Towards Improving Forecasts of Severe Convection Along the Dryline through Targeted Observing with Ensemble Sensitivity Analysis

Poor mesoscale numerical weather prediction forecasts of severe convection may be ameliorated through model physics and data assimilation improvements, or simply by obtaining more observational data to incorporate into the assimilation process. Intelligent methods to target additional observations have been available for decades, which take into account fast growing errors (i.e., singular vectors) and gradients of tangent linear models (i.e., adjoint sensitivity). Ensemble sensitivity analysis (ESA) is a more recently developed targeting method that combines information regarding the largest uncertainty of a forecast state along with areas ripe for error growth to estimate where observations could be gathered to improve forecasts. ESA linearly relates a scalar forecast metric to prior model states through a regression of an ensemble of non-linear model forecasts and can be carried out during forecast post-processing. Incorporating uncertainty in the initial state, and an estimate of observation error (see Ancell and Hakim 2007), ESA predicts the impact of a new observation on the chosen response as well as the variance of the ensemble's distribution of the response.

Less-than-ideal results have been realized when ESA has been utilized for mesoscale convection/precipitation forecasts. Recent literature has noted an average of near-neutral impacts of additional observations when targeted with ESA. Given the desire to exploit advantages of ESA over other targeting methodologies (e.g., execution time), it is imperative to understand factors, which may include forecast nonlinearity, data assimilation procedures, and model error, that influence the prediction of observation impacts and the impacts after assimilation.

A 50-member ensemble is generated over ten cases of severe storms along the dryline with the Advanced Research core of the Weather Research and Forecasting model (WRF) and Data Assimilation Research Testbed (DART) software. An independent "nature run" is conducted with WRF to generate observations that are assimilated with the ensemble analyses, consistent with practices in perfect-model and observing system simulation experiments (OSSEs). Following forecast integration, target observations are selected through ESA and assimilated at varying pressure levels and forecast hours. A number of experiment permutations will be discussed, including changes to the ensemble filter, localization thresholds, and inflation magnitudes, which aim to diagnose relative impacts of target observations on mesoscale convection forecasts. Preliminary results suggest an overprediction of observation impact when localization functions are used during assimilation. Moreover, positive observation impacts are strongly dependent on choice of response function.