Assessment of BSC-DREAM8b model using LIRIC (lidar and radiometer inversion code)

Resumen

Every year a large amount of mineral dust is transported from arid regions and injected into the atmosphere under specific weather conditions. The fact that mineral dust can produce a variety of problems to inhabitants both in and around desert areas (deaths and damage caused by traffic accidents, road disruption, aviation operations and impacts on human health, such us allergies, respiratory diseases and eye infections, among others) inspired the development of different dust forecast models as BSC-DREAM8b. Several approaches have been used to assess the columnar [Basart et al., 2012] and verticallyresolved [e.g. Guerrero-Rascado et al., 2009] performance of BSC-DREAM8b model using sunphotometric and lidar data, respectively. However, the available lidar data used (i.e. particle backscatter coefficient profiles) did not allow a quantitative evaluation up to now. Recently, the LIRIC software (lidar and radiometer inversion code) have been developed [Chaikovsky et al., 2008] and disseminated in the EARLINET community. After its exhaustive evaluation [Granados-Muñoz et al., 2014], a comparison with BSCDREAM8b skills in terms of dust vertical distribution have been performed at Granada using LIRIC retrievals during an intensive period from June to August 2012 (246 profiles). The LIRIC vs. BSC-DREAM8b comparison has been done in terms of direct profiles (concentrations) and normalized profiles (shape), considering statistical parameters for the whole profile and each level. After a preliminary evaluation, it is found that the direct comparison mainly shows underestimations in the range 25-125 μg/m3, slopes between 0 and 0.6, and only 55% cases with R>0.6. It is also found that BSCDREAM8b better reproduces the vertical layering. After normalization, 48% of profiles show mean relative deviation + 20% and slopes are closer to unity. BSC-DREAM8b better reproduces the profile shape between 2 and 4 km, and shows some limitations below 2 and very large differences > 4 km.