The first of the Atmospheric and Oceanic Sciences colloquium series began Sept. 25 spotlighting University of Wisconsin professor Daniel Vimont’s summer research. Vimont developed stochastic-dynamic models to understand and predict the 2023 El Niño event.
According to the International Encyclopedia of the Social and Behavioral Science, stochastic-dynamic models are designed based on the continuous noise and variance impacting a particular model. El Niño is crucial to understand as it impacts global climates, especially in the tropics where crop production and harvest may be stalled due to severe droughts as a byproduct of El Niño, Vimont said.
Vimont said El Niño-Southern Oscillation is a warming across the Tropical Pacific’s sea surface temperature. It tends to peak late in the year around November and December.
“Pacific trade winds relax, resulting in a slope of the thermocline decreasing and warm water is brought up to the surface. The ocean surface increases around one degree Celsius,” Vimont said.
Stochastic-dynamics physics popularly models El Niño using the laws of physics to forecast its evolution, Vimont said. This year, the dynamic physics model suggested a big El Niño event to occur. But, different statistical models forecasted no El Niño event.
NYT health reporter shares 5 lessons learned from reporting on COVID-19
Vimont said he built a statistical model with UW’s Physical Sciences Lab that was the only model in agreement with the International Research Institute for Climate and Society dynamical models. He created a system of nonlinear equations that govern the atmospheric system without knowing what equations there were to control the environment.
Many events and phenomena in the environment can’t be built with an equation because of high variability and unpredictability, Vimont said.
“So we took the most basic approximation of the system, using the Taylor System around some climatological state, resulting in a model with a linear approximation for nonlinear equations plus the remaining non-linearities,” Vimont said.
Vimont said he took an inverse approach and used data from the satellite basis assimilation data to predict the physics of El Niño. He would nudge the system through time, by separating it from the initial and moving forward.
Vimont’s model resulted in a deterministic forecast, as well as a forecast spread which evolves due to unknown stochastic forcing.
By forcing the system, there was an issue of having too much variability in the model, which Vimont said made them “under-confident in the forecast.” To control this, he incorporated seasonal control called cyclostationary limits to reduce the noise.
Vimont said the noise had the greatest amplitude between winter and spring, resulting in high error, though the error shouldn’t theoretically grow until late summer and early fall.
Vimont’s model has strengthened researcher’s ability to predict El Niño and how it may impact climates in the tropics, as well as how it may be predicted as climate change continues.
