全球变化知识资源中心

Tropical cyclones (TCs) represent one of the most destructive natural weather phenomena, with an annual global occurrence of about 90 events². As a large and complex convective system associated with enormous surface enthalpy fluxes, the frequency and intensity of TCs are regulated by several environmental conditions, including sea surface temperature (SST), vertical wind shear, vorticity, and humidity of the free troposphere⁶. Both observational and modelling studies have assessed changes in the frequency and intensity of TCs in a warming world, but the results often conflict because of limitations in the global historical records and uncertainties in global climate models. For example, although regional and global data analyses indicate a trend of more frequent and intense TCs during recent decades^{1, 2}, explicit and downscaled simulations using global climate models often project decreases in the globally averaged TC frequency because of greenhouse gas warming⁴. At present, the detection of long-term TC trends and their attribution to the rising level of greenhouse gases remain uncertain⁴.

In addition, atmospheric aerosols may also impact oceanic cyclones by modifying thermodynamic and microphysical conditions^{5, 7, 8, 9, 10, 11}. The warming effect due to accumulating greenhouse gases along with the cooling effect exerted by anthropogenic aerosols has been linked to long-term trends in tropical Atlantic warmth and increasingly stronger hurricanes in recent decades¹². Satellite analysis suggests that cooling over the tropical North Atlantic by elevated aerosols suppressed TC activities in the western Atlantic and the Caribbean during the 2006 hurricane season¹³. Recent studies of mineral dust reveal a large influence from the Sahara Desert on the formation and development of Atlantic hurricanes by modulating the cloud hydrometeor contents, diabatic heating distribution, and thermodynamic structure^{14, 15}.

The microphysical role of aerosols acting as cloud condensation nuclei (CCN) and altering TCs has been examined in several numerical studies. Cloud-resolving model simulations indicate that invigorated convection contributes to increased lightning activity at the periphery of TCs, but weakened convection and intensity in the eyewall region^{16, 17, 18}. Moreover, the plausible impact of submicron CCN seeding may suppress warm rain and weaken TCs as a result of lower-level evaporative cooling of unprecipitated raindrops¹⁹. However, few studies have examined the coupled microphysical and radiative effects of aerosols by considering the microphysical–radiative-dynamic interaction for TCs, even though aerosol-induced cooling of SST and the consequent feedback on the cyclogenesis in the ocean–atmosphere system have been suggested^{7, 13}.

In this study we quantify and isolate the microphysical and radiative effects of anthropogenic aerosols on TCs. Direct and semi-direct aerosol radiative forcings are evaluated by means of a radiative module interacting with explicit cloud microphysics^{20, 21}. Three aerosol scenarios are simulated in our numerical experiments (Methods and Supplementary Fig. 1), representing a clean maritime case (C-case), a polluted case (P-case), and a polluted case with the aerosol radiative forcing (PR-case) from light absorbing aerosols. The aerosol number concentrations in the P-case and PR-case are identical and are five times higher than that in the C-case, and the PR-case contains 5% of black carbon internally mixed with ammonium sulphate. Those aerosol properties are consistent with atmospheric field measurements throughout the Gulf of Mexico region^{22, 23} and satellite observations of maritime conditions under the influence of continental pollution^{9, 10}.

The TC intensity is commonly represented by the maximum surface wind speed and minimum surface pressure. Although all simulations reproduce the typical features of the TC evolution (Fig. 1), namely an intensification stage (<48 h) and a dissipation stage (>48 h), the simulated TC exhibits distinct development patterns under the different aerosol scenarios (Supplementary Figs 2–4). For example, upon reaching the maximal intensity at 48 h, the peak maximum surface wind speed is lower whereas the nadir minimum surface pressure is higher under the two polluted cases, both indicating a weakened TC. Also, the simulated TC in the two polluted cases exhibits noticeably delayed intensification, as evident in the slower increase (decrease) in the maximum surface wind speed (minimum surface pressure). The occurrences of the peak maximum surface wind speed and the nadir minimum surface pressure are delayed by about 10 h in the two polluted cases. The maximum surface wind speed and minimum surface pressure are comparable between the P-case and PR-case, except that the PR-case shows earlier dissipation. On the other hand, the domain average amount of precipitation is much larger in the two polluted cases than that in the clean maritime case. The largest difference in precipitation between the clean and polluted cases occurs after the TC reaches its peak intensity at 48 h. A comparison between the P-case and PR-case indicates that the aerosol radiative effect contributes to further enhanced precipitation.

Minimum surface pressure (a), maximum wind speed (b) and precipitation rate (c) for the three different aerosol scenarios: a clean maritime case (C-case) in blue, a polluted case (P-case) in dark red and a polluted case with aerosol radiative forcing (PR-case) in green. Error bars (one s.d.) are estimated from five ensemble simulations with different initial random temperature perturbations.

A two-moment bulk microphysical scheme under the framework of the WRF model (Version 3.1.1), initially developed by Li et al.²⁰, is employed in this study to assess the effects of aerosols. The microphysical scheme calculates the time-dependent mass-mixing ratios and the number concentrations of five types of hydrometeors—cloud water, rain water, ice crystals, snow flakes, and graupels—along with the aerosol mass-mixing ratio, surface area and number concentration. To consider the radiative effect of aerosols, a module is incorporated into the Goddard Shortwave Radiation Scheme to interact with aerosol and cloud microphysics and calculate online the wavelength-dependent aerosol optical properties, including aerosol optical depth, asymmetry factor and single scattering albedo (SSA; ref. 21).

We perform numerical simulations relevant to the conditions of Hurricane ‘Katrina’ from 00:00 UTC 27 August 2005 to 00:00 UTC 30 August 2005 using a two-way nested grid of the WRF model. A 9-km outer domain with a nest on a 3-km mesh is integrated (see Supplementary Fig. 1a). Simulations are initialized from six-hourly NCEP Final Operational Global Analysis (1°× 1°) and use the NCEP global SST data. In our current cloud-resolving WRF model, no convective parametrization is employed in the two-mesh domain.

In this study, clean maritime and polluted continental aerosols are assumed to consist of both seasalt (NaCl) and ammonium sulphate ((NH₄)₂SO₄). CCN activation is treated using the Khöler theory^{20, 31}. A seasalt production scheme is included to generate NaCl aerosols over the ocean surface^{9, 10}. For the initial and boundary concentrations at the surface level, the concentrations of (NH₄)₂SO₄ aerosols are 200 and 1000 cm⁻³, respectively, under the clean maritime condition (C-case) and the polluted conditions (P-case and PR-case). All simulations are initialized using a vertical aerosol profile with the maximal number concentration at the surface and a decreasing concentration with altitude (Supplementary Fig. 1b). In the case of the aerosol radiative effect (PR-case), aerosols are assumed to contain an internal black carbon core surrounded by ammonium sulphate and the mass-mixing ratios are assumed to be 5% for black carbon and 95% for ammonium sulphate. The SSA of internally mixed black carbon and ammonium sulphate aerosols in this study is calculated to be about 0.9 at the mid-visible range throughout the simulations. Atmospheric field measurements^{22, 23} reveal that black carbon accounts for 5–10% of the aerosol content and the measured SSA ranges from 0.7 to 0.9 throughout the Gulf of Mexico region, including from shipboard (NOAA R/V Ronald H. Brown) measurements during the 2006 Gulf of Mexico Atmospheric Composition and Climate Study. The aerosol loading is about 7 μg m⁻³ from the field measurements, consistent with those in our P- and PR-cases. On the global scale, the mass fraction of black carbon in anthropogenic aerosols is about 4%, which is simulated in the global climate model³² using the latest emission inventory from the Intergovernmental Panel for Climate Change (IPCC). In addition, significantly elevated aerosol loading has been documented over the Pacific under the influence of Asian pollution outflows on the basis of aircraft and satellite measurements^{8, 9, 33}. For the three aerosol cases, fixed aerosol concentrations (as the initial values) are assumed on the domain boundaries, with advection of aerosols from the lateral boundaries into the model domain occurring under favourable wind conditions. The calculated aerosol optical depth is around 0.55 at the model boundaries and about 0.20 averaged over the inner domain, consistent with satellite measurements from the Moderate Resolution Imaging Spectroradiometer (MODIS) in the Gulf of Mexico region. Several processes are considered for ice nucleation, including deposition and immersion, contact, and homogeneous freezing^{20, 31}. The removal of aerosols includes activation to form cloud droplets and ice crystals, but precipitation scavenging is not included in the present simulations. It is estimated that less than 10% of aerosols entrained from the model boundary are transported into the eyewall region, whereas the remaining amount is washed out by activation. A fixed model boundary and domain are assumed throughout our simulations. The aerosol radiative effect in the long-wave radiation is negligible compared to that in the short-wave radiation²¹.

1. Emanuel, K. A. Increasing destructiveness of tropical cyclones over the past 30 years. Nature 436, 686–688 (2005).
   • CAS
   • ADS
   • ISI
   • PubMed
   • Article
2. Webster, P. J., Holland, G. J., Curry, J. A. & Chang, H. R. Changes in tropical cyclone number, duration, and intensity in a warming environment. Science 309, 1844–1846 (2005).
   • CAS
   • ISI
   • PubMed
   • Article
3. Holland, G. J. & Webster, P. J. Heightened tropical cyclone activity in the North Atlantic: Natural variability or climate trend? Phil. Trans. R. Soc. A 365, 2695–2716 (2007).
   • PubMed
   • Article
4. Knutson, T. R. et al. Tropical cyclones and climate change. Nature Geosci. 3, 157–163 (2010).
   • CAS
   • ADS
   • Article
5. Rosenfeld, D. et al. Aerosol effects on microstructure and intensity of tropical cyclones. Bull. Amer. Meteorol. Soc. 93, 987–1001 (2012).
   • ADS
   • Article
   URL: