Use of Ensemble Sensitivity Analysis in the Forecasting of Convective Events

The recent increase in the use of ensemble forecasts has led to new information being available to forecasters, including uncertainty statistics and probabilistic guidance. Various techniques have been used to generate ensembles, one of them being the Ensemble Kalman Filter (EnKF), which produces analyses and subsequent forecasts that represent a probability distribution of the atmosphere with flow dependent errors. Ensemble Sensitivity Analysis (ESA) is an ensemble-based method that estimates the sensitivity of a longer-range forecast to early forecast or initial time weather features. In turn, this type of sensitivity analysis provides additional information from an ensemble that can highlight important dynamical relationships and can be used to increase forecaster awareness.

Previous work in this area has been primarily focused on synoptic-scale weather which involves mostly linear perturbation evolution at the 1-2 day forecast range. This study focuses on developing ESA within a multi-scale WRF model EnKF for severe convection, which is more influenced by nonlinear perturbation evolution and has a binary nature. Three model domains are used with grid spacing of 36, 12, and 4 km. Data assimilation is performed on all three domains, and a short range ensemble forecast is produced for each. Results applying ESA to multiple convective events will be shown, including both same and cross-grid sensitivity fields in order to capture both small and large scale features. Through real time sensitivity products, increased forecaster awareness and objective selection of sub-ensembles with higher skill can be accomplished. Also, adaptive observing systems can be designed to best reduce forecast uncertainty. Discussed will be how these applications of convective ESA can be used in an operational setting.