Evaluation of footprint models using surface fluxes over clearly defined heterogeneity

In this work, an approach to evaluate the performance of two footprint models is presented which makes use of experimental data from several eddy covariance measurement complexes set up in an area of different well defined land use types. The footprint models are the Lagrangian stochastic model of Thomson (1987) using atmospheric surface layer parameterisation without roughness layer effect on the turbulent flow (Rannik, 2003), and the two dimensional analytical flux source area model by Schmid (1997). The experimental data were measured during two field scale campaigns. The first campaign comprised of four eddy covariance complexes set up in a line, with heterogeneous fetches influenced by three different types of land use depending on the wind direction. The second campaign comprised of four measurement complexes one of which was moved twice during the period under consideration. In the latter campaign the fetches were influenced by two different types of land use depending on the wind direction. The relationship between fluxes measured at different positions in the field is dependent on the flux contributions of the different land use types as determined by the footprint models.

For the first measurement campaign, we compare the flux contribution emitted by a certain land use type (triticale field) as determined by the two footprint models for each measurement point as a function of wind direction under neutral stratification. Next we show the comparison of the measured momentum flux differences between pairs of measurement positions as functions of the predicted flux contribution differences. Because the triticale field has higher aerodynamic roughness we expect that the relative differences in measured momentum fluxes increase along with increasing differences in triticale contribution. Concerning the second measurement campaign, depending of the wind direction some of the measurement positions involved in the study are influenced exclusively by a single land use type, while others have a source area consisting of a mixture of grassland and bare soil. To test the performance of the two footprint models, we use the fluxes having sufficiently high modelled flux contribution from either grassland or bare soil as references. With this information, for the positions with mixed fetch a modelled value of the sensible heat flux is determined by linear combinations of fluxes from purely grassland and bare soil, which is subsequently compared to the flux which was actually measured there.