Improved Localization of an Ensemble-based Observation Impact Estimate

The goal of this project is to improve an ensemble-based estimation for forecast sensitivity to observations that is straightforward to apply using existing products of any ensemble data assimilation system. Observation impact depends on the instrument type, observation type, observation locations, as well as the availability of other observations. Due to limited ensemble sizes compared to the large degrees of freedom in typical models, it is necessary to apply localization techniques to obtain accurate estimates. Such localization needs to consider time-forecast dependency in addition to spatial and cross-variable dependencies. Here a dynamical localization method is applied to improve the observation impact estimate. We employ a Monte Carlo ‘group filter' technique to limit the effects of sampling error via regression confidence factor (RCF). This study employs a simple isentropic primitive two-layer model, to allow for reliable results without complications of large model errors, as has been done in several other EnKF studies. Results show that the shape, location, time-dependency, and variable-dependency of RCF localization functions are consistent with underlying dynamical processes of the model. Single-observation experiments displayed the evolving structures of the actual forecast error reduction with increasing forecast time and allowed for a more qualitative evaluation of dynamic versus static localization applied to ensemble impact estimates Application of RCF to observation impact estimates are qualitatively better at capturing actual forecast error reduction, especially at longer forecast lead times, compared to fixed localization methods. When verified against actual impact, the ensemble estimates using RCF localization showed marked improvement especially for longer forecasts at midlatitudes. However, deficiencies of the impact estimates occurred near the equator due to large discrepancies between the RCF function and the localization used at assimilation time. These latter results indicate that there exists an inherent relationship between the localization applied during the assimilation time and the proper localization choice for observation impact estimates. The optimal use of adaptive localization for he impact metric relies on the assumption that consistent localization was used during the assimilation as well.