Numerical modeling of Asian dust emission and transport with adjoint inversion using LIDAR network observations

Abstract

Abstract. A four-dimensional variational (4D-Var) data assimilation system for a regional dust model (RAMS/CFORS-4DVAR; RC4) is applied to a heavy dust event which occurred between 20 March and 4 April 2007 over eastern Asia. The vertical profiles of the dust extinction coefficients derived from NIES LIDAR observation network are directly assimilated. We conduct two experiments to evaluate impacts of selections of observation sites: Experiment A uses five Japanese observation sites located only downwind of dust source regions; the other Experiment B uses these sites together with two other sites near source regions (China and Korea). Validations using various observation data (e.g., PM10 concentration, MODIS AOT, OMI Aerosol Index, and the dust extinction coefficient derived by space-based LIDAR NASA/CALIPSO) are demonstrated. The modeled dust extinction coefficients are improved considerably through the assimilation. Assimilation results of Experiment A are consistent with those of Experiment B, indicating that observations of Experiment A can capture the dust event correctly and include sufficient information for dust emission inversion. Time series of dust AOT calculated by modeled and LIDAR dust extinction coefficients show good agreement. At Seoul, Matsue, and Toyama, assimilation reduces the root mean square errors of dust AOT by 31–32%. Vertical profiles of the dust layer observed by CALIPSO are also compared with assimilation results. The dense dust layer was trapped between θ=280–300 K and elevated higher toward the north; the model reproduces those characteristics well. The modeled dust AOT along the orbit paths agrees well with the CALIPSO dust AOT, OMI AI, and the coarse mode AOT retrieved from MODIS; especially the modeled dust AOT and the MODIS coarse mode AOT are consistent quantitatively. Assimilation results increase dust emissions over the Gobi Desert and Mongolia considerably; especially between 29 and 30 March, emission flux is increased by about 2–3 times. The heavy dust event is caused by the heavy dust uplift flux over the Gobi Desert and Mongolia during those days. We obtain the total optimized dust emissions of 57.9 Tg (Experiment A; 57.8% larger than before assimilation) and 56.3 Tg (Experiment B; 53.4% larger).

Assimilation of Both Column‐ and Layer‐Integrated Dust Opacity Observations in the Martian Atmosphere

“…Since MCS does not (in general) take data at the same levels as those used in the model, we need to pre-process the observed dust distribution. Our approach differs in approach from both Yumimoto et al (2008) and Sekiyama et al (2010), who both used the CALIPSO satellite LIDAR measurements of dust opacity in the Earth's atmosphere. Here, MCS dust retrievals are reported as dust opacities at atmospheric pressures typically 1-1.5 km apart, but their intrinsic vertical resolution is about 5 km (Kleinböhl et al, 2009), so the data set oversamples the actual MCS measurements.…”

“…Our approach differs in approach from both Yumimoto et al. ( 2008 ) and Sekiyama et al. ( 2010 ), who both used the CALIPSO satellite LIDAR measurements of dust opacity in the Earth's atmosphere.…”

“…Limited publications to date include the work of Yumimoto et al. ( 2008 ), who assimilated vertical profiles of the dust extinction coefficients in a regional dust transport model. In their study, the data from a ground‐based lidar network were interpolated to the vertical model levels for analyzing the model prognostic dust variables.…”

“…Few centers have assimilated dust profile observations, however, preferring instead to focus on achieving high horizontal resolution utilizing the much more abundant AOD measurements. Limited publications to date include the work of Yumimoto et al (2008), who assimilated vertical profiles of the dust extinction coefficients in a regional dust transport model. In their study, the data from a ground-based lidar network were interpolated to the vertical model levels for analyzing the model prognostic dust variables.…”