An efficient SPDE approach for El Niño
It provides a more efficient computational method for climate scientists modeling El Niño, though the improvement is incremental.
The paper develops an efficient numerical framework for approximating the mean and covariance of SPDE-based El Niño models, validated against 2014-2015 data. The approach outperforms existing Taylor and stochastic Galerkin methods.
We consider the numerical approximation of stochastic partial differential equations (SPDEs) based models for a quasi-periodic climate pattern in the tropical Pacific Ocean known as El Niño phenomenon. We show that for these models the mean and the covariance are given by a deterministic partial differential equation and by an operator differential equation, respectively. In this context we provide a numerical framework to approximate these parameters directly. We compare this method to stochastic differential equations and SPDEs based models from the literature solved by Taylor methods and stochastic Galerkin methods, respectively. Numerical results for different scenarios taking as a reference measured data of the years 2014 and 2015 (last Niño event) validate the efficiency of our approach.