PDA

View Full Version : Intern Project



Bill O
12-11-2007, 11:18 PM
Is there anymore data on the latest forecast verification project? It looks interesting.

I could swear we talked about this in the past but why is the 24 hour forecast better than the 12 hour? It defies most of the logic behind atmospheric modeling and forecasting.

Is it the weather models or bad human interpretation or both?

If you know nothing about the weather you could make surprisingly excellent 12 hour forecasts by using the persistence method. That is, forecasting the weather in 12 hours to be exactly the same as it is right now. In San Diego you could make a living off that forecast. But maybe that theory falls apart on Mount Washington.

Knapper
12-12-2007, 01:18 AM
Persistance works for valley forecasts and in areas like LA and San Diego but when you are dealing with the 850 mb level of the atmosphere (MWO), things don't play out that easily. It's like the notion that daytime temperatures are normally warmer than night time temperatures in most locations. On MW, temperatures cool and warm on different time scales. We are in what is called free air (not to be confused with the free atmosphere) so therefore, we are not under the influence of normal variables like valley locations.

Brad
12-12-2007, 05:38 AM
Persistance works for valley forecasts and in areas like LA and San Diego but when you are dealing with the 850 mb level of the atmosphere (MWO), things don't play out that easily. It's like the notion that daytime temperatures are normally warmer than night time temperatures in most locations. On MW, temperatures cool and warm on different time scales. We are in what is called free air (not to be confused with the free atmosphere) so therefore, we are not under the influence of normal variables like valley locations.
Ah, so you guys are above it all.

Bill O
12-12-2007, 06:44 AM
Persistance works for valley forecasts and in areas like LA and San Diego but when you are dealing with the 850 mb level of the atmosphere (MWO), things don't play out that easily. It's like the notion that daytime temperatures are normally warmer than night time temperatures in most locations. On MW, temperatures cool and warm on different time scales. We are in what is called free air (not to be confused with the free atmosphere) so therefore, we are not under the influence of normal variables like valley locations.

I know, I know. I'll leave persistence to Brick Tamland.

But what's the theory behind a more accurate mid-range forecast?

Mike D
12-12-2007, 09:37 AM
I know, I know. I'll leave persistence to Brick Tamland.

But what's the theory behind a more accurate mid-range forecast?

Could it be due to the periodicity of the day? If you think of the day as a sine curve, 12 hours from "now" will put you in the exact opposite position or direction on that curve. On the flip side, conditions at 24 hour intervals might be very persistent.

Is "Brick Tamland" a name from America: The Book? You take a building material and a breed of horse and that's your TV name.

Bill O
12-12-2007, 12:35 PM
Could it be due to the periodicity of the day? If you think of the day as a sine curve, 12 hours from "now" will put you in the exact opposite position or direction on that curve. On the flip side, conditions at 24 hour intervals might be very persistent.

Is "Brick Tamland" a name from America: The Book? You take a building material and a breed of horse and that's your TV name.

I should note that my persistence comments were unrelated to MWO forecasting, just some interesting data.

On the summit they are still using model data for the 12 hour forecast. Maybe too much human emotion is going into the forecast, or not enough. I don't know. Or maybe its something else entirely.

Brick Tamland is the weather man from Anchorman.

Mike D
12-12-2007, 01:28 PM
And I should add that I took Weather 3 times.

Knapper
12-12-2007, 07:53 PM
We are threading this topic due to one image saying that we are better at 24 hour forecast than 12 hour forecasts. First off, the score is out of two so if you examine closely, the variability is not that much between one period to the next. But more importantly, the image posted is just for one element examined by our intern. In reality, when all things are factored together, there is a downward trend in accuracy like one would expect. When all was said an done with the months examined, our intern found that we are 78% accurate for the 12 hour, 76% for the 24 hour and 72% for the 36 hour. So overall not bad for a place with such variability. But getting back to the 12 hour vs 24 hour debate for sky cover, it has to do with convective clouds (or in more simpler terms, clouds that form during daytime heating) more than anything else. As to the quote by Bill O: "On the summit they are still using model data for the 12 hour forecast. Maybe too much human emotion is going into the forecast, or not enough. I don't know. Or maybe its something else entirely." We forecast how we were taught in college and most NWS and weather agencies use the same methods that we use. If you want a second opinion, check out NWS Grey ME's 12 hour summit forecast put out every morning for comparison. It is not as easy as it looks.

Bill O
12-12-2007, 10:42 PM
We are threading this topic due to one image saying that we are better at 24 hour forecast than 12 hour forecasts. First off, the score is out of two so if you examine closely, the variability is not that much between one period to the next. But more importantly, the image posted is just for one element examined by our intern. In reality, when all things are factored together, there is a downward trend in accuracy like one would expect. When all was said an done with the months examined, our intern found that we are 78% accurate for the 12 hour, 76% for the 24 hour and 72% for the 36 hour. So overall not bad for a place with such variability. But getting back to the 12 hour vs 24 hour debate for sky cover, it has to do with convective clouds (or in more simpler terms, clouds that form during daytime heating) more than anything else. As to the quote by Bill O: "On the summit they are still using model data for the 12 hour forecast. Maybe too much human emotion is going into the forecast, or not enough. I don't know. Or maybe its something else entirely." We forecast how we were taught in college and most NWS and weather agencies use the same methods that we use. If you want a second opinion, check out NWS Grey ME's 12 hour summit forecast put out every morning for comparison. It is not as easy as it looks.

Ryan, you've misunderstood me entirely. You should check who invented the higher summits forecast that Mount Washington issues. I'm curious about the data, how the project is coming and how its been improved.

I think the MWO forecasts are excellent on all time frames and variables. As a meteorologist I find it fascinating that the 24 hour forecast is more accurate than the 12 hour forecast, for this one variable and time frame. It's interesting and nothing more. Your answer about the models not resolving convective clouds answers everything.

I'm not looking for a second opinion, and I know its not easy. I've done it.

Bill O
12-12-2007, 11:06 PM
..... and I know its not easy. I've done it.

That's not entirely true. When the project was started it was setup to be as easy and repeatable as possible. That way an intern could put out very good results. I'm sure it had flaws, after all it was version 1.0.

It was heavily model dependent and only as precise as the situation required. As Ryan noted, MWO sits in the free atmosphere. Models are surprisingly good at forecasting the state of the free atmosphere, no trees and butterflies to mess things up at 800mb.

Mike D
12-13-2007, 10:07 AM
Yikes, meteorologists fight so dirty.