The Cold-air Pooling Experiment (COLPEX)
April 2012 - The COLd-air Pooling EXperiment (COLPEX) is a collaborative research programme concerned with understanding and improving predictions of cold-air pools which form in valleys during stable night time conditions.
COLPEX is a joint activity between the Met Office and NCAS.
During calm, clear nights the air temperatures in hollows and valleys can fall several degrees lower than on the surrounding hill tops. These cold pools are associated with hazardous conditions such as localised fog, icy road surfaces and poor air quality. Accurate forecasting of such phenomena is important. However, the small-scale nature of cold pools means they are often poorly represented in even the highest resolution operational numerical weather prediction (NWP) models.
The COLPEX programme has involved long-term detailed measurements of atmospheric flows in and around a small-scale valley. This field phase was carried out in the rolling hills of Shropshire, UK, specifically the Clun valley (see the image to the right - this photograph was taken during the snowy conditions of January 2010 and shows an episode of fog which formed in an adjacent valley and then spilled over into the Clun valley itself). Near-continuous measurements were made for a period of 15 months during 2009-2010. The measurements were made by a team of scientists from the Met Office and the Universities of Leeds and Salford. The experiment involved an array of sophisticated meteorological instruments. These included a wide variety of ground based sensors such as turbulence probes, custom-built automatic weather stations (measuring wind, temperature, pressure and humidity), visibility and radiation sensors, soil temperature probes, ground heat flux plates and a Doppler lidar. The measurement sites were distributed within the Clun valley in order that a detailed picture of the flow within the valley could be established. Some instruments were placed on the surrounding hills and also in an adjacent valley in order that comparisons could be made with the Clun valley. During periods of specific interest these measurements were supplemented by regular radiosonde releases and an instrumented vehicle which was used to map the air temperatures along various routes around the field site. A photograph of one of the main COLPEX field measurement sites, located at Upper Duffryn within the Clun valley, is shown in the image to the left. Two tall (50 m and 10 m) masts were erected at this site, hosting a variety of temperature, turbulence and radiation sensors.
A complementary aspect of the project involves the use of the Unified Model (MetUM), run at extremely fine horizontal resolution. The MetUM is the main numerical modelling system developed and used at the Met Office. It is a "seamless" system, in that different configurations of the same modelling system are used across all time and space scales, from global climate simulations, extending decades and centuries into the future, to regional weather prediction for only a few hours ahead. Different MetUM configurations are designed to best represent the processes which have most influence on the timescale of interest.
The horizontal grid spacing of the Met Office's highest resolution operational weather forecast is now 1.5 km. Whilst this represents the state-of-the-art in NWP, even this resolution is insufficient to capture the details of the flow on sub-kilometre scales, such as those within the Clun valley. The impact of these small (sub-grid) scales must be represented within the model however and this is accomplished through sub-grid physical parametrizations. NWP forecast products must also be corrected, or post-processed, in order that local detail such as the temperature changes and fog patches associated with cold pools can be represented.
The MetUM has been configured for the COLPEX field site, using a nested suite of domains, the inner-most of which has a variable horizontal grid spacing with a finest resolution of 100 m. Model simulations have been conducted for several cold-pool and valley-fog episodes that were observed during the field phase of COLPEX. Comparisons between the model predictions and the COLPEX measurements are being used to assess the ability of the model to capture the details of the flow within the valley. The overall aims of the modelling exercise are:
1. To investigate the extent to which a very high resolution modelling system can be used to provide accurate detailed local-scale forecasts for sites of specific importance (e.g. airfields).
2. Produce a comprehensive model dataset which can be used to complement and aid the analysis of the field measurements.
3. Use the fine resolution simulations, for example by coarse-graining the results, to improve the parametrizations of sub-grid processes at the coarser resolutions used in NWP and Climate simulations.
An example animated sequence of model wind and temperature across the COLPEX field site is presented in Figure 1. The results shown are from a 100 m resolution simulation of an observed cold-pool case during 10-11 September 2009, during clear-sky light-wind conditions. (Colour shading indicates the potential temperature in degrees Celsius. Line contours denote the terrain heights at an interval of 50 m. Wind vectors are denoted by arrows (units are metres per second) and the time interval between frames is 10 minutes.) A distinct cold pool forms in the valley during the evening and this continues to strengthen during the night. Close examination reveals that a gradual down-valley drainage flow forms along the valley bottom. The model temperature predictions are compared with the COLPEX observations for this case in Figure 2. The observed temperatures are reproduced by the model with remarkable accuracy.
The COLPEX programme will increase our understanding of how valley cold pools form and why, for instance, some valleys offer a more favourable environment for their formation than others. There are fundamental questions which currently remain largely unanswered. For example, to what extent is the cooling controlled by local processes at the valley bottom, such as radiative loss of heat from the ground and turbulent mixing of heat in the air above? Or is the drainage of cold dense air down from the hill tops and side valleys important in establishing the cold pool in the valley? Through the combination of measurements and numerical modelling we are aiming to address fundamental questions such as these. More generally, the results are being used to improve our understanding and ability to predict airflow in complex terrain.