Hurricane Forecasts said by the experts: "More Art than Science"

[TheDanold]

Hurricane researchers, who forecast seven more storms this season, have flubbed the past two annual estimates because of unusual El Nino and La Nina weather phenomena in the Atlantic and Pacific Oceans.

The predictions reflect variables that make this kind of weather forecasting ``more art than science,'' said Eric Blake, a hurricane specialist at the National Hurricane Center in Miami. Two of the nine Atlantic hurricanes predicted already have occurred for the season that ends Nov 30. Last year, five storms emerged after nine were anticipated.
http://www.bloomberg.com/apps/news?pid=20601103&sid=aWuWjvhiv8Y4&refer=us


I'm not criticizing them, there is a lot of unknown variables like El Nino and La Nina, etc... that effect the results, it makes it very, VERY hard to predict what can come only months ahead.

Based on that, who on here would actually trust the models that make predictions for years or even decades later? There are always going to be far more unknown variables in a longer period of time and that is why long-term predictions are always harder to make. It would be foolish to trust them when the short-term ones are so inexact.

 

[Bonestorm]

Hurricane researchers, who forecast seven more storms this season, have flubbed the past two annual estimates because of unusual El Nino and La Nina weather phenomena in the Atlantic and Pacific Oceans.

The predictions reflect variables that make this kind of weather forecasting ``more art than science,'' said Eric Blake, a hurricane specialist at the National Hurricane Center in Miami. Two of the nine Atlantic hurricanes predicted already have occurred for the season that ends Nov 30. Last year, five storms emerged after nine were anticipated.
http://www.bloomberg.com/apps/news?pid=20601103&sid=aWuWjvhiv8Y4&refer=us


I'm not criticizing them, there is a lot of unknown variables like El Nino and La Nina, etc... that effect the results, it makes it very, VERY hard to predict what can come only months ahead.

Based on that, who on here would actually trust the models that make predictions for years or even decades later? There are always going to be far more unknown variables in a longer period of time and that is why long-term predictions are always harder to make. It would be foolish to trust them when the short-term ones are so inexact.

This is akin to saying that all climate science is bunk because the weathermen can't get weekly forecasts right. It's nonsense.

There are methods for testing the accuracy of climate models. "Hindcasting" is one particularly relevant method.

 

[TheDanold]

This is akin to saying that all climate science is bunk because the weathermen can't get weekly forecasts right. It's nonsense.

There are methods for testing the accuracy of climate models. "Hindcasting" is one particularly relevant method.

Not weekly, but months down the road are more long term forecasts. Weekly forecasts are more accurate because they have had trial and error and been refined and more bugs worked out.
Long term models are not even in their infancy yet, they have yet to be tried out even once. If weekly forecasts took decades to perfect, then long term models could take centuries to perfect.

I write software for a living, I have infinite more confidence in past software that has been in the field and tried out for years to work better than new software in new releases that has never been tested in real time. These models use software too.

 

[Bonestorm]

Not weekly, but months down the road are more long term forecasts. Weekly forecasts are more accurate because they have had trial and error and been refined and more bugs worked out.
Long term models are not even in their infancy yet, they have yet to be tried out even once. If weekly forecasts took decades to perfect, then long term models could take centuries to perfect.

I write software for a living, I have infinite more confidence in past software that has been in the field and tried out for years to work better than new software in new releases that has never been tested in real time. These models use software too.

I didn't say that the models are perfect. They're not. But they are accurate. Again, see hindcasting for more as to their accuracy.

Short-term predictions are much more difficult than long-term predictions.

 

[Topspin]

they are getting much better at it I'll tell you that. We watch them very close down here post Katrina and they nailed the last two to the nats ass.

 

[TheDanold]

I didn't say that the models are perfect. They're not. But they are accurate. Again, see hindcasting for more as to their accuracy.

Short-term predictions are much more difficult than long-term predictions.

How can you say that when the long-term predictions (over a few months) have yielded so much worse results? Short term forecasts aren't bad at all because you can at least see existing weather coming from elsewhere and know it won't change too much.

Hindcasting can help if the variables are known, but are they? Variables that affect future weather are going to vary just as much as future weather itself. As an example, you don't know that there is going to be certain wind currents in 2030, even if that would help you determine that those wind currents mean more hurricanes based on past weather.