The aim of this study is therefore to assess the impact of various numerical aspects on the UM simulation of Super-typhoon Haiyan. The first aspect is the dynamical core since a new one, known as ENDGame, is now available for testing alongside the existing dynamical core. The second aspect is the horizontal resolution and so, to simultaneously investigate both these aspects, a set of limited area model (LAM) simulations have been run using both dynamical cores at horizontal resolutions ranging between 25 km and 1.5 km. The configuration of the 4 km and finer resolution models is based on the 1.5 km model used operationally over the UK and hence convection is treated explicitly, whilst the coarser resolution LAMs are configured in the same way as the operational global forecast model and hence convection is parametrized. The third aspect is the relative importance of the representation of convection compared to the horizontal resolution. In order to investigate this, additional simulations have been performed for horizontal resolutions in the range 12 km to 4 km so that there is a pair of parametrized and explicit convection simulations for each of these resolutions.
The simulations are all performed on a domain of approximately 5500 km x 2500 km which spans the full lifecycle of the typhoon. Most of the simulations are initialised approximately four days prior to landfall, when the observed central pressure was close to 1000 hPa. The initial data for the simulations is the Met Office operational analysis with the 25 km global forecast generating hourly lateral boundary conditions for the LAMs which are all nested directly inside it.
In this presentation we describe and illustrate results summarising the characteristics of the simulated typhoon, in particular its simulated intensity, and their sensitivity to the changes in model numerics described above. Consistent with the systematic tropical cyclone behaviour of the global forecast model, the 25 km LAM using the current dynamical core produces only a relatively modest deepening. The 4 km and finer explicit convection simulations produce a more intense typhoon, with the simulation using the new dynamical core being most intense (within 20 hPa of the peak observed intensity) and approximately 30 hPa deeper than the simulation using the existing dynamical core. The pairs of explicit/parametrized convection simulations indicate that the representation of convection dominates over the choice of horizontal resolution.
Finally, an analysis of the diabatic heating tendencies will be presented with the aim of explaining the reasons behind the large range in peak intensity across the simulations.