Evaluation of Limited Area Model


In this Chapter the Limited Area Model (LAM) used in this study is evaluated to determine whether it is able to reproduce the intensities of intense precipitation events that are observed. The model is evaluated by looking at two past intense precipitation events, those of June and July 2007 in the UK, that were associated with extra-tropical cyclones. The June and July 2007 events were chosen because the intense precipitation resulted in extensive flooding in the UK, causing widespread disruption as well as fatalities. They are also well documented cases (e.g. Marsh and Hannaford, 2007; Stuart-Menteth, 2007; Blackburn et al., 2008; Grahame and Davies, 2008; Hanna et al., 2008; Marsh, 2008; Mayes, 2008; Prior and Beswick, 2008), therefore the atmospheric conditions predicted by the model can also be evaluated. Previous studies using LAMs have been shown in x2.3.2, to be useful tools when investigating large-scale atmospheric conditions on a regional scale (e.g. Colin et al., 2010). In this Chapter the LAM is evaluated specifically for the purpose of this study, i.e. the ability to reproduce intense precipitation associated with a synoptic scale extra-tropical cyclone.

The two events will be discussed individually, July 2007 is discussed in x5.2, and June 2007 is discussed in x5.3. The meteorological conditions of these events are discussed in xx5.2.1 and 5.3.1. The LAM used in this study is the UK Met Office’s (UKMO) Unified Model (UM) which is a Numerical Weather Prediction (NWP) model run as a limited area version for this study, as described in x3.4. The model is run at both a 12 km and a 4 km resolution, which is higher, or at similar scales, to other studies using LAMs to investigate intense precipitation events (Gustafsson et al., 2010; Duliere et al., 2011; Tryhorn and DeGaetano, 2011), although a coarser resolution than used by Roberts (2008b). The 4 km resolution is run both as one-way nesting within the 12 km run, and also directly from the global forcing data. The impact of an increase in resolution, and the different methods used to drive the 4 km runs, are discussed in x5.2.2, to determine whether the precipitation distribution and intensity is sensitive to these choices. The model output is compared to raingauge observations for both events in xx5.2.4 and 5.3.3, using the methods explained in x3.5. The July event was also compared to radar data (not available for June) to investigate the distribution of the precipitation, these results are presented in x5.2.3. The conclusions are presented in x5.4. The Chapter proceeds with a discussion of the July 2007 meteorological conditions that led to the intense precipitation followed by an evaluation of the model for this particular event. The July event was investigated first due to the flooding being more widespread and causing more damage than the June event. Several requirements for running the model to achieve realistic precipitation intensities were learned from investigating the July event, and these were applied to the June event discussed later.


--

Champion, A.J. and Hodges, K.I., 2014: Importance of resolution and model configuration when downscaling extreme precipitation, Tellus 66A, 23993, doi:10.3402/tellusa.v66.23993. http://dx.doi.org/10.3402/tellusa.v66.23993