TL;DR
This paper introduces two versions of a new modal aerosol module (MAM) for the Community Atmosphere Model (CAM5), aiming to balance complexity with computational efficiency for climate simulations. The study evaluates MAM7 (more comprehensive) and MAM3 (simplified for long-term simulations), finding that MAM3 can largely reproduce MAM7's results for key species like sulfate and SOA, while highlighting areas needing improvement in aerosol processes and cloud interactions.
What Problem Does This Paper Address?
Accurately representing aerosols in global climate models (GCMs) is crucial because they significantly influence Earth's radiative balance and climate. However, aerosols are complex, with diverse chemical compositions, size distributions, and short atmospheric lifetimes leading to large spatial variations. This complexity poses a challenge for GCMs, which need to balance detailed aerosol processes with the computational demands of long-term simulations. The authors address this by developing and evaluating two versions of a modal aerosol module (MAM): a detailed version (MAM7) and a simplified version (MAM3) designed for efficiency in long-term climate projections, aiming to capture the essential aerosol forcing on climate with minimal complexity.
Methods
The core of this paper is the development and evaluation of two versions of the Modal Aerosol Module (MAM):
- MAM7: A more comprehensive version with seven lognormal modes, explicitly treating aerosol size distribution, internal/external mixing, and various physical/chemical processes. It serves as a benchmark.
- MAM3: A simplified version with three lognormal modes, designed for computational efficiency in long-term simulations. Key simplifications include:
- Immediate mixing of primary organic matter (POM) and black carbon (BC) with other aerosol components.
- Merging of fine dust/sea salt modes into the accumulation mode.
- Merging of coarse dust/sea salt modes into a single coarse mode.
- Neglecting explicit ammonia/ammonium cycles.
The paper describes the implementation of various aerosol processes within these models, including emissions, gas-phase chemistry, SOA formation, nucleation, condensation, coagulation, water uptake, vertical transport, activation, and deposition. The evaluation involves comparing simulations from MAM3 and MAM7 against observations and other model results.
Datasets
The paper utilizes several datasets for emissions and evaluation:
- Emissions Data:
- Anthropogenic emissions: Lamarque et al. (2010) IPCC AR5 dataset.
- Black carbon (BC) and organic carbon (OC) emissions: Updates from Bond et al. (2007) and Junker and Liousse (2008).
- Sulfur dioxide (SO2) emissions: Update from Smith et al. (2001, 2004).
- Natural aerosols and precursor gases (volcanic SO2, sulfate, DMS, NH3, biogenic VOCs): Utilized AeroCom emission fluxes, injection heights, and size distributions. NH3 emissions from MOZART-4 data set (Emmons et al., 2010). Isoprene, monoterpenes, toluene, etc., from MOZART-2 data set (Horowitz et al., 2003).
- Observational Data for Evaluation:
- SO2 and sulfate concentrations: IMPROVE sites (US) and EMEP sites (Europe).
- Sulfate concentrations: Ocean sites operated by the University of Miami.
- BC and OC/OM concentrations: IMPROVE, EMEP sites, and compilations by Liousse et al. (1996), Cooke et al. (1999), and Zhang et al. (2007).
- BC vertical profiles: Aircraft measurements from AVE Houston, CR-AVE, TC4, CARB campaigns, and HIPPO campaign.
- Mineral dust concentrations and deposition fluxes: Mahowald et al. (2009).
- Sea salt concentrations: Ocean sites operated by the University of Miami.
- Aerosol number size distributions: Marine boundary layer observations from Heintzenberg et al. (2000).
- Aitken and accumulation mode particle vertical profiles: Punta Arenas, Chile, and Prestwick, Scotland.
- CCN number concentrations: Field experiments reported in Ghan et al. (2001).
- Aerosol optical depth (AOD) and single scattering albedo (SSA): AERONET data.
- Cloud properties (SWCF, LWCF, LWP, IWP, CLDTOT, CDNUMC, CINUMC, CCN): CERES, SSM/I, ISCCP, and HIRS data.
Key Insights and Ideas
- MAM3 is a viable simplification for long-term simulations: The simplified MAM3 model largely reproduces the behavior of the more complex MAM7 for key species like sulfate and SOA, suggesting it can capture essential aerosol forcing on climate with significantly reduced computational cost.
- Aerosol aging and hygroscopicity are critical for POM/BC: The treatment of POM and BC aging and their hygroscopicity significantly impacts their concentrations, especially at high latitudes. Sensitivity tests highlight the importance of these processes.
- Sea salt treatment leads to notable differences: Differences in the treatment of coarse-mode sea salt size distributions and mixing states between MAM3 and MAM7 result in substantial differences in sea salt burden and AOD over oceans.
- Cloud interactions are a major source of bias: The paper repeatedly points to biases in aerosol simulations being linked to issues in cloud processes, particularly wet scavenging and cloud water lifetime. This underscores the tight coupling between aerosols and clouds.
- Ammonia/ammonium treatment matters for sulfate: Explicitly treating ammonia and ammonium cycles in MAM7, unlike in MAM3, leads to differences in aqueous phase SO2 oxidation and thus sulfate production, especially in regions with higher NH3 concentrations.
Key Results
- Sulfate and SOA: Simulated mass concentrations are remarkably similar between MAM3 and MAM7 (~90% in accumulation mode).
- POM and BC: Differences are small globally (within 10%) due to POM's hygroscopic nature and wet removal. However, sensitivity tests show increases of several times at high latitudes with hydrophobic POM or slower aging.
- Mineral Dust: Global burden differs by about 10% between MAM3 and MAM7.
- Sea Salt: Global burden differs by 30-40% between MAM3 and MAM7, primarily due to differences in size ranges and standard deviations of log-normal distributions.
- Model Performance: The models qualitatively capture observed geographical and temporal variations. However, biases exist, including low simulated sulfate and mineral dust over oceans, low BC in the Arctic, and low global AOD.
- Computational Cost: CAM5 with MAM7 is ~30% slower than with MAM3. MAM3 is ~35% slower than using the prognostic bulk aerosol module (BAM).
Limitations and Future Work
- Model Biases: The paper acknowledges several biases, including:
- Underestimation of sulfate and mineral dust concentrations over oceans.
- Underestimation of BC concentrations in the Arctic.
- Low bias in global aerosol optical depth, especially in developing countries.
- Underestimation of Aitken mode number concentrations in remote marine boundary layers.
- Overestimation of BC in the free troposphere in tropics and subtropics.
- Too fast cloud water conversion to precipitation, leading to too short cloud water lifetime and thus too fast wet removal of aerosols.
- Future Work: The authors suggest several areas for improvement:
- Improvements in aerosol processes (emission fluxes, boundary layer nucleation).
- Refining aerosol properties (primary emission size, POM hygroscopicity).
- Better representation of cloud properties (liquid water content, cloud fraction) for improved wet scavenging.
- More accurate treatment of aerosol transport and scavenging in convective clouds.
- Separating POM and BC properties by source (biomass burning vs. fossil fuel).
- Improving SOA formation, partitioning, and aging.
- Including nitrate chemistry.
- Quantifying model sensitivity to approximations in aerosol modal parameters, mixing state, internal structure, shape, production, transformation, and loss processes.
- Developing a more integrated treatment of aerosol transport and scavenging by convective clouds.
- Implementing a double-moment cloud microphysics parameterization for deep convective clouds.
Summary updated