General Weather/Earth Sciences Topics => Weather Conditions Discussion => Topic started by: Selby8 on October 15, 2017, 06:22:18 PM

Title: Why does it seem like the forecast only gets revised upwards? (Lapland, Finland)
Post by: Selby8 on October 15, 2017, 06:22:18 PM
Hi everyone, I'm far from sure that this is the right place to ask, but I Googled weather forum and here I am.
So I've been following the weather forecast for Rovaniemi, Finland on a daily basis since the end of last month, because I'll be visiting next month and I'm hoping for some snow.
I've been surfing mostly on Wunderground.com but also briefly had a look in some other sites.
They're having a very warm October(3 degrees Celsius above average), but that's not what I wanted to ask, as the reasons for that are perfectly clear.
I'm also not asking why a weather forecast is inaccurate, I know how hard that can be to achieve a good forecast.
However, I did notice a strange pattern. The 10 day forecast would generate average seasonal temperatures for 6-10 days ahead and then as we move to 5-7 days before a given date the temperatures would be revised upwards. 2-4 days before the date it would be revised even more significantly and then finally, the actual temperature is higher than the revised forecast.
To put it into real numbers, it was originally forecasted to drop below freezing and snow on October 2nd, and then on many days since, but it never happened, not even close. In fact, it got even warmer in the last few days and will get warmer tomorrow and on Tuesday. Then there's a cold spell in sight, but it gets "skimmed" by the hour until soon I believe there'll be nothing left. From this pattern I'm convinced the permanent snow there will be at least a month late.
So why is the inaccuracy always in a single direction?

I'll give another example from where I live in coastal Israel. Rain starts in the forecast as a LOT of rain. Heatwaves start as average temperatures.

One would expect the weather sites interest would be to give an accurate prediction for what would actually take place and not to rely solely on no longer relevant statistics from the past. That's what an almanac is for.
Honestly, a better strategy would be to take the average temperatures and add a few degrees :)

Thanks and sorry for the long post.
Title: Re: Why does it seem like the forecast only gets revised upwards? (Lapland, Finland)
Post by: ocala on October 16, 2017, 06:37:20 AM
I think you answered your own question above when you said "I know how hard that can be to achieve a good forecast."
It's a thankless job. Forecastors get no credit when they make a good forecast but get criticized when they make a bad one. There is too many factors that can change in a long term forecast.
Basically anything past 3 days is useless.
As for the issue with the temps going up I think that just may be an anomaly. Better data as the day/days of interest get closer lets them refine the forecast. I have seen what you are describing but have also seen it go the other way.
When they make a forecast they use models and local knowledge. They know from past experience that some models aren't accurate in certain situations and can't be trusted. So basically they are making a consensus forecast based on available data from models they do trust and their local knowledge of past weather conditions for the area. It's basically a scientific guess. You have to remember that the dynamics are constantly changing in the atmosphere. Nothing is static. A forecast made 6 hours ago could change. Sometimes they get it right sometimes they don't.
I don't know about in Israel but in the states the weather service puts out a product called a weather discussion. What it says is that this model says this, this model says that, this high pressure is here, the jet stream is doing this etc etc. From this long discussion you get a forecast that says,
Partly cloudy with a 30% chance of rain. High around 80. 
Bottom line is they do the best they can from the best available data they have.
If you want to see for yourself these models have graphical outputs of temp, rain, clouds etc and you can see for any time frame how they are all different.