Affiliation:
1. Center for Analysis and Prediction of Storms, Norman, Oklahoma
2. Center for Analysis and Prediction of Storms, and School of Meteorology, University of Oklahoma, Norman, Oklahoma
Abstract
Abstract
This paper introduces the use of very large ensembles for detailed sensitivity analysis and applies this technique to study the sensitivity of model forecast rainfall to initial boundary layer and soil moisture fields for a particular case from the International H2O Project (IHOP_2002) field program. In total, an aggregate ensemble of over 12 000 mesoscale model forecasts are made, with each forecast having different perturbations of boundary layer moisture, boundary layer wind, or soil moisture. Sensitivity fields are constructed from this ensemble, producing detailed sensitivity fields of defined forecast functions to initial perturbations. This work is based on mesoscale model forecasts of convection of the 24 May 2002 IHOP case, which saw convective initiation along a dryline in western Texas as well as precipitation along a cold front. The large ensemble technique proves to be highly sensitive, and both significant and subtle connections between model state variables are revealed. A number of interesting sensitivity results are obtained. It is found that soil moisture and ABL moisture have opposite effects on the amount of precipitation along the dryline; that moisture on both the dry and moist sides of the dryline was equally important; and that some small perturbations were alone responsible for entire convective storm cells near the cold front, a result implying a high level of nonlinearity. These sensitivity analyses strongly indicate the importance of accurate low-level water vapor characterization for quantitative precipitation forecasting.
Publisher
American Meteorological Society
Cited by
56 articles.
订阅此论文施引文献
订阅此论文施引文献,注册后可以免费订阅5篇论文的施引文献,订阅后可以查看论文全部施引文献