North African dust absorbs less solar radiation than estimated by models and remote-sensing retrievals

Adeyemi A. Adebiyi, Yue Huang, B. H. Samset, Jasper F. Kok2022 · DOI

TL;DR

New research reveals that current climate models and remote-sensing retrievals overestimate the solar radiation absorbed by North African dust by up to a factor of two. By leveraging in-situ aircraft measurements, the study finds that dust is less "absorbing" than previously assumed, suggesting a significant warm bias in our current understanding of dust's radiative effects on the global climate.

What Problem Does This Paper Address?

Mineral dust is a major player in Earth’s energy balance. It can reflect sunlight (cooling the planet) or absorb it (warming the atmosphere). North Africa is the world's largest source, accounting for over half of the global dust burden. However, quantifying exactly how much "shortwave" (solar) radiation this dust absorbs remains a massive challenge.

The primary metric for this is Absorption Aerosol Optical Depth (AAOD). If AAOD is high, the dust is dark and heat-absorbing; if it's low, the dust is more reflective. Currently, climate models and ground-based sensors (like AERONET) disagree on these values. This uncertainty stems from three poorly constrained microphysical properties: the dust's size distribution, its shape, and its "darkness" (quantified by the imaginary refractive index). This paper aims to reconcile these discrepancies to provide a more accurate picture of dust's role in climate warming.

Methods

The authors developed a framework to constrain dust absorption by working backward from direct observations.

  1. Optimization: They used an optimization method to find the imaginary refractive index that best matches a compilation of 14 in-situ (direct aircraft) measurements of Single-Scattering Albedo (SSA)—the ratio of scattering to total extinction—over North Africa.
  2. Accounting for Shape: Unlike most models that assume dust particles are perfect spheres (Lorenz-Mie theory), this study used a database that accounts for the complex, aspherical shapes of real dust.
  3. Size and Loading: They integrated these findings with the DustCOMM dataset, which provides observationally constrained estimates of dust mass loading and size distribution.
  4. Comparison: Finally, they compared their observationally-derived constraints against an ensemble of six selected climate models, eight AeroCom Phase III models, and retrievals from 23 AERONET stations.

Datasets

  • In-situ Measurements: 14 direct measurements of dust SSA and 12 measurements of size distribution from various field campaigns.
  • DustCOMM: A dataset combining aircraft data, satellite observations, and reanalysis to constrain dust size and mass.
  • AERONET: Data from 23 ground-based stations across North Africa.
  • Model Ensembles:
    • An ensemble of 6 climate models (including GISS, GEOS-Chem, and CESM).
  • An ensemble of 8 models from the AeroCom Phase III project.

Key Insights and Ideas

  • The "Darkness" Discrepancy: The most striking finding is that the actual imaginary refractive index of North African dust (0.0012) is less than half of what is typically assumed in climate models (~0.0026 to 0.0029).
  • The Masking Effect: Models suffer from two errors that partially cancel each other out. They overestimate how much the material itself absorbs (the refractive index), but they underestimate the number of large, coarse dust particles (which are more efficient absorbers). Because the "over-darkening" error is stronger, the net result is still a massive overestimation of total absorption.
  • AERONET Bias: Even AERONET, often considered the "gold standard" for ground-based aerosol data, appears to overestimate dust absorption. The authors suggest this is because AERONET's inversion algorithms may be mischaracterizing the size and refractive index of dust in these regions.
  • Climate Implications: Because models think dust is more absorbing than it actually is, they likely have a "warm bias." Correcting this could change our understanding of how dust influences everything from tropical cyclone intensification to North Atlantic sea surface temperatures.

Key Results

  • AAOD Overestimation: Climate models overestimate North African dust AAOD by up to a factor of two.
  • Regional Values: The study constrained the dust AAOD over the North African continent to approximately 0.0094, while AeroCom models estimated it at 0.0180.
  • Refractive Index: The mean imaginary refractive index at 550 nm was found to be 0.0012, significantly lower than the values used in the AeroCom ensemble.
  • Iron Content: Since iron oxides (hematite and goethite) drive dust absorption, the results imply that models likely overestimate the iron content in deposited dust, which has further implications for ocean biogeochemistry.

Limitations and Future Work

  • Geographic Scope: The study focuses specifically on North Africa. While this is the largest source, dust from other regions (like East Asia or the Middle East) may have different mineralogical compositions and absorption properties.
  • Spectral Range: The primary focus was on the 550 nm (mid-visible) wavelength. While the authors suggest these results can be scaled across the solar spectrum, more work is needed to confirm the spectral distribution of absorption.
  • Vertical Distribution: The paper notes that the impact of dust on clouds and precipitation depends heavily on its vertical position in the atmosphere, a factor that remains difficult for models to simulate accurately.

Summary updated

Abstract

Desert dust accounts for a large fraction of shortwave radiation absorbed by aerosols, which adds to the climate warming produced by greenhouse gases. However, it remains uncertain exactly how much shortwave radiation dust absorbs. We leverage in-situ measurements of dust single-scattering albedo to constrain absorption at mid-visible wavelength by North African dust, which accounts for approximately half of the world's dust. We find that models overestimate North African dust absorption aerosol optical depth (AAOD) by up to a factor of two. This occurs primarily because models overestimate the dust imaginary refractive index, the effect of which is partially masked by an underestimation of large dust particles. Additionally, similar factors contribute to an overestimation of AAOD retrieved by the ground-based Aerosol Robotic Network over North Africa. We conclude that the overestimation of simulated and retrieved dust absorption suggests substantial biases in current estimates of dust impacts on the Earth system, including a warm bias in dust radiative effects.
BibTeX
@misc{adebiyi2022north,
  author = {Adeyemi A. Adebiyi and Yue Huang and B. H. Samset and Jasper F. Kok},
  title = {{North African dust absorbs less solar radiation than estimated by models and remote-sensing retrievals}},
  year = {2022},
  doi = {10.31223/x5zh21},
}
Figure