Can any model really predict the weather to a tee?

[Karl Lijnders]

In lieu of discussions happening in the Breaking weather section, I think we need to setup a discussion on where people stand when it comes to computer modelling of weather. How do you forecast? What level of scrutiny do you place the computer science of meteorology under and how much does it factor into your forecasting?

Do you focus on nature and natural signs to give you the inside scoop when it comes to weather approaching. I would really like to see what people think. Please be passionate about it, because it seems to be striking a raw nerve with some which is great, but attack the arguement and not the person!!

Play nice....

[droughtbreaker]

Most of the time models are really good but as I've learned over the last couple of years they are only a guide (a very good one) and not meteorological gospel. It is a good idea to use the models in conjunction with observations you make yourself, what sort of clouds are about, what the synoptic pattern is doing etc.

Sometimes models can really over do or under do things. They over did things a bit today although nowhere near as bad as they have repeatedly done for the last several years before we got this extended period of wet weather over the past several months. Remember it is mid to late December now and you just don't get 10-30mm from every cold front that passes through. AS we head into summer we will notice that 10mm is a very good fall from a front and that we need cut off lows, troughs with storms etc. to give us the really heavy falls. There are very few rain days in summer in comparison to spring and when it rains it is either heavy and we get average to above average falls or it is a very dry month if the rainfall when it comes is not heavy. That's the way things are in summer.

[Anthony Violi]

I think today's events are the perfect talking point. Which model predicted today well? None, because models cannot possibly factor in things like what happened today, with the trough stalling for most of the day, building in the west with decent upslide, then when it got to the sweet spot no upslide came, then it decided to rocket through allowing for no development pretty much. Yes, models get totals pretty close, and most of todays were ok. But not to a tee, no.

They are certainly better, late 80s you would wake up and the forecast was totally different to the night before. Then i distinctly remember one event early 90s that was forecast to be fine and at 9pm there was mayhem bearing down on the city, David Brown says some cold air in the mids was missed by the forecasters???? How would that happen? Certainly not today.

I know its a hot topic, but EC clearly on % is the best. GFS takes the best drugs. Laps is mortally wounded. JMA and Uk are very good. Nogaps has its moments. GASP is shocking, yet once a year provides a stunning upset and takes the cake, TWC Mesocast, actually no this a computer model discussion, it belongs in the bin, Access i have no idea about only David can answer that.

[Karl Lijnders]

I know its a hot topic, but EC clearly on % is the best. GFS takes the best drugs. Laps is mortally wounded. JMA and Uk are very good. Nogaps has its moments. GASP is shocking, yet once a year provides a stunning upset and takes the cake, TWC Mesocast, actually no this a computer model discussion, it belongs in the bin, Access i have no idea about only David can answer that.

LMAO! Nice one!! I think that should be the mantra attached to most of the forecasting threads!!

I agree Andrew 100% with what you have said. It is a guide. Out here there have been signs for a while, natural ones, that I look too. Rain trees have been going beserk, ants building mounds, black cockatoos sqwaking etc. But also the night before a system, I love to get a picture of the sky and see what it is saying. Just like with a sunrise too. This morning I could tell things were not as unstable as I liked, flat As and Acu. But I revised in my mind the 30mm I thought we would get to 15-20mm. Just off an observation with the eye and this was before consulting the computer models.

I think the EC vs the world of models is sometimes valid, but again this is a natural science and even EC stuffs up. I mean how can it go from 60mm here last night to 0mm tonight for the next system??

[Blackee]

I am finding that GFS has been on the ball of late.
It has been pretty accurate with Olga and now the low, which is now sliding SW towards NE SA. The extended GFS picked this a long way out also, so I'd have to dip me lid to GFS as compared to EC.

[droughtbreaker]

I'm sort of thinking the opposite Sniper. EC had ex Olga following the path it has to date pretty much from day one whereas GFS was holding it inland and then eventually shifting it off east and just brushing us. Of course GFS has come on board now. I reckon EC has handled it a lot better though.

In relation to the storms along the trough on Sunday, both of the main models (GFS and EC) missed it despite correctly predicting that conditions would be conducive to storm development. LAPS and MLAPS were all over it. There are some scenarios where you can't trust any of the global models, and I'd say thunderstorm development on marginal days or days where the instability signature is not all that obvious, and thunderstorm activity in general basically and drizzle areas which the global models rarely pick up well.
With these sort of events you have to rely on MLAPS and LAPS, and even then they can overdo or under do rainfall.

[Blackee]

True Andrew.
I thought GFS did pretty well with TC Laurence yet missed todays activity or lack of in SA.
I have found EC frustrating at times really ramping up totals early in the run and then abandoning the idea.
Each to their own hey!

[Karl Lijnders]

Finding a trend that models are intially going for the systems in the first few runs, backing away and going off with the fairies and then coming back towards the original plan.

I wouldn't be suprised if EC develops a low from the NW on this system which could see falls come back towards 100mm in some spots.

[Greg Sorenson]

I find it rather exciting to see which one gets the cake. It would be an interesting idea to keep a points score over a 12 month period to see which one wins. I take each though with a grain of salt though. They all seem to have trouble making up their minds for ex tropical lows.

[Blackee]

I find that GFS does a pretty good job for QLD and NSW. This weeks rain event was picked a long way out and predicted 200m+ falls.
Having said that, we have low confidence in GFS down here. EC has won most battles in 2010.