behind it though for low lying areas. The contours are a forecast snow level which dynamically changes based on how heavy the model precipitation is (heavier snow can penetrate closer to the ground). The front moves through after dark. A separate feature moves through Scotland on the weekend bringing a convincing signal for snow to low levels in the west. Imagery ECMWF via MetraWeather.
  4. forecaster

    Model output discussions pm 31/12/15

    The temperature news from the latest EC32: Below average for 4th - 10th, then above average thereafter. It also makes all of northern Europe, eastern Europe and half of Russia above average as well.
  5. Not a personal criticism but I find this, along with almost all discussions of AAM to be a bit "hand wavey". I don't understand many of the mechanisms behind AAM and how it affects or is affected by the weather and no explanations I've seen quite cut it. I would ask: 1) WHY does it aid to push the PV to lower latitudes? What's the mechanism? 2) HOW does it split the jet? Why would an increase in AAM split the jet?
  6. Why does an increase in AAM help disrupt the polar vortex?
  7. Using the mean along with well calibrated probabilities is quite sound in many situations. The mean smooths things out, the probabilities tell you about what's been removed by the smoothing. If you care about cold spells, you can do things like showing the mean and % of members showing 850hPa T < xC. Even 5% will then provide you with a hint. Cascading probabilities of multiple thresholds and parameters then give you a pretty good view of the ensemble. My only wariness of stamp maps themselves is that you are overloading the human with information and they may sometimes filter it out subconsciously. Clustering is probably superior IMO, unless you are looking at a very specific feature, in which case stamp maps are the king (I remember seeing ECMWF stamps for a famous low in 1999 or something like that, it was very interesting). Bad times to use the mean are if you have very distinct possible outcomes, with the mean simply provided a physically meaningless and non-representative "MOR solution". In this case it's very dangerous, and misleading. But it's also rare.
  8. This is just hyperbole. The meteorological (both operational and research) consensus is strongly in favour of ensembles. They are not crystal balls but they are essential to forecasting. http://onlinelibrary.wiley.com/doi/10.1002/wea.2069/epdf " Practical experience and statistical verifications indicate that the average (or median) output from the ensemble is indeed the superior provider of ‘the most likely outcome’. The spread around this average tells how much trust can be put into this output, and, converted into probabilities, can warn about high-impact weather developments. It would therefore be quite possible to base the entire weather forecast guidance on the EPS. "
  9. I had a look at the ECMWF Extreme Forecast Index for temperature in Days 10-15. Not even a hint of cold even over Eastern Europe and Russia. I found that pretty surprising, it looks cold even on the ensemble but maybe it's going to be relatively cold compared with now, but still fairly nea average....
  10. That's probably a sound strategy in some situations but not in all situations. The difficulty is in knowing what kind of situation it is to begin with
  11. Essentially agree. Always will be hard for ensembles to pick the extent of the extremity, but the ECMWF do have their Extreme Forecast Index which is pretty good. The way they decide how to "duff the data" is quite interesting in itself (and way beyond me). I think they invest a lot of computing time into picking which things to perturb, and where they should be perturbed. If the atmosphere weren't chaotic, and if we could analyse it perfectly, would be no need to duff things up
  12. That's called "confirmation bias". Ensembles aren't perfect and they are underdispersive. That's what you need 100-200 members for. They are NOT "as prone to flipping as the op". Science backs this up.
  13. There's no solid evidence to back up such opinions. You need to be careful with netweather forum folklore. Not specific to NW of course; it's a human trait. http://journals.ametsoc.org/doi/abs/10.1175/2009MWR2960.1 One thing the EC always try to drum home is that model jumpiness does not correlate to accuracy. That's important but doesn't make our job muchy easier.
  14. Not an individual ensemble member, no. You would always expect an ens member to be worse than it's op. Their power lies in considering them as a family. The idea is that they represent initial uncertainty in the atmosphere well enough to also forecast future uncertainty. Doesn't work quite that well in reality* but they aren't bad. Deterministics flip and flop much more frequently than ensembles. *ensembles are usually underdispersive which means they never quite capture or forecast the full extent of uncertainty in the atmosphere. But they do a better job at uncertainty than deterministics. I heard 100-200 members would be needed realistically - which is beyond even the ECMWF right now!
×