Why cant the weathermen get it right?

Born_2_Kill_83

Born_2_Kill_83

Born_2_Kill_83

Soldato
Joined
10 Nov 2006
Posts
8,563
Location
Lincolnshire
I’ve been wondering this for sometime, when checking the weather why is it so unreliable?

For example, I checked the weather last Friday afternoon for my area, the outlook for Saturday was heavy rain by 10am in the morning clearing up in the afternoon. 10am Saturday it was nice and sunny, with bright clear weather all day, I again checked the weather Saturday which showed rain all day Sunday. On Sunday the weather was clear and sunny till 8pm and we finally got some rain.

It also seems if they're not sure what the weather is doing they try and cover all angles, take the following week each day has a picture of a cloud, bit of sun and a drop of rain, wtf!??
 
Because to qccurately predict the weather days in advance, you need ENORMOUS computing power and hyper-accurate measurements of temp, pressure, humidity etc. A difference of half a degree may make the difference between a nice, sunny day and a dull rainy one a week in the future. This was the cause of the famous 1987 storm that Michael Fish assured us would never happen.

Modern medium-range forecasts by agencies such as ECMWF run many scenarios side by side altering the data slightly to one side of the recorded values, and then the other to see how errors in measurements (which can NEVER be perfectly accurate) will pan out. After all these models have been run for the required future date, they use an average of the results to determine what is most likely to happen. For example if 26 of 31 forecasts run say the temp over central England will be 19.5C next Wednesday, then that's what they'll forecast. It's a LOT more accurate than a single forecast method, but still in its infancy. It's also widely thought of as being THE most computationally intensive procedure in the world. Supercomputer budgets are ENORMOUS.

In a nutshell, no matter how powerful your computers are, you will never be able to accurately predict the future weather, because you can never measure the existing weather to an infinite degree of accuracy. You can get pretty close, however...
 
Lies!
rain.jpg
 
Haircut said:
Welcome to chaos theory.
Click to expand...

Indeed...../me strokes MSc dissertation....

There's also the notion taht the 'grid' of measurements is only about 1km on a side for the UK - such a size could easily conceal a thunderstorm or such like, which can have dramatic effects on local weather, as well as introducving significant local perturbations whcih contribute to the chatic variation.
 
12.5Km is the highest resolution the Met Office use at the moment for UK forecasts, 40Km for global. Thunderstorms are usually a LOT bigger than 1Km across, but even with a very, very high resolution model, you could easily miss small cumulus clouds that may produce showers. There's a bit of info about numerical weather analysis here:

Linky
 
Last edited:
MrFunex said:
12.5Km is the highest resolution the Met Office use at the moment for UK forecasts, 40Km for global. Thunderstorms are usually a LOT bigger than 1Km across, but even with a very, very high resolution model, you could easily miss small cumulus clouds that may produce showers. There's a bit of info about numerical weather analysis here:

Linky
Click to expand...

For paid-for forecasting they go down to 1km - but I agree regarding 'normal' forecasting.
 
I think they get it right more often than not, I only curse them on the occasion when they get it wrong and I've made plans for that day, but every other day I haven't paid much attention to the forecast and wouldn't realise if they got it right or not. Maybe we should start a weather watch and collect some hard data :)
 
AcidHell2 said:
I don't know, but weather has thousands of variables and only the most influential ones are measured. We still don't have a full understanding of weather and as such, can't do a full working model.
Click to expand...

Weather doesnt have thousands of variables. Its basically temperature, pressure, humidity, wind and precipitation. The limits on forecasting are set (as mentioned) by the chaotic nature of the atmosphere, and the limitations set by the computational power available. There is no point measuring further variables as the limitations on the main variables are still the gross factors that limit accuracy.
 
Variables can include:

Temperature
Humidity
Pressure
Cloud heights, amounts, types
Wind direction and speed
Sunlight
Rainfall
Atmospheric composition (rarely collected, but still important)

Most of these variables will need to be collected for different altitudes by radiosonde ascents.
 
Visage said:
Weather doesnt have thousands of variables. Its basically temperature, pressure, humidity, wind and precipitation. The limits on forecasting are set (as mentioned) by the chaotic nature of the atmosphere, and the limitations set by the computational power available. There is no point measuring further variables as the limitations on the main variables are still the gross factors that limit accuracy.
Click to expand...

To get it correct all the time, you have to include all variables. Not just the main ones. But yes until we get the main ones to a decent level I agree.
 
AcidHell2 said:
To get it correct all the time, you have to include all variables. Not just the main ones. But yes until we get the main ones to a decent level I agree.
Click to expand...

The problem is that there is no point getting accurate data for the lesser variables - their effect is so small that after only a few iterations the variation in forecast caused by uncertainties on the main variables will swamp any variability caused by the lesser ones.
 
You must log in or register to reply here.
Back
Top Bottom