Very large ensemble ocean forecasting experiment using the Grid computing infrastructure

Abstract

Atmospheric and oceanic ensemble forecasting is a way to deal with uncertainty related to inaccurate knowledge of the initial state of the atmosphere and the ocean, the lateral and vertical boundary condition errors and the model physics shortfalls (Lewis, 2005, Epstein, 1969). Since the atmosphere and the ocean are extremely non-linear systems (Lorenz, 1993, Saravanan et al., 2000) initial uncertainties can amplify and limit the predictability of short term forecasts (Kleeman and Majda, 2005). For the ocean, ensemble forecasting is a novel field. Ensemble methods are used to compute the background error covariance matrix in data assimilation schemes (Evensen, 2003) but are not used yet to quantify the forecast uncertainty in short term ocean forecasting systems. Initial conditions uncertainty is a major source of unpredictability for ocean currents due to the limited observations available for nowcasting and the highly non-linear physics. In this study we explore the short term ensemble forecast variance generated by perturbing the initial conditions using a new computational facility, so-called Grid infrastructure (http://grid.infn.it/), distributed over the Italian territory. This infrastructure allowed us to perform several ensemble forecast experiments with 1000 members: they are completed within 5 hours of wall-clock time after their submission and the ensemble variance peaks at the mesoscales.