Jul 1, 2015 06:19 PM
https://plus.maths.org/content/el-nino-w...appen-next
EXCERPT: [...] Even a simple model has some predictive power. In reality, to make forecasts a few months ahead, climate agencies use more sophisticated models that describe climate processes in a lot more detail. To monitor El Niño they use the TAO/TRITON array of buoys that was placed in the tropical Pacific in the 1990s and reports information of temperature, winds and currents in real time. (If you'd like to know what a lonely buoy floating on the Pacific is recording right now, visit the website.)
These sophisticated models, which include the Met Office's global seasonal prediction system GloSea5, make use of fluid mechanics, which involves the famous Navier-Stokes equations (you can read about them here). The models find solutions to special forms of the equations, using observed data as input. Obtaining detailed solutions, however, requires an enormous amount of computing power. The models also suffer from what is known as the butterfly effect: to predict the future they need to take as input climate information from today, and even a small difference in these initial values can snowball into a massively different prediction.
To counteract this effect you can run a model many times, each time with slightly different initial values. The result is an ensemble of simulated futures, which also gives you a way of computing the odds of what will happen in reality: if the bulk of the simulations predict an El Niño event, then you conclude that such an event is indeed very likely....
EXCERPT: [...] Even a simple model has some predictive power. In reality, to make forecasts a few months ahead, climate agencies use more sophisticated models that describe climate processes in a lot more detail. To monitor El Niño they use the TAO/TRITON array of buoys that was placed in the tropical Pacific in the 1990s and reports information of temperature, winds and currents in real time. (If you'd like to know what a lonely buoy floating on the Pacific is recording right now, visit the website.)
These sophisticated models, which include the Met Office's global seasonal prediction system GloSea5, make use of fluid mechanics, which involves the famous Navier-Stokes equations (you can read about them here). The models find solutions to special forms of the equations, using observed data as input. Obtaining detailed solutions, however, requires an enormous amount of computing power. The models also suffer from what is known as the butterfly effect: to predict the future they need to take as input climate information from today, and even a small difference in these initial values can snowball into a massively different prediction.
To counteract this effect you can run a model many times, each time with slightly different initial values. The result is an ensemble of simulated futures, which also gives you a way of computing the odds of what will happen in reality: if the bulk of the simulations predict an El Niño event, then you conclude that such an event is indeed very likely....