How accurate are ten day forecasts?

Off Topic
#1

Last Monday I wondered how accurate the ten day forecasts would be and also how often they are updated. I use Weather Channel, weather.com, for the Addison airport. The shaded numbers are the low for that time and it tends to shift between four days. It’s not perfect but I’m impressed. On Friday, they missed this morning by 20 degrees but the forecast was showing the low temp about right if two days late. That’s what happens when the front speeds up.

Following is an image of my screen with color followed by a text dump.

image
image1008×990 140 KB

Weather Channel - low temerature forecasts
When Sat Sun Mon Tue Wed Thu Fri Sat
Mon 7pm 35 16 15 25 31 27 31 37 39 42
Mon 10pm 34 16 15 24 31 27 31 37 39 42
Tue 7am 33 14 11 21 36 30 26 30 36 39
Tue 10am 34 13 11 20 36 30 24 29 35 36
Tue noon 35 12 12 21 32 28 25 33 36 39
Tue 2pm 35 12 12 21 32 28 26 33 36 39
Tue 10pm 34 12 12 21 29 28 24 27 34 36
Wed 10am 30 8 12 20 36 28 24 27 34 37
Wed 1pm 31 9 12 21 35 28 28 28 36 38
Wed 4pm 28 13 11 20 32 28 25 28 35 39
Wed 8pm 28 14 10 19 32 28 26 28 33 38
Wed 10pm 30 14 11 19 31 28 26 28 35 38
Thu 10am 21 14 9 18 32 28 25 27 36 40
Thu 4pm 18 15 11 17 31 27 24 25 36 39
Thu 10pm 18 15 10 16 31 30 25 26 35 40
Fri 7am 35 17 19 11 14 32 23 23 29 36
Fri 10am 33 17 19 10 14 32 23 22 29 41
Fri noon 34 16 18 11 14 33 27 23 30 35
Fri 3pm 33 17 18 11 16 33 27 20 25 33
Fri 5pm 32 16 17 11 16 33 27 20 24 33
Fri 8pm 34 16 18 12 16 33 27 21 25 34
Sat 8am 17 17 13 16 33 24 20 25 37 45
Sat noon 18 17 13 16 34 27 20 26 37 44
Sat 3pm 17 17 12 16 34 27 21 26 37 42
Sat 5pm 17 17 13 17 35 27 21 25 36 42
Sat 10pm 17 16 13 17 35 27 21 26 36 43
Sun 7am 14 15 14 17 36 25 21 28 40 47
4 Likes
#2

Weather Channel is an IBM company. While I loved working for IBM, I can assure you that like ALL weather forecasts, it’s voodoo. Nothing more. We can predict the passing of an asteroid that’s spent 4.5 billion years circling in space with FAR better accuracy than we can predict the path of a single thunderstorm across a county.

3 Likes
#3

The asteroid has a limited number of events that might change its path. Not so the weather. I once thought that if we actually collect ALL the events that can change what weather and plug them into a program on the biggest, fastest computer out there, we should be able to accurately predict the weather. Just before it gets here.

1 Like
#4

That’s not true. it’s an n-body problem and the calculations can get as complex as you want. It just sounds plausible because the average person doesn’t have a framework to apply or assess the asteroid model.
Just wait till we have spaceships. I guarantee the driver will be saying stuff like “navigating under relativistic velocities isn’t as simple as being on planet’s surface. I could predict the exact path of the storm on Jupiter, but nobody knows for sure how many generations it’ll take for us to reach alpha centauri” and people will still believe it.

2 Likes
#5

I wouldn’t call it voodoo. It’s pretty sophisticated and pretty accurate. I wish I knew what other things were going to do 5 days out with 90% accuracy. Say perhaps the stock market?

The process of gathering the data to feed the simulation models is really impressive.

3 Likes
#6

I like when all the models out there diverge… :thinking:

1 Like
#7

Lol. 90% accurate 10% of the time.

2 Likes
#8

I assume the 90% accuracy stats include/are bolstered by predicting the weather in places like San Diego (let me guess…it’s sunny and warm) or Cochise County, AZ (let me guess…it’s hot and dry?) or Caribou Balls, Nunavut (let me guess…it’s cold and cloudy?) and similar locales.

I only care if they are accurate 90% of the time where I live (because I am selfish and self-centered), and here in Dallas, the weather forecasts suck, especially in regards to predicting rain even within the same day.

1 Like