In recent years, several footprint models have been proposed. For evaluation of these models, footprint estimates of different approaches have been compared. Yet, suitable experimental data are necessary to decide whether the footprint models yield the correct predictions. Although several attempts to evaluate footprint models against observations have been undertaken in the past, there is still a lack of reliable full-scale data. The reason often lies in the constraints that are given by the model assumptions (e.g., restriction to the surface layer). In particular, stationary turbulence and horizontal homogeneity are fundamental requirements for most of the footprint approaches and are difficult to meet.

The requirements of stationary turbulence and, if necessary, homogeneous surfaces can easily be fulfilled in wind-tunnel or water-tank experiments. Moreover, measurements in laboratory studies have often been performed with high spatial and temporal resolution.

In this contribution, we present an extensive evaluation of a 3-dimensional Lagrangian footprint model using high-resolution wind-tunnel datasets. The present footprint model (LPDM-B) is based on a 3-dimensional Lagrangian stochastic particle dispersion model valid for stable to convective conditions, as well as for receptors above the surface layer. It employs an approach using backward trajectories of particles, the density kernel estimation and a spin-up routine. Simulations with LPDM-B have been performed with two versions of the model: (i) employing the original turbulence parameterisations, and (ii) employing the profiles retrieved from the wind-tunnel observations.

The wind-tunnel experiments have been conducted for a ground-level source in a convective boundary layer with significant shear as well as in a neutral boundary layer. For further evaluation of LPDM-B, vertical flux estimates are derived from the experiment of the neutral wind-tunnel applying K-theory.

Footprint predictions of LPDM-B show satisfactory correspondence to the observations, both in peak location and shape. Summary statistics confirm the good performance of the model for receptor heights ranging from close to the ground to the top of the boundary layer. Furthermore, it is shown that the sensitivity of the footprint predictions on the implemented turbulence statistics is not large. The LPDM-B version employing the standard turbulence parameterisations agrees well with the observations.