The impact of upstream blocking, drainage flow and the geostrophic pressure gradient on the persistence of cold-air pools

Idealized numerical simulations are performed to investigate dynamical mechanisms affecting the persistence of cold-air pools in basins and valleys. The first orography type considered is a shallow elongated basin located upstream of a mountain ridge. For sensitivity tests, the mountain ridge is removed. The second type is a basin embedded in a plateau-like mountain ridge. In part of the simulations, this basin has an outflow towards the leeside plain so as to assess the impact of the drainage flow. The large-scale flow is taken to be in geostrophic balance. In the standard setting, it is perpendicular to the basin and the ridge. The main effect of a large-scale pressure gradient is to induce a circulation within a cold-air pool until the upper boundary of the cold pool is inclined such as to compensate for the ambient pressure gradient. The cold air accumulates where the ambient pressure is lowest. For a shallow basin, this means that part of the cold air may be lost due to advection but of the basin. The upstream influence of a mountain ridge in the lee of a shallow basin is found to be twofold. It tends to deflect the low-level flow towards the lower pressure, leading to an additional ridge-parallel force on the cold-air pool. On the other hand, the absolute wind speed is reduced, diminishing the turbulent mixing near the top of the cold pool. The simulations show that the first effect prevails for ridge-normal flow while second effect may dominate for other flow directions. Drainage flow out of a valley is found to be very important as it promotes the penetration of warm air into valleys very effectively. It may cause a cold pool in a deep valley to disappear more quickly than a cold pool in a shallow basin. Sensitivity tests show that the persistence of a cold pool depends on its depth, on its vertically integrated heat deficit, and on the maximum heat deficit at the bottom of the cold pool.