Mahowald et al 2006

Climate response and radiative forcing from mineral aerosols during the last glacialmaximum, pre-industrial, current and doubled-carbon dioxide climatesNatalie M. Mahowald 1,2 , Masaru Yoshioka 1,2 , William D. Collins 1 , Andrew J. Conley 1 ,David W. Fillmore 1 , and Danielle B. Coleman 11 National Center for Atmospheric Research, Boulder, Colorado, USA, 80307.2 Institute of Computational Earth System Science, University of California, SantaBarbara, Santa Barbara, California, USA, 93106.For submission to GRL (please note all submitted or in press papers are available onlinethrough web links in the reference list).AbstractMineral aerosol impacts on climate through radiative forcing by natural dust sources areexamined in the current, last glacial maximum, pre-industrial and doubled-carbon dioxideclimate. Modeled globally averaged dust loadings change by +88%, +31% and –60% inthe last glacial maximum, pre-industrial and future climates, respectively, relative to thecurrent climate. Model results show globally averaged dust radiative forcing at the top ofatmosphere is -1.0,-0.4 and +0.14 W/m 2 for the last glacial maximum, pre-industrial anddoubled-carbon dioxide climates, respectively, relative to the current climate. Globallyaveraged surface temperature changed by –0.85, -0.22, and +0.06 °C relative to thecurrent climate in the last glacial maximum, pre-industrial and doubled carbon dioxideclimates, respectively, due solely to the dust radiative forcing changes simulated here.

These simulations only include natural dust source response to climate change, andneglect possible impacts by human land and water use.1.0 IntroductionMineral aerosols both absorb and scatter shortwave and long wave radiation, andthus are thought to modify climate [e.g. Miller and Tegen, 1998]. Additionally, mineralaerosol deposition to high latitude ice cores are 2-20 times higher during cold periods,such as the last glacial maximum, than warm periods like today [e.g. Petit and al., 1999];and marine sediment records suggest that globally dust deposition is 3-4 times higherduring cold periods than warm periods [Rea, 1994]. Thus, the large expected changes indust deposition in response to climate can play an important role in modulating climate.Although climate models can include the impact of mineral aerosols on climate[e.g. Miller and Tegen, 1998], there have not been published simulations of the climateresponse of mineral aerosols under different climate regimes using realistic dustdistributions. Additionally, there have been limited estimates of radiative forcing underdifferent climates [Claquin, et al., 2002; Woodward et al., 2005]. Here, we include bothestimates of radiative forcing for four different time periods, as well as the climateresponse to those dust radiative forcings in the National Center for AtmosphericResearch’s (NCAR) Community Climate System Model (CCSM3) [Collins et al., inpress-a]. We include only the climate response to ‘natural’ sources of desert dust,because current estimates of the land use or water use impacts are not well constrained(see Mahowald, et al. [2005] for review).

2.0 Modeling methodologyClimate models simulations are conducted in the Community Atmospheric Model(CAM3) of the CCSM [Collins, et al., 2006]. The dust source, transport, depositionparameterizations used for these simulations is described briefly in the online supplementand in detail in Mahowald, et al. [2006] and is based on the Dust Entrainment andDeposition module [Zender, et al., 2003]. Radiative forcings and climate response arecalculated as described in Yoshioka, et al. [submitted, available athttp:www.cgd.ucar.edu/tss/staff/mahowald]; a brief description is given in the onlinesupplement. We use the slab ocean model version of the model to allow dust feedbacksonto the sea surface temperature, but this does not allow for changes in the oceancirculation from the dust. The simulations in this paper have exactly the same boundaryconditions and radiative forcing (except for dust) as the slab ocean model simulationsthey are an extension of, which are described in more detail in Kiehl, et al. [2006] andOtto-Bliesner, et al. 2006]. Please note that for the last glacial maximum and preindustrialsimulations, the slab ocean model heat fluxes are based on a fully coupledsimulation of these climates, [Otto-Bliesner, et al., 2006]. Simulations for the current anddoubled-carbon dioxide climate are based on the ocean heat fluxes of the fully coupledsystem in the current climate [Kiehl et al., 2006].The radiative forcing of dust is calculated online using just one year ofsimulations (interannual variability in loading is less than 10% in this simulation), asdescribed in more detail in Yoshioka et al. [submitted]. The default version of the CAM3includes a prescribed dust climatology for shortwave radiative forcing [Collins et al.,2006b]. For this paper, climate response is calculated relative to the case where this

forcing is turned off, described in Yoshioka et al. [submitted]. Longwave and shortwaveforcings from dust are included in these results. The radiative forcings reported in thispaper are estimated as instantaneous net forcings from both long wave and short wavewithout stratospheric adjustment.The sources of dust are allowed to change with climate based on changes invegetation calculated in the BIOME3 equilibrium vegetation model [Haxeltine andPrentice, 1996], following the methodology of Mahowald, et al. [1999]. We include theimpact of carbon dioxide on vegetation in modifying the sources for the simulationsincluded here [see Mahowald et al., 2006 for more cases]. These cases are called SOMB,SOMBLGMC, SOMBPIC and SOMBDC for the BIOME3 vegetation based simulationsfor the current climate (SOMB, B meaning using the BIOME3 output), last glacialmaximum (LGM), pre-industrial (PI) and doubled-carbon dioxide climates (D),respectively (the C at the end of the acronym means that the carbon dioxide fertilizationeffect is included in the vegetation change). We also include a case where the radiativeforcing was examined in more detail, compared to other models and observations[Yoshioka et al., submitted], using the default vegetation in the model (SOM). For the lastglacial maximum, we include a case where glaciogenic sources are included and tuned tomatch observations (SOMBLGMT). We deduced the size of these sources using anoptimization algorithm and available observations (terrestrial sediment, ice and marinesediment cores), and use those source areas for this study. These simulations, includingcomparisons to observations (concentration, deposition and aerosol optical depth) aredescribed in more detail in Mahowald et al. [2006].

3.0 ResultsA comparison of the results of this study to available observations is given inMahowald et al. [2006]. The climate response of dust to these different climates includesthe impact of changes in vegetation (from carbon dioxide fertilization, precipitation,temperature and insolation changes), surface winds, precipitation and soil moisture. Thedust response to source area is much stronger than to surface winds or precipitation inthis model, as described in Mahowald et al. [2006]. The globally averaged dust sourcechanges by +105%, +164%, +51% and –64% relative to the current climate (SOMB) forthe SOMBLGMC, SOMBLGMT, SOMBPIC, and SOMBDC cases, respectively, whendust direct radiative feedback is included in the models (Table 1). The latitudinaldistribution of the sources expands to northern high latitudes under last glacial maximumconditions, similar to previous studies [e.g. Mahowald et al., 1999]. Notice that addingthe impact of the glaciogenic sources on dust (and tuning the model to best match theobservations) increases our dust source and loading in the extra-tropics in the last glacialmaximum in our simulations (SOMBLGMT vs. SOMBLGMC). Atmospheric desert dustaerosol optical depth (500 nm) changes dramatically during the different climates due tothe changes in source strength (Figure 1a). Globally averaged dust aerosol optical depth(AOD) change by +81%, +88%, +31% and –60% in the SOMBLGMC, SOMBLGMT,SOMBPIC and SOMBDC cases relative to the SOMB case, respectively (Table 1). Theresponse of dust sources and deposition to climate change is discussed in more detail andcompared to available observations in Mahowald et al. [2006]. Note that the dust sourcestrength and loadings in this study will differ slightly from that in Mahowald et al. [2006]

ecause we are reporting the results from simulations including the radiative feedbacks ofprognostic dust, while those results were without radiative feedback of predicted dust.Table 1: Dust impacts on climateSOM SOMB SOMLGMC SOMLGMT SOMBPIC SOMBDCDust Source (Tg/year) 4650 5320 10880 14020 8040 1930Dust AOD (500 nm) 0.039 0.053 0.096 0.099 0.069 0.021TOA RF (W/m 2 ) -0.46 -0.43 -0.96 -1.47 -0.87 -0.29Sfc RF (W/m 2 ) -0.60 -0.56 -1.28 -1.59 -1.03 -0.36Change from dust radiative feedbacksSfc. T (°K) -0.20 -0.19 -1.00 -1.04 -0.41 -0.13Precip (mm/day) -0.035 -0.033 -0.095 -0.091 -0.049 -0.017Change relative to current from dust radiative feedbacksTOA RF (W/m 2 ) -0.53 -1.04 -0.43 0.14Sfc RF (W/m 2 ) -0.72 -1.02 -0.47 0.20Sfc. T (°K) -0.81 -0.85 -0.22 0.06Dust radiative forcing changes with climate and the amount of dust, and theradiative forcings are highest at the latitudes with the most dust, although the relationshipis not strictly linear (Figure 1a, 1b and 1c). The globally averaged largest radiativeforcings are seen in climates with the largest change in aerosol optical depth (Table 1).Our top of atmosphere radiative forcing tend to have larger longwave and smaller shortwave components than previous modeling studies, although these results are consistentwith available observations, as discussed in detail in Yoshioka et al. [submitted], andbriefly in the online supplement. The case where we consider the impact of glaciogenicsources in the LGM has higher dust loading than without these sources (case

SOMBLGMT vs. SOMBLGMC in Figure 1a) and a larger magnitude of radiative forcing(Figure 1c and d). If we look in more detail at the radiative forcings, we see that thisincrease is a balance between large negative radiative forcings over ocean, small negativeforcings over land and positive radiative forcings over the ice sheets (see onlinesupplement for spatial plots). Even though adding glaciogenic sources increases our dustover ice sheets, it increases the dust loading more over oceans, and thus is more negative(SOMLGMT vs. SOMLGMC—see online supplement for figures). More discussion onthe comparison of our radiative forcing results to published results is in the onlinesupplement.The climate response to dust can occur at different latitudes than the radiativeforcing. Here we show the difference in surface temperature (radiative) between the casewhere dust radiative forcing is included in the climate simulation minus the case wherethere is no radiative forcing, for each climate. The response of surface temperature todust radiative forcing tends to be between 0.1 and 2.1 °C cooling, and these values arestatistically significant across a broad range of latitudes (Figure 1d). The strongestcooling occurs in northern midlatitudes (Figure 1d), even when the dust optical depth andradiative forcing has a larger magnitude in the tropics (Figure 1a, b, and c). The strongestresponse in temperature occurs at 40-90 °N in the last glacial maximum cases, and isapproximately 2 °C. It is unclear why the response should be largest at northernmidlatitudes, but it could be due to the lower heat capacity of land versus ocean, and thelarge portion of land at this latitude.The climate response of precipitation to the dust radiative forcing tends to be ashift in the precipitation to the hemisphere with less dust (Figure 1e) and a decrease in the

globally averaged precipitation (Table 1). Our results suggest that further study of theimportance of climate feedbacks of desert dust aerosols is warranted, and suggests somepatterns that are robust across different climates in our simulations.4.0 DiscussionRadiative forcings and climate feedbacks from atmospheric desert dust for lastglacial maximum, pre-industrial, current and doubled-carbon dioxide climates arecalculated for the first time in one single modeling framework. Included in these dustscenarios are changes in dust sources due to changes in vegetation using the BIOME3equilibrum vegetation model [Haxeltine and Prentice, 1996] and glaciogenic sources inthe last glacial maximum. We include in the model only the response due to vegetationchanges, soil moisture and surface winds in this simulation, and ignore possible changesdue to human land or water use [e.g. Mahowald and Luo, 2003; Mahowald et al, 2005];estimates of anthropogenic radiative forcing and climate feedbacks including land use aregiven in the online supplement, based on previous studies. Comparisons of the dustdistribution in this study to available observations is conducted in a separate paper[Mahowald et al., 2006]; and shows a good performance of the model. The netinstantaneous top-of-atmosphere radiative forcing differences due to dust between the lastglacial maximum, pre-industrial and doubled-carbon dioxide climates and the currentclimate are –0.53, -0.43, and +0.14 W/m 2 , respectively. If we include the impact ofglaciogenic sources, the net top-of-atmosphere radiative forcing difference between thelast glacial maximum and the current climate increases in magnitude to –1.04 W/m 2 . Inthe future we simulate a 0.14 W/m 2 increase in radiative forcing because of a reduction indust from carbon dioxide fertilization of the vegetation. It is uncertain that the carbon

dioxide fertilization effect will continue in the future, when other nutrients may becomelimiting [Smith et al., 2000; Schlesinger and Lichter, 2001]. Indeed other studies includethe impact of climate induced desertification and obtain a significant increase in desertdust in the future [Woodward et al, 2005] while others simulate little change [Tegen etal., 2004], suggesting that more studies of vegetation and climate interactions are vital topredicting future dust.There is roughly a linear relationship between annually averaged aerosol opticaldepth and radiative forcing under different climates in the model simulations (Table 1and see Online Supplement for figure); the correlation coefficients are –0.96 and –0.92and the slopes are –14.7 and –13.0 W/m 2 / AOD, for top-of-atmosphere and surfaceradiative forcing, respectively. Globally averaged surface temperature changed by –0.85, -0.22, and +0.06 °C relative to the current climate in the last glacial maximum, preindustrialand doubled carbon dioxide climates, respectively, due solely to the dustradiative forcing changes simulated here. The response in surface temperature orprecipitation to dust forcing is approximately linear in this model in these differentclimates, with correlation coefficients of -0.94 and –0.98, respectively (Table 1 and seeOnline Supplement for figure). The slope in the global response is –12.7 °C/AOD and -1.0 mm/day/AOD for surface temperature and precipitation, respectively. These valuesshould be considered tentative, due to both the shortness of these simulations, and thedependence of these values on specific model characteristics. However, that the responseremains linear while the surface albedo changes so strongly, suggests a robust response ofsurface temperature and precipitation to aerosol optical depth with this set of opticalproperties. More analysis of the differences between dust impacts on climate in this

model and other models are discussed in Yoshioka et al. [submitted]. The sensitivity ofdust feedback results to optical properties is explored in Miller et al. [2004]. Theseresults will ignore the impact of dust radiation on changes in ocean circulation, since weuse a slab ocean model. We explore the efficacy of dust forcing within our model in theonline supplement.In the last glacial maximum, our results suggest a radiative forcing relativeto the current climate of –1.04 W/m 2 at the top of atmosphere, for the case includingglaciogenic sources and best matches available deposition observations. Compared to theforcings from carbon dioxide (-1.7 W/m 2 ) and insolation and albedo (-5.2 W/m 2 ) [e.g.Hewitt and Mitchell, 1997], the dust forcings are about 15% of the total of other forcings.The surface temperature response from dust is –0.85 ºC relative to the current climate,while the changes in carbon dioxide, insolation and albedo in the same simulations causea change in surface temperature of 5.6 ºC.Despite the possible sensitivity of the results to our model specifications, ourresults suggest some interesting relationships across the different climate studies.Radiative forcing at the top-of-atmosphere and surface is linear with aerosol opticaldepth, even in different climates. Climate response in surface temperature andprecipitation are roughly linear with aerosol optical depth in our model, with a decreasein both surface temperature and precipitation associated with increasing optical depth.Finally, our model predicts statistically significant decreases in temperature at manylatitudes (not just close to the dust sources) when dust is added in the different climates,and a shift in precipitation from the northern part of the ITCZ to the southern part of theITCZ.

AcknowledgementsThis work was supported by NASA-IDS (NAG5-9671), NSF-Biocomplexity (OCE-9981398) and DOE (DE-FG02-02ER-63387). Simulations were conducted at theNational Center for Atmospheric Research, which is funded by the National ScienceFoundation. This manuscript was improved by comments from the anonymous reviewersand Ron Miller.ReferencesClaquin, T., et al. (2002), Radiative forcing of climate by ice-age atmospheric dust,Climate Dynamics, 20, 193-202.W. D. Collins, C. M. Bitz, M. L. Blackmon, G. B . Bonan, C. S. Bretherton, J. A. Carton,P. Chang, S. C. Doney, J. J. Hack, T. B. Henderson, J. T. Kiehl, W. G. Large, D.S. McKenna, B. D. Santer, and R. D. Smith (2006-a), The Community ClimateSystem Model: CCSM3, Journal of Climate, (19) 2122-2143.Collins, W. D., et al. (2006-b), The formulation and atmospheric simulation of theCommunity Atmopshere Model, CAM3., Journal of Climate, (19) 2144-2161.Haxeltine, A., and I. C. Prentice (1996), BIOME3: An equilibrium terrestrial biospheremodel based on ecophysiological constraints, resource availability, andcompetition among plant functional types, Global Biogeochemical Cycles, 10,693-709.

Hewitt, C., and J. Mitchell (1997), Radiative forcing and repsonse of a GCM to ice ageboundary conditions: cloud feedback and climate sensitivity, Climate Dynamics,13, 821-834.Kiehl, J., C. Shields, J. Hack, and W.Collins (in press), The climate sensitivity of theCommunity Climate System Model: CCSM3, Journal of Climate (11) 2584-2596.Mahowald, N., A. Baker, G. Bergametti, N. Brooks, R. Duce, T. Jickells, N. Kubilay, J.Prospero, and I. Tegen (2005), The atmospheric global dust cycle and iron inputsto the ocean, Global Biogeochemical Cycles, 19, GB4025,10.1029/2004GB002402, 2005.Mahowald, N., K. Kohfeld, M. Hansson, Y. Balkanski, S. P. Harrison, I. C. Prentice, M.Schulz, and H. Rodhe (1999), Dust sources and deposition during the last glacialmaximum and current climate: A comparison of model results with paleodatafrom ice cores and marine sediments, Journal of Geophysical Research, 104,15,895-815,916.Mahowald, N. M., and C. Luo (2003), A less dusty future? Geophysical Research Letters,30, 1903, doi: 1910.1029/2003GRL017880.Mahowald, N., D. Muhs, S. Levis, P. Rasch, M. Yoshioka, and C. Zender (2006), Changein atmospheric mineral aerosols in response to climate: last glacial period, preindustrial,modern and doubled-carbon dioxide climates, Journal of GeophysicalResearch (111) D10202, doi:10.1029/2005JD006653.Mahowald, N., G. Rivera, and C. Luo (2004), Comment on "Relative importance ofclimate and land use in determining present and future global soil dust emission",Geophysical Research Letters, 31, 10.1029/2004GL021272.

Miller, R., and I. Tegen (1998), Climate Response to Soil Dust Aerosols, Journal ofClimate, 11, 3247-3267.Miller, R. I. Tegen, J. Perlwitz, Surface radiative forcing by soil aerosols and thehydrological cycle, J. Geophysical Res. (109)D04203/doi:10.1029/2003JD004085, 2004.Otto-Bliesner, B., E. Brady, G. Clauzet, R. Tomas, S. Levis, and Z. Kothavala Lastglacial maximum and holocene climate in CCSM3, Journal of Climate (11),2526-2544.Petit, J. R., and e. al. (1999), Climate and atmospheric history of hte past 420,000 yearsfrom teh Vostok Ice core, Antarctica, Nature, 399, 429-436.Rea, D. K. (1994), The Paleoclimatic record provided by eolian deposition in the deepsea: The geologic history of wind, Reviews of Geophysics, 32, 159-195.Schlesinger, W. H., and J. Lichter (2001), Limited carbon storage in soil and litter ofexperimental forest plots under increased atmospheric CO 2 , Nature, 411, 466-469.Smith, S., T. Huxman, S. Zitzer, T. Charlet, D. Housman, J. Coleman, L. Fenstermaker, J.Seemann, and R. Nowak (2000), Elevated CO2 increases productivity andinvasive species success in an arid ecosystem, Nature, 408, 79-82.Tegen, I., M. Werner, S. P. Harrison, and K. E. Kohfeld (2004), Relative importance ofclimate and land use in determining present and future global soil dust emission,Geophysical Research Letters, 31, L05105, doi:05110.01029/02003GL019216,012004.

Woodward, S., D. Roberts, and R. Betts (2005), A simulation of the effect of climatechanged-induced desertification on mineral dust aerosol, Geophysical ResearchLetters, 32, L18810, doi:18810.11029/12005GL023482.Yoshioka, M., N. Mahowald, A. Conley, W. Collins, D. Fillmore, C. Zender, and D.Coleman, Impact of dust radiative forcing on Sahel Precipitation, Journal ofClimate (submitted—available at www.cgd.ucar.edu/tss/staff/mahowald).Zender, C., H. Bian, and D. Newman (2003), Mineral Dust Entrainment and Deposition(DEAD) model: Description and 1990s dust climatology, Journal of GeophysicalResearch, 108, 4416.Figure captionsFigure 1: Zonally averaged values for the SOM, SOMB, SOMBLGMC, SOMBLGMT,SOMBPIC and SOMBDC cases for a) dust optical depth (500 nm), b) dust netradiative forcing at the top of atmosphere (TOA), c) dust net radiative forcing atthe surface (SFC), d) surface temperature (radiative) response (only valuessignificant at 95% are shown) and e) precipitation response (only valuessignificant at 95% are shown).

Dust aerosol optical depthLatitudeSurface radiative forcing (W/m2)Top-of-Atmosphere radiative forcing (W/m2)Change in surface temperature (C)Change in precipitation (mm/day)Latitude

Surface Temperature response from dustSurface Temperature (K)Years of simulation

RF (W/m2)AOD vs. RFAOD vs. TsAOD vs. Precip0.00.00.0-0.2-0.02-0.5-1.0-0.4-0.6-0.04-0.06-1.5-0.8-0.08-1.0-0.100.0 0.04 0.08 0.12AOD0.0 0.04 0.08AOD0.12 0.0 0.04 0.08AOD0.12Ts (C)Precip (mm/day)