A Backscatter Lidar Forward Operator for Particle-Representing Atmospheric Chemistry Models
Abstract. State-of-the-art atmospheric chemistry models are capable of simulating the transport and evolution of particles and trace gases but there is a lack of reliable methods for model validation and data assimilation. Networks of automated ceilometer lidar systems (ACLs) could be used to fill this gap. These are already used for the detection of clouds and aerosols, providing a 3D dataset of atmospheric backscatter profiles. However, as the aerosol number concentration cannot be obtained from the ACL data alone, one needs a backscatter-lidar forward model to simulate lidar profiles from the model variables. Such an operator allows then for a qualitative and quantitative model validation based on ACL data. In this work, we present a new backscatter-lidar operator which contains most of the microphysical properties of aerosol particles, discuss sensitivity studies and compare simulated with measured ACL profiles. A major challenge is the high sensitivity of the optical cross sections to the particle size and shape: A slightly different particle radius may lead to quite a large change of the scattering properties. As most particle size distributions are continuous in reality, the optical cross sections are averaged over certain size-intervals which also reduces the problematic and unrealistic sensitivity significantly. To calculate the attenuated backscatter coefficient, the size-dependent particle number concentration and the scattering properties of each particle type and size have to be simulated. While the particle number concentration is a model output variable, the scattering properties have to be determined by extensive scattering calculations. As these scattering calculations require assumptions about the particle refractive indices and shapes, sensitivity studies were performed to estimate the uncertainties related to the particle properties as represented by the model system. The strong sensitivity of the scattering characteristics to the particle radius was largely reduced by size-averaging algorithms. We focus on a case study of the eruption of the Islandic volcano Eyjafjallajökull from 20 March 2010 to 24 May 2010. The Consortium for Small-scale Modeling – Aerosols and Reactive Trace gases (COSMO-ART) model of DWD (Deutscher Wetterdienst) and KIT (Karlsruhe Institute of Technology) was used during this event for ash-dispersion simulation over Europe. For the forward model, the attenuated backscatter coefficient is used as lidar-independent variable, as it only relies on the laser wavelength. Finally, the forward modeled lidar profiles were compared to ACL measurements. Significant differences between ACL profiles and the output of the forward operator applied to the COSMO-ART data were found but also several identical features have been observed. Comparing the data quantitatively revealed that the model-predicted ash number concentration is slightly too high. We identified the following key issues which are mandatory for performing quantitative comparisons between forward modeled and measured ACL profiles: First, it is suggested that the ACL systems perform automatic calibration and return the attenuated backscatter coefficient directly. Second, the particles' scattering properties have to be analyzed even more extensively as significant differences of the backscatter efficiency occur depending on the shape and refractive indices. Nevertheless, the results of this study allow for a quantitative estimation of the volcanic ash mass concentration from ACL data as well as the related uncertainties. We consider this forward operator development as a crucial step for the future assimilation of the huge information contents of lidar backscatter data in chemical weather forecast models.