Spatial and temporal correlation length as a measure for the stationarity of atmospheric dust aerosol distribution

Abstract

Abstract Fields of dust aerosol optical depth (AOD) from numerical models and satellite observations are widely used data sets for evaluating the actual distribution of atmospheric dust aerosol. In this study we investigate the use of estimates of spatial and temporal correlation lengths (CLs) calculated from simulations using the regional model system COSMO-MUSCAT (COSMO: Consortium for Small-scale Modelling; MUSCAT: MUltiScale Chemistry Aerosol Transport Model) to characterize the spatial and temporal variability of atmospheric aerosol distribution, here mineral dust, and to provide an estimate on the temporal model output interval required in order to represent the local evolution of atmospheric dustiness. The CLs indicate the scales of variability for dust and thus provide an estimate for the stationarity of dust conditions in space and time. Additionally, CLs can be an estimate for the required resolution in time and space of observational systems to observe changes in atmospheric dust conditions that would be relevant for dust forecasts. Here, two years of dust simulations using COSMO-MUSCAT are analyzed. CLs for the individual years 2007 and 2008 are compared to the entire two-year period illustrating the impact of the length of time series on statistical analysis. The two years are chosen as they are contrasting with regard to mineral dust loads and thus provide additional information on the representativeness of the statistical analysis. Results from the COSMO-MUSCAT CL analysis are compared against CL estimates from satellite observations, here dust AOD inferred from IASI (Infrared Atmospheric Sounding Interferometer), which provides bi-daily information of atmospheric dust loading over desert land and ocean. Although CLs estimated from the satellite observations are at a generally lower level of values, the results demonstrate the applicability of daily observations for assessing the atmospheric dust distribution. Main outcomes of this study illustrate the applicability of CL for characterizing the spatio-temporal variability in atmospheric dustiness. This is in particular of interest for determining time intervals at which for example dust forecasts need to be provided. Results from this study further demonstrate that bi-daily satellite dust observations are sufficient for assessing the dust distribution over regions such as the Mediterranean region that are far from the dust sources.