Ask A Stupid Question
One of the professors in the Department thinks he's found a way to empirically determine the predictability of the atmosphere based on completely atmospheric data. It's a really clever method that finds situations of similar geopotential height over the world and determines based on how they diverge from another according to a Lotka-Volterra model (typical predator-prey model). Unfortunately, he didn't quite explain how one gets the ten day predictability limit, but if he's right, a theory that could derive such wonderful elements from first principles would be very cool.
To explain it more simply: imagine that the system of the world is a computer: the laws of nature are the processor and it has various forms of memory whose storage time you can calculate. Basically, the claim is that the memory of the atmosphere is around ten days. Looking in the memory of the atmosphere, you can say something about its behavior for up to ten days. Otherwise, the memory off which the program of weather calls is wiped on a ten-day cycle. The argument is that the memory of the ocean, land, etc. is more important in determining long-term atmospheric behavior.
Anyway, the experiment was run on data that is for a low-level of the atmosphere, where atmospheric turbulence might be expected to disturb "information" very quickly. After the talk, I asked the professor whether higher levels might have longer memories. He answered that much longer memories than ten days seem unlikely. At the moment, he is working on doing his experiment on a vorticity factor (insert hand-waving explanation here) that is known to be isentropic, that is unaffected by turbulence enough, only to be affected by thermodynamic considerations or something to that effect. Next week, he promises me the results. He believes the memory will be limited to twelve days. We shall see.
ESA(20031003.1)
One of the professors in the Department thinks he's found a way to empirically determine the predictability of the atmosphere based on completely atmospheric data. It's a really clever method that finds situations of similar geopotential height over the world and determines based on how they diverge from another according to a Lotka-Volterra model (typical predator-prey model). Unfortunately, he didn't quite explain how one gets the ten day predictability limit, but if he's right, a theory that could derive such wonderful elements from first principles would be very cool.
To explain it more simply: imagine that the system of the world is a computer: the laws of nature are the processor and it has various forms of memory whose storage time you can calculate. Basically, the claim is that the memory of the atmosphere is around ten days. Looking in the memory of the atmosphere, you can say something about its behavior for up to ten days. Otherwise, the memory off which the program of weather calls is wiped on a ten-day cycle. The argument is that the memory of the ocean, land, etc. is more important in determining long-term atmospheric behavior.
Anyway, the experiment was run on data that is for a low-level of the atmosphere, where atmospheric turbulence might be expected to disturb "information" very quickly. After the talk, I asked the professor whether higher levels might have longer memories. He answered that much longer memories than ten days seem unlikely. At the moment, he is working on doing his experiment on a vorticity factor (insert hand-waving explanation here) that is known to be isentropic, that is unaffected by turbulence enough, only to be affected by thermodynamic considerations or something to that effect. Next week, he promises me the results. He believes the memory will be limited to twelve days. We shall see.
ESA(20031003.1)
0 Comments:
Post a Comment
Subscribe to Post Comments [Atom]
<< Home