Abstract/Summary

Progress in numerical weather prediction (NWP) is made through better understand¬ing of the physical processes represented in numerical models and their impacts on the dynamics of large-scal~ weather systems. Here, potential vorticity (PV) tracer diagnostics are used to investigate the representation of processes in the Met Office Unified Model (MetG:l1). An exact budget of the PV tracers is derived and a "dynamics-tracer inconsistency" diagnostic implemented to quantify non-conservation of PV by the dynamical core which was not previously accounted for. It is shown that non-conservation of PV by the dy¬namical core can have comparable tendencies to the dominant physical processes implying that non-conservation of PV by a dynamical core can, and should, be quantified alongside PV modification by physical processes. Recent work has shown that the sharpness of the extratropical tropopause declines with lead time in KWP models. In the MetUM, the advection scheme is shown to result in an exponential decay of tropopause sharpness and non-conservative processes are shown to sharpen the tropopause. The systematic errors in tropopause-level PV are comparable to the tendencies associated with physical processes, suggesting that the systematic error in tropopause sharpness could be significantly rednced through realistic adjustments to the model physics. I' Turbulent mixing within the boundary layer has been previously shown to produce positive PV anomalies that can be advected into cyclones and reduce growth rates through an increase in static stability; however, it is unclear whether N\VP models correctly represent this mechanism. In the MetUM, the generation of these positive PV anomalies is found to be less clear due to large cancellations with other physical processes in the cold sector. Front-relative compositing .is used to separate the cold and warm sectors, providing the basis for investigating PV generation in the boundary layer systematically by compositing over many fronts.