Elucidating Hidden and Enduring Weaknesses in Dust Emission Modeling

• Published in: Journal of geophysical research. Atmospheres Vol. 128; no. 17
• Main Authors: Chappell, Adrian, Webb, Nicholas P., Hennen, Mark, Zender, Charles S., Ciais, Philippe, Schepanski, Kerstin, Edwards, Brandon L., Ziegler, Nancy P., Balkanski, Yves, Tong, Daniel, Leys, John F., Heidenreich, Stephan, Hynes, Robert, Fuchs, David, Zeng, Zhenzhong, Baddock, Matthew C., Lee, Jeffrey A., Kandakji, Tarek
• Format: Journal Article
• Language: English
• Published: Washington Blackwell Publishing Ltd 16-09-2023; American Geophysical Union
• Subjects: Aerosol effects; Albedo; Atmospheric models; Atmospheric particulates; Calibration; Climate; Climate models; Continental interfaces, environment; Dust; Dust control; Dust cycle; Dust emission; Earth; Emissions; Emissions control; Environmental changes; Estimates; Frequency dependence; Geophysics; Mathematical models; Modelling; Numerical models; Ocean, Atmosphere; Optical analysis; Optical thickness; Plant cover; Probability theory; Roughness; Satellite observation; Satellites; Sciences of the Universe; Statistical analysis; Vegetation; Vegetation cover; drag partition; Dust emission; vegetation; aerodynamic sheltering; MODIS; albedo
• ISSN: 2169-897X; 2169-8996
• DOI: 10.1029/2023JD038584

Large‐scale classical dust cycle models, developed more than two decades ago, assume for simplicity that the Earth's land surface is devoid of vegetation, reduce dust emission estimates using a vegetation cover complement, and calibrate estimates to observed atmospheric dust optical depth (DOD). Consequently, these models are expected to be valid for use with dust‐climate projections in Earth System Models. We reveal little spatial relation between DOD frequency and satellite observed dust emission from point sources (DPS) and a difference of up to 2 orders of magnitude. We compared DPS data to an exemplar traditional dust emission model (TEM) and the albedo‐based dust emission model (AEM) which represents aerodynamic roughness over space and time. Both models overestimated dust emission probability but showed strong spatial relations to DPS, suitable for calibration. Relative to the AEM calibrated to the DPS, the TEM overestimated large dust emission over vast vegetated areas and produced considerable false change in dust emission. It is difficult to avoid the conclusion that calibrating dust cycle models to DOD has hidden for more than two decades, these TEM modeling weaknesses. The AEM overcomes these weaknesses without using masks or vegetation cover data. Considerable potential therefore exists for ESMs driven by prognostic albedo, to reveal new insights of aerosol effects on, and responses to, contemporary and environmental change projections.
Mineral dust influences Earth's systems, and understanding its impacts relies on numerical models which include large uncertainties. We compared measurements of dust optical depth (DOD) frequency of occurrence (probability) and satellite observed dust emission frequency from point sources (DPS) across North America. We found up to 2 orders of magnitude difference between DOD probability and DPS probability. Compared with DPS probability, we found an exemplar traditional dust emission model (TEM) and the albedo‐based dust emission model (AEM) both overestimated dust emission probability by up to 1 order of magnitude with statistically significant relations, suitable for calibration. Relative to the AEM calibrated to DPS, the exemplar TEM overestimated large dust emission over vast vegetated areas and produced considerable false change in dust emission. Tuning dust cycle models to DOD has very likely hidden, for more than two decades, these TEM weaknesses with implications for our understanding of Earth's systems. Considerable potential exists for new insights of dust‐climate in Earth System Models by using AEM with prognostic albedo.
Tuning dust models to dust optical depth (DOD) hides dust emission model weaknesses including overestimates and false change in vegetated areas
New shadow‐shelter model calibrated to observed dust emission circumvents unrealistic model assumptions
Two orders of magnitude difference between DOD and observed dust emission without significant relation