Demonstration of ensemble sensitivity-based targeted observing for convective-scale applications: Perfect-model experiments
Hill, Aaron J.
MetadataShow full item record
A series of experiments are designed and executed to evaluate predicted and actual impacts from targeted observations on idealized high-resolution ensemble forecasts of convection. Ensemble Sensitivity Analysis (ESA) is used to target single observations at prescribed times and vertical levels in advance of convection to systematically evaluate the effectiveness of ESA on convective scales. Observing System Simulation Experiments are utilized to generate synthetic target observations from a reference forecast at any prescribed location or time and control for model-error influences on ensemble forecasts. Performed experiments perturb the assimilation con figuration, vary observation type, assimilate non-targeted observations, and modify assimilation times to comprehensively assess the ESA-based targeting algorithm for convection. The ESA targeting method effectively predicts changes to the mean of response distributions under certain experiment confi gurations, but fails to adequately predict changes to response variance under all experiment permutations. Targeted observations produce equal impacts on response distributions as non-targeted observations for all cases considered, due to non-linear effects of the moist dynamics. In addition, localization applied during data assimilation negatively influences observation impacts by restricting analysis and forecast increments from assimilated observations. Targeted observations assimilated with an ensemble Kalman lter (i.e., no localization employed) produce predictable impacts, particularly at short lead times. Moist dynamics in the model and chaos seeding (i.e., numerical noise) contribute to poor predictability of forecast impacts. The results discussed, under idealized conditions and simpli ed ensemble con gurations, indicate potential pitfalls if ESA-based targeting methods are employed with real-time ensemble systems.