There are a lot of things that are incorrect in this post. I’ll just hit the highlights. Climate scientists do not predict short term weather events. While short and medium range (typically up to about 15 days) forecast are derived from numerical models that are fundamentally the same as the models used for longer term climate scenarios (N-S fluid equation plus radiative transfer, turbulence, phase change of moisture, and heat, momentum, and moisture exchange with the surface of the earth), the application is entirely different. Short and medium range forecasts are concerned with deterministic events while climate simulations are concerned with running various CO2 scenarios to compute quantities such as, but not limited to, global means. You cannot simulate deterministic events more than a week or two in advance so don’t pretend that anyone (who actually knows what they are doing) is trying to do that. Climate and weather simulations are two different endeavors.
Two days in advance, Ian was very well simulated and the landfall was very close. Some of the longer term simulations had landfall anywhere from the panhandle of Florida to the Keys. With the hurricane approaching from the south and the Florida coastline oriented north-south, slight differences in the simulated tracks can result in large geographic differences in landfall. This was a much different geometry than if Ian was approaching land from a right angle. Regardless, similarly to many complex fields, what is really being forecasted is a probability function. The atmosphere is inherently chaotic and slight differences in the initial state can produce differences in the simulations. That is the reason why ensembles, both with different initial states and different models, are run. Essentially, an envelope of possible outcomes is created. Large spreads in the ensembles is an indicator of the inherent predictability of the event. This event had fairly large variations in the results until about 2 days before landfall. Some events are more predictable than others. Numerical Weather Prediction is a difficult problem because it is a unsteady fluid problem with an enormous number of degrees of freedom.
Quote:
Originally Posted by TrapX
Climate scientists didn't predict hurricane Ian 30 days in advance.
Climate scientists predicted a direct hit to Tampa 2 days in advance. They evacuated to be safe from total destruction.
Climate scientists didn't predict the location for the eye to hit until it was only a few miles away from land.
Their models were close, but very wrong and incomplete. I'm glad they are as good as they are, but realize there are many factors that cannot be predicted.
So climate scientists... What is the exact date when the next hurricane will hit Florida? And where will it hit? How strong? What will the high and low temperatures be on that day?
Climate scientists didn't even predict the 15deg drop in daytime high temperatures until after Ian passed and it was happening.
And now try to convince me that you all know what will happen 100 years from now?
But wait. I can predict the next hurricane. I'll write it down, seal it in an envelope. I'll open it the day it happens to prove I am better at predicting than you. Of course I will predict every combination of a hurricane to hit for every day, impacting every city, and for every intensity. Thousands of prediction envelopes. One WILL be right. That's the one I'll pull out and show the world how great I am.
Analyzing data points to predict the future is a lot like that scenario. Pick the data points you want to use that show your sponsor's theories are spot on. Ignore the rest. Extrapolate into the future.
|