Quantification of the dust optical depth across spatiotemporal scales with the MIDAS global dataset (2003–2017)

Abstract. Quantifying the dust optical depth (DOD) and its uncertainty across spatiotemporal scales is key to understanding and constraining the dust cycle and its interactions with the Earth System. This study quantifies the DOD along with its monthly and year-to-year variability between 2003 and 2017 at global and regional levels based on the MIDAS (ModIs Dust AeroSol) dataset, which combines MODIS-Aqua retrievals and MERRA-2 reanalysis products. We also describe the annual and seasonal geographical distributions of DOD across the main dust source regions and transport pathways. MIDAS provides columnar mid-visible (550 nm) DOD at fine spatial resolution (0.1° × 0.1°), expanding the current observational capabilities for monitoring the highly variable spatiotemporal features of the dust burden. We obtain a global DOD of 0.032 ± 0.003 – approximately a quarter (23.4 % ± 2.4 %) of the global AOD – with about one order of magnitude more DOD in the northern hemisphere (0.056 ± 0.004; 31.8 % ± 2.7 %) than in the southern hemisphere (0.008 ± 0.001; 8.2 % ± 1.1 %) and about 3.5 times more DOD over land (0.070 ± 0.005) than over ocean (0.019 ± 0.002). The northern hemisphere monthly DOD is highly correlated with the corresponding monthly AOD (R2 = 0.94) and contributes 20 % to 48 % of it, both indicating a dominant dust contribution. In contrast, the contribution of dust to the monthly AOD does not exceed 17 % in the southern hemisphere, although the uncertainty in this region is larger. Among the major dust sources of the planet, the maximum DODs (~1.2) are recorded in the Bodele Depression of the northern Lake Chad Basin, whereas moderate-to-high intensities are encountered in the Western Sahara (boreal summer), along the eastern parts of the Middle East (boreal summer) and in the Taklamakan Desert (spring). Over oceans, major long-range dust transport is observed primarily along the Tropical Atlantic (intensified during boreal summer) and secondarily in the North Pacific (intensified during boreal spring). Our calculated global and regional averages and associated uncertainties are consistent with some but not all recent observationally based studies. Our work provides a simple, yet flexible method to estimate consistent uncertainties across spatiotemporal scales, which will enhance the use of the MIDAS dataset in future studies.