COLPEX: Cold air pooling in small-scale valleys

Cold air pooling and fog formation in valleys is an important practical problem in terms of road safety, agriculture and air quality for example. Many previous studies have focussed on relatively large scale mountain valleys. In contrast this study looks at smaller scale valleys typical of much of the UK. This is a challenging forecasting problem due to the small (often sub-grid) scale of the terrain and the sensitivity to the land surface and stable boundary layer turbulence schemes. This talk will present observational and modelling results from the recent COLPEX (COLd-air Pooling EXperiment) campaign conducted in the Clun Valley, Shropshire, UK. The valley has a typical width of ~1-2km and a depth of ~150m. In the operational Met Office 1.5km model the valley is completely unresolved. The talk will describe the instrumentation deployed for the field campaign over the autumn and winter of 2009/10 and give examples of observed cold air pooling events. Cold air pooling events were surprisingly common, but events with fog forming in the valley but not on the surrounding hills were very rare. Numerical simulations have been conducted using the UK Met Office Unified Model run at a resolution of 100m over the study area. Results from some of the IOPs demonstrate that at this resolution the model reproduces observed cooling rates in the valley to a high degree of accuracy.

Focussing on one or two of the IOPs during the field campaign will allow a detailed study of the processes controlling cold air pooling in such moderate scale valleys. Recent idealised numerical simulations have suggested that for smaller scale valleys drainage effects are relatively unimportant and the sheltering effect of the valley dominates the cooling. This is an idea that needs testing in more realistic 3-d valleys. The observations are important in validating the high resolution numerical modelling, but it is hard to measure the effects of advection on the energy balance in the valley using observations. Using the model a detailed energy budget can be conducted. This shows that advection and boundary layer terms in the temperature equation are both large, but nearly cancel. The sum of these terms is primarily responsible for the observed temperature change in the model. In-situ radiative cooling is generally a smaller contribution. The use of back trajectories allows the source of the colder air to be tracked. This demonstrates that the cold air in the valley is generally decoupled from the flow aloft. Trajectories generally lead back up the valley floor towards the head of the valley. Understanding of the processes controlling cold air pooling will be valuable in developing new downscaling methodologies for forecasting cold air pooling events based on the existing 1.5km operational forecasts. Details of the cold pool climatology in both the observations and in a long (2-month) model simulation will be discussed in a separate talk.