Correlations between thermal surface heterogeneities and secondary circulations during LITFASS-2003 - An LES study

In the framework of the LITFASS-2003 experiment, measurements in a heterogeneous area of 20km x 20km in the south-east of Berlin (Germany) were used to initialize and drive the parallelized large-eddy simulation (LES) model PALM . Four simulation cases were defined, in consideration of requirements of suitable synoptic conditions for the LES, to investigate the atmospheric boundary layer over realistic surface heterogeneities in the diurnal cycle. Eddy-covariance measurements from energy-balance stations over different land use classes in the LITFASS area (e.g. forest, lakes, different crops, grassland) as well as radiosondes and other meteorological instruments were used to prescribe initial profiles of temperature and humidity and surface heterogeneity properties in the course of day. The horizontal model domain covered an area, which varied between 40km x 40km and 56km x 56km with a resolution of 100m horizontally and 50m vertically (up to 2.5*10⁷ grid points).

The simulation data shows the development and evolution of heterogeneity induced mesoscale circulations (secondary circulations, SC), that are superimposed on the turbulent flow and spanning the entire boundary layer. Depending on the background wind velocity, which varied between 2m/s and 6m/s, local or roll-like structures are observed and extracted with the aid of an ensemble-averaging method. This study follows the hypothesis, that the bulk of the SC structures is explainable by a linear correlation between the underlying surface heat flux and the vertical velocity of the SCs, especially for distinct roll-like structures. To prove this hypothesis, a correlation analysis method was developed, that takes into account the background wind and thus the effect of advection. The method was applied to data from an idealized study with a constant heterogeneous surface forcing and to one realistic simulation case with time-dependent surface forcing. We show that there is evidence to suggest, that the hypothesis is valid for both, the idealized and the realistic cases and that it is possible to calculate a length, which determines the upstream fetch of the surface heat flux, which is responsible for SC structures.