Investigating the Performance of an Ensemble Prediction System in a Changing Climate
Elizabeth Satterfield, Justin McLay, Carolyn Reynolds Naval Research Laboratory, Monterey, CA
The aim of this study is to assess how predictability of an ensemble forecast system changes in potential new climates, which differ from that of present day. To achieve this we conduct numerical experiments using a version of the Navy Global Environmental Model (NAVGEM) in which the boundary forcing fields (e.g SST, ice cover) have been modified to be consistent with future climate states. Analyses that are physically consistent with a potential new climate state are produced through the assimilation of synthetic observations generated from climate model data. The local Ensemble Transform (ET) scheme is used to perturb the NAVGEM analyses to generate a 20-member ensemble, which is used as initial conditions for 16-day ensemble forecasts.
To assess the performance of the ensemble in a changing climate we apply linear diagnostics to the ensemble perturbations in a small local neighborhood of each model grid point. A local error covariance matrix is defined for each local region, and the diagnostics are applied to the linear space defined by the range of the ensemble-based estimate of the local error covariance matrix. The particular diagnostics are chosen to help investigate how well the ensemble performs in capturing the space of forecast or analysis uncertainties, predicting the magnitude of forecast and analysis uncertainties, and accurately representing the importance of different state space directions. The performance of the ensemble is assessed over 16-day forecast integrations for winter season dates for the baseline climate period (2015-2017) and future extreme climate period (2098-2100).