Effect of the Atlantic hurricanes on the oceanic meridional overturning

circulation and heat transport

Aixue Hu1 and Gerald A. Meehl1

Received 14 November 2008; accepted 31 December 2008; published 7 February 2009.

[1] Hurricanes have traditionally been perceived as intensebut relatively small scale phenomena, with little effect onthe large scale climate system. However, recent evidencehas suggested that hurricanes could play a much moresignificant role in global climate. Here we prescribe Atlantichurricanes in a global coupled climate model to show that,climatically, the strong hurricane winds can strengthen theAtlantic meridional overturning circulation (MOC) that isresponsible for an increased northward meridonal heattransport (MHT), and the hurricane rainfall tends to weakenthe MOC and to reduce the MHT. The net effect of thehurricanes on the MOC and MHT depends on the outcomeof these two competing processes. This result implies thathurricanes may indeed play an important role in the coupledclimate system and need to be studied further in highresolution global coupled models. Citation: Hu, A., and G. A.

Meehl (2009), Effect of the Atlantic hurricanes on the oceanic

meridional overturning circulation and heat transport, Geophys.

Res. Lett., 36, L03702, doi:10.1029/2008GL036680.

1. Introduction

[2] Recent observed evidence has suggested that hurri-canes could be connected to climate change [Emanuel,2001, 2005; Webster et al., 2005; Kossin et al., 2007;Landsea, 2005; Knutson et al., 2007]. Estimates based onobservations show that hurricanes can transfer a significantamount of momentum into the ocean [Liu et al., 2008], andthey can also cause the downward pumping of hugeamounts of heat into the subsurface ocean along their paths[Emanuel, 2001; Sriver and Huber, 2007; Pasquero andEmanuel, 2008]. These papers also raised the more funda-mental issue of what role hurricanes actually play in theglobal climate system and how hurricanes could affect thelong-term oceanic meridional overturning circulation(MOC) and the meridional heat transport (MHT) in theAtlantic.[3] There are some indications that hurricanes could be

contributing to global climate through connections to large-scale fluctuations of sea surface temperatures (SSTs) as wellas in possible influences on northward MHT and the oceanconveyor belt circulation in the Atlantic. For example thereis a linkage between hurricane counts and low-frequencyvariations of tropical SSTs [Emanuel, 2007; Mann andEmanuel, 2006; Holland and Webster, 2007]. These SSTscould be connected to the MOC which is a global scaleoceanic circulation that plays a very important role in the

redistribution of heat and freshwater globally on decadal,centennial, and millennium timescales [Broecker, 1998].The collapse and re-establishment of this circulation isthought to be the major cause of past abrupt climate changes,such as Heinrich events [Heinrich, 1988; Timmermann etal., 2005]. Many studies also show that the changes of thiscirculation may be important for future climate change[Manabe and Stouffer, 1988; Hu et al., 2004; Gregory etal., 2005; Broccoli et al., 2006; Stouffer et al., 2006; Meehlet al., 2007]. If hurricanes affect MHT or MOC, this couldbe a significant contribution to large scale fluctuations ofclimate. Without the associated precipitation effect, a coarseresolution ocean general circulation model coupled to azonal mean atmospheric model that included a parameter-ized hurricane ocean mixing effect did show increases inboth northward heat transport and meridional overturningdue to hurricanes under extremely high CO2 conditions[Korty et al., 2008].[4] Here, we focus on the potential climatic role of the

Atlantic hurricane-induced wind, as well as precipitation, onthe Atlantic MOC and MHT by comparing global coupledclimate model simulations with prescribed hurricanes in theAtlantic to a control simulation that does not resolvehurricanes. The state-of-art global coupled climate modelused here is the National Center for Atmospheric Research(NCAR) Community Climate System Model version 3(CCSM3) [Collins et al., 2006] that has been used forsimulations of 20th and 21st century climate and paleocli-mate assessed by the IPCC AR4. As a first attempt to tacklethis problem using a fully coupled model, many simplifi-cations are applied in our experiments. However the overallgood agreement of the hurricane effect on the Atlantictropical-subtropical oceanic stratification in the model com-pared to observations does suggest that our simulations mayhave captured the major effects of hurricanes on the AtlanticMOC and MHT.

2. Model and Experiments

[5] CCSM3 includes the Community Atmospheric Model(CAM3) at T42 resolution (roughly 2.8�) and 26 hybridlevels in the vertical, a version of the Parallel OceanProgram (POP) with 1� horizontal resolution and enhancedmeridional resolution (1/2�) in the equatorial tropics, andwith 40 vertical levels, the Community Sea Ice Model(CSIM5) with Elastic-viscous-plastic dynamics, a subgrid-scale thickness distribution, and energy conserving thermo-dynamics, and the Community Land Model (CLM3). Themean Atlantic MOC defined as the maximum value of theAtlantic meridional streamfunction below 500 meter depthis 19.5 Sv (Sv � 106 m3s�1) in CCSM3 for the 240-yearperiod of the control simulation analyzed here, agreeing

GEOPHYSICAL RESEARCH LETTERS, VOL. 36, L03702, doi:10.1029/2008GL036680, 2009ClickHere

for

FullArticle

1Climate and Global Dynamics Division, National Center for Atmo-spheric Research, Boulder, Colorado, USA.

Copyright 2009 by the American Geophysical Union.0094-8276/09/2008GL036680$05.00

L03702 1 of 6

well with recent observations [Ganachaud and Wunsch,2000]. The mean MHT in the model at 24�N is 1.02 PW(PW = 1015 W), a bit lower than the observational values of1.2 PW [Ganachaud and Wunsch, 2000; Talley et al., 2003].[6] Here we prescribe hurricanes in the tropical Atlantic

in the model (Figure 1a) and make comparisons with thecontrol simulation without prescribed hurricanes. First, onlyhurricane induced wind forcing is considered throughmodifying the model produced wind field. Historically overthe last 100 years, on average there are 10 hurricanes eachyear in the Atlantic [Landsea, 2007]. We use this meannumber of hurricanes as our base case where we only

prescribe winds from each of the climatological hurricanes(termed ‘‘hurricane wind’’, HW). The size of the hurricanesvaries from a radius of 200 km at the time of formation to aradius of 500 km at the end of the track which is the typicalsize of tropical cyclones [Liu and Chan, 1999]. Themaximum wind speed is 18 m/s initially and increases upto about 53 m/s, equivalent to a category 3 hurricane. InJune, July and October, the maximum strength of thehurricanes is category 2 with a maximum wind speed of43 m/s. The lifetime of the hurricanes varies from 7 to 8 daysin June, July, and October to a maximum of 11 days inAugust and September. In the second experiment, in addi-

Figure 1. (a) Idealized climatological hurricane tracks used in the model simulations. In our experiments, there is onehurricane in June, one in July, three in August, three in September, and two in October. Color lines indicate the differenthurricane tracks. (b) Changes of the daily mean vertical temperature profile east of Florida before (black), during (blue), theday after (red) hurricane passage, and the recovery (a week after hurricane passage) of the upper ocean temperature (green).The region east of Florida is defined as the area 25.5–26.5�N and 73–74�W which is shown as a blue hollow circle inFigure 1a. (c) Vertical profiles of the century mean temperature anomalies in the upper 1200 meters of the ocean for the twomodel experiments relative to the control simulation and observations from an active hurricane year (2005) minus aninactive year (2006) for the tropical-subtropical Atlantic from equator to 40�N. (d) Same as Figure 1c except for salinityanomalies. (e) Same as Figure 1c except for the subpolar North Atlantic (40�N–65�N). (f) Same as Figure 1e except forsalinity anomalies. (g) Same as Figure 1e except for density anomalies.

L03702 HU AND MEEHL: ATLANTIC HURRICANES AND MOC, MHT L03702

2 of 6

tion to the same hurricane wind forcing, the torrentialprecipitation associated with hurricanes (termed ‘‘hurricanewinds and rainfall’’, HWR) is also added. Precipitation ratesare derived from TRMM 3B42 3 hourly data [Huffman etal., 2001] and this additional freshwater flux is compensatedby a uniformly increased evaporation of an equivalentamount in the domain of the trade winds in the Atlanticbetween 20�S and 35�N. These additional fluxes are addedto the ocean model by modifying the same fluxes producedby the model. The exact water source for season-averagedhurricane rainfall is still an open question from the obser-vations, though there are some indications that individualhurricane can drawmoisture from a radius as large as 1600 km[e.g., Trenberth and Fasullo, 2007]. From a general circu-lation perspective, the largescale tropical Atlantic moistureconvergence must be balanced partly by precipitation, thussupporting our method of treating the freshwater balance inthe tropical Atlantic as covering the climatological domainof the trade winds.[7] These experiments are branched from a millennium

control run of the CCSM3 and are run for 100 years eachwith the same set of hurricanes repeating every year. Unlessotherwise explicitly stated, the numbers mentioned beloware century or 50-year mean anomalies which are used toassess the climatic effect of hurricanes on the MOC.

3. Results

[8] The effect of high winds associated with the imposedhurricanes is to cool the ocean surface and to warm thesubsurface water along the hurricane tracks (Figure 1b,daily averaged vertical temperature changes). This coolingeffect is caused by increased evaporation from the strongwinds, as well as the strengthened vertical mixing withcooler subsurface water mixed upward and warmer surfacewater mixed downward, leading to a deepening of thethermocline. For example, in an area east of Florida (theblue hollow circle in Figure 1a), a mature hurricane in ourmodel can cool the surface ocean by more than 2�C on dailyaverage and warm the subsurface water by the samemagnitude (Figure 1b), and this surface cooling can be asmuch as 6�C in some regions in the model. After thepassage of the hurricane, the surface water is warmed upby solar heating, and the subsurface water is still warmerthan before the passage of the hurricane (Figure 1b),resulting in a net downward heat transport in the upperocean, agreeing well with observations [e.g., Emanuel,2001; Sriver and Huber, 2007]. This feature appears inour model simulations in all the regions along the hurricanetracks with somewhat different magnitudes of the surfacecooling and the subsurface warming. The qualitative agree-ment of our model simulations with the observationsindicates our experiment setup captures the major effect ofhurricane winds on upper ocean stratification.[9] Climatically, the century mean surface temperature

anomaly in experiment HW relative to the control simula-tion without prescribed hurricanes shows a cooling up to1�C (Figure 2a). As the strong hurricane winds enhanceevaporation in HW, salinity increases by about 0.1 psu (upto 0.5 psu) in most areas of the tropical Atlantic (Figure 2c).In the tropical-subtropical region (Figure 1c, blue lines), theregional averaged century mean surface cooling is about

�0.1 to �0.15�C and the subsurface warming peaks around+0.3�C centered at 100 meters depth, qualitatively agreeingwith observations [Emanuel, 2001; Sriver and Huber,2007]. The salinity increases by up to 0.04 psu in the upper800 m ocean (Figure 1d). As this more salty water is carriednorthward, the regional mean upper ocean salinity increasesby about 0.02 psu in the subpolar North Atlantic (Figures 1fand 2c). It is mainly these increases of salinity that raise thedensity in the upper ocean in the model in that region(Figure 1g). When the upper ocean density increases, theMOC is intensified by 3% (significant at 99% level,Figure 3a), the northward heat transport increase is 2% at24�N (significant at 90% level, Figure 3b), and SSTs in thesubpolar North Atlantic warm by several tenths of a degree(Figure 2a).[10] A further breakdown of the contributions from

changes of the MOC and the changes of the oceanicstratification in the Atlantic to this MHT increase indicatesthat it is the strengthening of the MOC that leads to theincreased MHT. By comparing the MHT in the HW run andthe control run that both are calculated by using the controlrun’s MOC, the changes of the oceanic stratification inducea slightly reduced northward heat transport in the HW runcompared to the control run. This explains why the MOCchanges are larger than the MHT changes in the HW runrelative to those in the control run. Additional sensitivityexperiments show that as the hurricane winds strengthen orthe number of the hurricanes increases, the changes shownabove are also larger and stronger, implying that after aperiod of more active hurricane seasons, the MOC wouldspin up and transport more heat into the subpolar NorthAtlantic if only the hurricane wind effect is considered.[11] On the other hand, if the climatic effects of hurricane

rainfall are included, the pattern of surface cooling in thetropical-subtropical Atlantic is similar to HW case withslightly larger amplitude (Figure 2b) and tropical Atlanticsalinity shows mostly a net increase mainly from the regionsof enhanced evaporation in the eastern tropical Atlantic(Figure 2d). However, the freshwater from the hurricane-related precipitation contributes to a larger salinity decreasein the Gulf of Mexico and parts of the eastern tropicalAtlantic. This freshwater is advected to the subpolar NorthAtlantic lowering the upper ocean density there (Figures 1fand 2d). That contributes to a reduction of MOC by 5% anda decrease in MHT of 3% relative to the HW case (bothchanges are significant at 99% level, Figure 3) which leadsto cooler SSTs in the subpolar North Atlantic by up to 0.5�C(Figure 2b). Sensitivity experiments indicate that the neteffect of the hurricane winds and rainfall on MOC and MHTdepends on the source regions for where the freshwater forrainfall is evaporated. If the water vapor is from a largeregion relative to the size of the hurricane, the MOC wouldweaken more relative to the HW case and transport less heatnorthward (see auxiliary material).1 Although the imple-mentation of the hurricane rainfall and the source of theprecipitable water in our experiments is very simple andidealized, our results do indicate that hurricane winds tendto strengthen the MOC and hurricane rainfall tends toweaken it, and the net effect of hurricanes on the MOC

1Auxiliary materials are available in the HTML. doi:10.1029/2008GL036680.

L03702 HU AND MEEHL: ATLANTIC HURRICANES AND MOC, MHT L03702

3 of 6

Figure 3. (a) Time averaged percent changes (over the last 50-years) for the Atlantic MOC from the two modelexperiments. (b) Same as Figure 3a except for MHT at 24�N. Error bars are plus and minus one standard deviation of themulti-century control run. The MOC and MHT changes in HW relative to control simulation (HW–CON) are 97% and90% significant, and the changes in HWR relative to HW (HWR–HW) are both significant at 99% level, respectively,according to a student t-test.

Figure 2. (a) Century mean sea surface temperature anomalies (�C) from the hurricane winds-only experiment (HW)relative to the control run. (b) Same as Figure 2a except for the hurricane winds and precipitation experiment (HWR).(c) Same as Figure 2a except for salinity anomalies (psu). (d) Same as Figure 2b except for salinity anomalies (psu).Stippling indicates the anomalies are significant at greater than the 95% level based on a student t-test.

L03702 HU AND MEEHL: ATLANTIC HURRICANES AND MOC, MHT L03702

4 of 6

and MHT relies on the outcome of these two competingprocesses.[12] To put these model sensitivity results into context,

we compare the anomalies in the HWR run relative tocontrol run with the observed anomalies between a relativeactive hurricane year (2005) and a relatively inactivehurricane year (2006). Though this must necessarily notbe considered a quantitative comparison, the sign of thechanges should be indicative of whether the model isqualitatively capturing the sense of the changes hurricanescould induce in the climate system.[13] The observations show increased heat mixed into the

upper ocean in the tropical-subtropical Atlantic (Figure 1c,black lines) with a relative warming for the active hurricaneyear compared to the inactive year of nearly 0.1�C arounddepths of 50 to 100 m, while HWR also warms but withgreater amplitude. Meanwhile, observations for the activehurricane year show relative cooling of the subsurface in thesubpolar North Atlantic with values of �0.1�C near thesurface comparable to the cooling in that region in HWR(Figure 1e). This low amplitude cooling in observations andthe model simulation extends through the depth of the upperocean to below 1000m. For changes of salinity and densityin the subpolar North Atlantic (Figures 1f and 1d), the HWRsimulation agrees well with observations, such that theocean column is fresher and less dense, except the upper100 meters, but the HW simulation (less realistic becauserainfall effects are excluded) is opposite to the observations.[14] Further study indicates that the initial surface salinity

changes in HWR also show a salinity increase in this region.When the negative surface salinity anomaly from thesubtropics is transported into the subpolar North Atlanticin the second decade, it becomes fresher there. This sug-gests that the freshwater anomaly associated with morehurricanes in 2005 may not have arrived yet at the subpolarocean in 2006 with observations. This result may alsosuggest that, in reality, after a period of active hurricaneseasons, the cumulative anomalous freshwater flux due tohurricane rainfall would propagate into the subpolar NorthAtlantic, and lead to a weakening of the MOC.[15] After a period of inactive hurricane seasons, the

MOC may strengthen due to the lack of a cumulativehurricane rainfall induced negative salinity anomaly. Thesedelayed changes of the MOC could also feed back tohurricane activity by modulating the SSTs in the tropical-subtropical Atlantic region. This effect needs to be studiedfurther.

4. Conclusion and Discussion

[16] Results from our model sensitivity experimentsindicate that climatologically, Atlantic hurricane winds mixheat downward into the subsurface ocean and the associatedevaporation increases the surface salinity in the tropical-subtropical region. When this salinity anomaly is propagatedinto the subpolar North Atlantic, it destabilizes the oceanicstratification there and strengthens theMOCwhich transportsmore heat northward. On the other hand, the hurricanerainfall freshens the western subtropical Atlantic. When thisfreshwater anomaly reaches the subpolar North Atlantic, itstabilizes the oceanic stratification there and weakens theMOC, reducing the northward heat transport. The net effect

of the hurricane winds and rainfall on the Atlantic MOC andMHT depends on the outcome of these two competingprocesses. In spite of the limitations of our model, thesehurricane effects on the climate system are physically con-sistent from a process point of view. Although these clima-tological changes are not large, they do statisticallysignificant at least at 90% level, indicating that hurricanescould be an active agent in the mean climate system.[17] A caveat that must accompany this study is that

atmospheric dynamical effects of hurricanes are not includedsince the hurricanes are specified only as local surface windand precipitation changes. The vertical structure of the hurri-canes would modulate (interact with) the large scale atmo-spheric circulation which could add somewhat to the effectswe describe here.

[18] Acknowledgments. The authors thank Gokhan Danabasoglu andWilliam Large for their comments. And thank to Kerry Emanuel, GregHolland, and Zhengyu Liu for constructive discussions. The authors alsothank Brian Kauffman for his coding help. The objectively analyzed ARGOdata used here is provided by the Coriolis Data Center of France. We thankAiguo Dai for providing TRMM data in netCDF format. A portion of thisstudy is supported by the Office of Science (BER), U.S. Department ofEnergy, Cooperative Agreement DE-FC02-97ER62402. The National Cen-ter for Atmospheric Research is sponsored by the National ScienceFoundation. Correspondence and requests for materials should be addressedto Aixue Hu (ahu@ucar.edu).

ReferencesBroccoli, A. J., K. A. Dahl, and R. J. Stouffer (2006), Response of the ITCZto Northern Hemisphere cooling, Geophys. Res. Lett., 33, L01702,doi:10.1029/2005GL024546.

Broecker, W. S. (1998), Paleocean circulation during the last deglaciation:A bipolar seesaw?, Paleoceanography, 13, 119–121.

Collins, W. D., et al. (2006), The Community Climate System Model:CCSM3, J. Clim., 19, 2122–2143.

Emanuel, K. (2001), Contribution of tropical cyclones to meridional heattransport by the oceans, J. Geophys. Res., 106, 14,771–14,781.

Emanuel, K. A. (2005), Increasing destructiveness of tropical cyclones overthe past 30 years, Nature, 436, 686–688.

Emanuel, K. A. (2007), Environmental factors affecting tropical cyclonepower dissipation, J. Clim., 20, 5497–5509.

Ganachaud, A., and C. Wunsch (2000), Improved estimates of global oceancirculation, heat transport and mixing from hydrographic data, Nature,408, 453–457.

Gregory, J. M., et al. (2005), A model intercomparison of changes in theAtlantic thermohaline circulation in response to increasing atmosphericCO2 concentration, Geophys. Res. Lett., 32, L12703, doi:10.1029/2005GL023209.

Heinrich, H. (1988), Origin and consequences of cyclic ice rafting in thenortheast Atlantic Ocean during the past 130,000 years, Quat. Res., 29,143–152.

Holland, G. J., and P. J. Webster (2007), Heightened tropical cycloneactivity in the North Atlantic: Nature variability or climate trend?, Philos.Trans. R. Soc., Ser. A, 365, 2695–2716, doi:10.1098/rsta.2007.2083.

Hu, A., et al. (2004), Response of the Atlantic thermohaline circulation toincreased atmospheric CO2 in a coupled model, J. Clim., 17, 4267–4279.

Huffman, G. J., et al. (2001), Global precipitation at one-degree dailyresolution from multisatellite observations, J. Hydrometeorol., 2, 36–50.

Knutson, T. R., et al. (2007), Simulation of the recent multidecadal increaseof Atlantic hurricane activity using an 18-km-grid regional model, Bull.Am. Meteorol. Soc., 88, 1549–1565.

Korty, R. L., K. A. Emanuel, and J. R. Scott (2008), Tropical cyclone-induced mixing and climate: Application to equable climates, J. Clim.,21, 638–654.

Kossin, J. P., K. R. Knapp, D. J. Vimont, R. J. Murnane, and B. A. Harper(2007), A globally consistent reanalysis of hurricane variability andtrends, Geophys. Res. Lett., 34, L04815, doi:10.1029/2006GL028836.

Landsea, C. W. (2005), Hurricane and global warming, Nature, 438, E11–E13, doi:10.1038/nature04477.

Landsea, C. W. (2007), Counting Atlantic tropical cyclones back to 1900,Eos Trans. AGU, 88(18), doi:10.1029/2007EO180001.

Liu, K. S., and J. C. L. Chan (1999), Size of tropical cyclones as inferredfrom ERS-1 and ERS-2 data, Mon. Weather Rev., 127, 2992–3001.

L03702 HU AND MEEHL: ATLANTIC HURRICANES AND MOC, MHT L03702

5 of 6

Liu, L. L., W. Wang, and R. X. Huang (2008), The mechanical energy inputto the ocean induced by tropical cyclones, J. Phys. Oceanogr., 38, 1253–1266.

Manabe, S., and R. J. Stouffer (1988), Two stable equilibria of a coupledocean-atmosphere model, J. Clim., 1, 841–866.

Mann, M. E., and K. A. Emanuel (2006), Atlantic hurricane trends linked toclimate change, Eos Trans. AGU, 87(24), doi:10.1029/2006EO240001.

Meehl, G. A., et al. (2007), Climate Change 2007: The Physical ScienceBasis, Contribution of Working Group I to the Fourth Assessment Reportof the Intergovernmental Panel on Climate Change, edited by S. Solomonet al., Cambridge Univ. Press, Cambridge, U. K.

Pasquero, C., and K. Emanuel (2008), Tropical cyclones and transientupper-ocean warming, J. Clim., 21, 149–162.

Sriver, R. L., and M. Huber (2007), Observational evidence for an oceanheat pump induced by tropical cyclones, Nature, 447, 577 – 580,doi:10.1038/nature05785.

Stouffer, R. J., et al. (2006), Investigating the causes of the response of thethermohaline circulation to past and future climate changes, J. Clim., 19,1365–1387.

Talley, L. D., J. L. Reid, and P. E. Robbins (2003), Data-based meridionaloverturning streamfunctions for the global ocean, J. Clim., 16, 3213–3226.

Timmermann, A., U. Krebs, F. Justino, H. Goosse, and T. Ivanochko(2005), Mechanisms for millennial-scale global synchronization duringthe last glacial period, Paleoceanography, 20, PA4008, doi:10.1029/2004PA001090.

Trenberth, K. E., and J. Fasullo (2007), Water and energy budgets ofhurricanes and implications for climate change, J. Geophys. Res., 112,D23107, doi:10.1029/2006JD008304.

Webster, P. J., et al. (2005), Changes in tropical cyclone number, duration,and intensity in a warming environment, Science, 309, 1844–1846,doi:10.1126/science.1116448.

�����������������������A. Hu and G. A. Meehl, Climate and Global Dynamics Division,

National Center for Atmospheric Research, P.O. Box 3000, Boulder, CO80305, USA. (ahu@ucar.edu)

L03702 HU AND MEEHL: ATLANTIC HURRICANES AND MOC, MHT L03702

6 of 6