The sensitivity of numerical simulations to initial modifications of potential vorticity—a case‐study

Abstract

AbstractModifications of numerical potential vorticity (PV) fields according to features in water vapour (WV) images are combined with singular‐vector dynamics in an attempt to improve the numerical analysis in the study of a severe winter storm. The apparent mismatch between features in the WV images and upper‐level PV anomalies in the numerical analysis is corrected at levels indicated as sensitive by the fastest growing singular vectors. Model re‐runs, based on the inverted PV fields, are then carried out. Though numerical weather prediction (NWP) re‐runs are highly successful in improving the simulation of the storm, they produce a degraded simulation of a second wave which followed rapidly in the wake of the first one. Some success is achieved by enhancing its associated upper‐level PV anomaly at sensitive levels. The impact on the large‐scale flow arising from localized PV modifications is discussed and may point to a reason why the NWP runs based on PV modification occasionally fail to improve the simulation, e.g. the second wave in this case. Copyright © 2003 Royal Meteorological Society