Sources and Growth of Initial Condition Errors in Convection-resolving Forecasts in the Mesoscale Predictability Experiment (MPEX)

点击翻译按钮获取中文摘要

This research focuses on understanding how errors associated with particular features prior to convective initiation can influence forecasts of convection over the Great Plains region of the United States within the Mesoscale Predictability Experiment (MPEX). In particular, this study will evaluate the hypothesis that errors associated with upper-level synoptic features, midtropospheric moisture, midtropospheric lapse rate, and boundary layer moisture and shear at 18-24 h and 1 h prior to convective initiation are responsible for the lack of predictability in convection forecasts. These hypotheses will be validated by running a series of experiments whereby different sets of observations are assimilated within the most sensitive regions and compared to the control where these observations are withheld. In addition, the ensemble-based sensitivity technique will be applied to 3 km Weather Research and Forecasting (WRF) model ensemble forecasts that use initial conditions from a cycling ensemble Kalman filter. The sensitivities that are obtained from this method will also be evaluated by producing perturbed initial conditions that are consistent with adjusting a feature of interest (i.e., midtropospheric moisture) and comparing the resulting forecasts against the control where no perturbation is applied. The results from several cases will be compared to each other to evaluate whether convective forecasts are consistently sensitive to particular fields and features and the degree to which each case is characterized by unique sensitivities.Intellectual Merit:This project will enhance our understanding of the dynamical processes that limit the predictability of convective systems over a range of cases using an ensemble of convection-resolving forecasts. Moreover, this study will also demonstrate whether sensitivity analysis, which has typically been applied to phenomena characterized by linear error growth over longer time scales can be applicable to convection, which is characterized by non-linear dynamics and error growth over short time scales.Broader Impacts:The results from this study could provide guidance of where to take observations during future convective events that could be assimilated into numerical prediction models, which will hopefully produce better forecasts, greater lead time for warnings, and reduced loss of life. The results will be communicated with operational forecasters and others in the research communities to design better ways to observe the atmosphere and evaluate numerical models. Moreover, this study will also allow for the training of a graduate student in predictability, convective dynamics and data assimilation.