Abstract. Mineral dust impacts key processes in the Earth system, including the radiation budget, clouds, and nutrient cycles. We evaluate dust aerosols in 16 models participating in the sixth phase of the Coupled Model Intercomparison Project (CMIP6) against multiple reanalyses and satellite observations. Most models, and particularly the multi-model ensemble mean (MEM), capture the spatial patterns and seasonal cycles of global dust processes well. However, large uncertainties and inter-model diversity are found. For example, global dust emissions, primarily driven by model-simulated surface winds, vary by a factor of 5 across models, while the MEM estimate is double the amount in reanalyses. The ranges of CMIP6 model-simulated global dust emission, deposition, burden and optical depth (DOD) are larger than previous generations of models. Models present considerable disagreement in dust seasonal cycles over North China and North America. Here, DOD values are overestimated by most CMIP6 models, with the MEM estimate 1.2–1.7 times larger compared to satellite and reanalysis datasets. Such overestimates can reach up to a factor of 5 in individual models. Models also fail to reproduce some key features of the regional dust distribution, such as dust accumulation along the southern edge of the Himalayas. Overall, there are still large uncertainties in CMIP6 models’ simulated dust processes, which feature inconsistent biases throughout the dust lifecycle between models, particularly in the relationship connecting dust mass to DOD. Our results imply that modelled dust processes are becoming more uncertain as models become more sophisticated. More detailed output relating to the dust cycle in future intercomparison projects will enable better constraints of global dust cycles, and enable the potential identification of observationally-constrained links between dust cycles and optical properties.

中文翻译：

---

CMIP6 模型模拟尘埃气溶胶的效果如何？

摘要。矿物粉尘影响地球系统中的关键过程，包括辐射收支、云和营养循环。我们根据多次再分析和卫星观测评估了参与耦合模型比对项目 (CMIP6) 第六阶段的 16 个模型中的尘埃气溶胶。大多数模型，尤其是多模型集合平均值 (MEM)，都能很好地捕捉全球沙尘过程的空间模式和季节周期。然而，发现了很大的不确定性和模型间的多样性。例如，主要由模型模拟的地表风驱动的全球粉尘排放在不同模型中相差 5 倍，而 MEM 估计值是再分析中的两倍。CMIP6 模型模拟的全球尘埃排放、沉积、负荷和光学深度 (DOD) 的范围比前几代模型更大。模型在华北和北美的沙尘季节周期方面存在很大差异。在这里，大多数 CMIP6 模型高估了 DOD 值，与卫星和再分析数据集相比，MEM 估计值高 1.2-1.7 倍。在单个模型中，这种高估可以达到 5 倍。模型也无法再现区域尘埃分布的一些关键特征，例如沿喜马拉雅山脉南缘的尘埃堆积。总体而言，CMIP6 模型模拟的沙尘过程仍然存在较大的不确定性，其特点是在模型之间的整个沙尘生命周期中存在不一致的偏差，特别是在连接沙尘质量与 DOD 的关系方面。我们的结果意味着，随着模型变得更加复杂，模拟的尘埃过程变得越来越不确定。