Contribution of solar radiation to decadal temperaturevariability over landKaicun Wanga,1 and Robert E. Dickinsonb

aState Key Laboratory of Earth Surface Processes and Resource Ecology, College of Global Change and Earth System Science, Beijing Normal University, Beijing100875, China; and bDepartment of Geological Sciences, Jackson School of Geosciences, The University of Texas at Austin, Austin, TX 78712

Edited by Mark H. Thiemens, University of California, San Diego, La Jolla, CA, and approved August 2, 2013 (received for review June 18, 2013)

Global air temperature has become the primary metric for judgingglobal climate change. The variability of global temperature ona decadal timescale is still poorly understood. This paper examinesfurther one suggested hypothesis, that variations in solar radia-tion reaching the surface (Rs) have caused much of the observeddecadal temperature variability. Because Rs only heats air duringthe day, its variability is plausibly related to the variability of di-urnal temperature range (daily maximum temperature minus itsminimum). We show that the variability of diurnal temperaturerange is consistent with the variability of Rs at timescales frommonthly to decadal. This paper uses long comprehensive datasetsfor diurnal temperature range to establish what has been thecontribution of Rs to decadal temperature variability. It shows thatRs over land globally peaked in the 1930s, substantially decreasedfrom the 1940s to the 1970s, and changed little after that. Reduc-tion of Rs caused a reduction of more than 0.2 °C in mean temper-ature during May to October from the 1940s through the 1970s,and a reduction of nearly 0.2 °C in mean air temperature duringNovember to April from the 1960s through the 1970s. This coolingaccounts in part for the near-constant temperature from the 1930sinto the 1970s. Since then, neither the rapid increase in tempera-ture from the 1970s through the 1990s nor the slowdown ofwarming in the early twenty-first century appear to be signifi-cantly related to changes of Rs.

global dimming | global brightening | global warming |surface incident solar radiation | decadal variability

Global temperature has become the primary metric forjudging global climate change, although many other factors

are recognized to be of comparable importance. The overallincrease of global temperature over the last century has beenlargely attributed to the increase of greenhouse gases (1). Lesswell understood are the reasons for the variability of this increaseon a decadal timescale. In particular, warming from 1900 to 1940was followed by three decades of flat or slightly decreasingtemperature, then three decades of very rapid temperature in-crease, then so far in this century, very little additional increase.The two most plausible explanations for the decadal variabilityare natural climate variability and variable degrees of coolingfrom anthropogenic releases of sulfur gas producing sulfateaerosols (2). This effect has long been proposed as a mechanismto counter greenhouse warming (3), has become the basis formany geoengineering proposals (4), and has been used to attri-bute the lack of warming so far this century to the rapid growthof aerosols in Asia (5).Besides the difference in sign of their temperature effects,

sulfate aerosols are distinguished from greenhouse gases in thatthey only affect daytime radiation, i.e., surface incident solarradiation (Rs). Some kinds of natural variability can also actthrough affecting Rs, i.e., those involving cloud properties.Changes of aerosol loading and cloud properties likely caused

the rapid decrease of Rs, measured at the surface from the 1950sto the 1980s, referred to as “global dimming,” and its partialrecovery after that (6). The plausible suggestion was made byWild et al. (7) that the rapid warming in the late twentieth

century was a consequence of the cessation of global dimming,possibly in part from the imposition of controls on sulfur emis-sion in the industrialized nations (8, 9).This paper examines further the hypothesis that variations in

Rs have caused much of the observed decadal variability in therate of warming. Direct measurements of Rs cannot be quanti-tatively related to such variability because they have been limitedin their geographical coverage. The approach used here is toexamine a global land dataset of diurnal temperature range(DTR). This concept is not new, indeed, Wild et al. (7) noted(compare with their figure 2) that the global pattern of DTR wassimilar to that of their global dimming and brightening. Thepresent paper develops the longest and most comprehensivedataset for DTR possible, and, with some plausible assumptions,establishes what the contribution of Rs has been to decadaltemperature variability. It indicates that a decrease of Rs fromthe 1940s through the 1970s reduced the global temperaturetrend over that period. However, global temperature does notappear to have been significantly affected by changing Rs afterthat. The method is limited in that it is only applicable over land.As the effects of aerosols are likely to be less over ocean, es-pecially in the Southern Hemisphere, this approach may exag-gerate the actual effect of aerosols on global temperature trends.

ResultsRelationship Between Rs and DTR. This section establishes that lo-cally DTR is highly correlated with Rs, but that spatial and sea-sonal variability precludes direct use of this correlation to inferRs where it is not already measured. In the absence of weathervariability, near-surface air temperature Ta over land decreaseswith time at night from longwave radiative cooling and reachesTmin before sunrise. After sunrise, the surface is heated by Rs and

Significance

Global air temperature has become the primary metric forjudging global climate change. The variability of global tem-perature on a decadal timescale is still poorly understood. Thispaper shows that surface incident solar radiation (Rs) over landglobally peaked in the 1930s, substantially decreased from the1940s to the 1970s, and changed little after that. The coolingeffect of this reduction of Rs accounts in part for the near-constant temperature from the 1930s into the 1970s. Sincethen, neither the rapid increase in temperature from the 1970sthrough the 1990s nor the slowdown of warming in the earlytwenty-first century appear to be significantly related tochanges of Rs.

Author contributions: K.W. designed research; K.W. performed research; K.W. and R.E.D.analyzed data; and K.W. and R.E.D. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.

Freely available online through the PNAS open access option.1To whom correspondence should be addressed. E-mail: kcwang@bnu.edu.cn.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1311433110/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1311433110 PNAS | September 10, 2013 | vol. 110 | no. 37 | 14877–14882

EART

H,A

TMOSP

HER

IC,

ANDPL

ANET

ARY

SCIENCE

S

Dow

nloa

ded

by g

uest

on

Aug

ust 5

, 202

0

this heat is transferred as sensible heat H to the overlying air,raising Ta to Tmax in early afternoon. Therefore, changes ofDTR =Tmax − Tmin, have been interpreted as directly related to changes ofRs (6, 7, 10–13). Here we explain how Rs and DTR connectphysically and how their relationship varies with environment.Fig. 1 shows the correlation of monthly anomalies of Rs col-

lected by the Global Energy Balance Archive (GEBA) (14) withDTR from 1950 to 2005 at 524 globally distributed stations (seeSI Text and Fig. S1 for DTR data sources and their qualitycontrol). The correlation coefficients between Rs and DTR arethe highest in humid areas and lower in arid or semiarid areasbecause the fraction of absorbed Rs generating H also dependson variable soil moisture resulting from the frequency and in-tensity of precipitation (15, 16). Besides its dependence on sur-face wetness (17), the partitioning of surface-absorbed Rsbetween H and latent heat flux (λE) depends on land-coverconditions (18, 19) and atmospheric evaporative demand (20). Inhumid areas, both H and λE generally increase with Rs (21, 22),but under warm conditions the latter increases more (23). In aridor semiarid regions, λE is limited by soil water supply and H canaccount for a higher portion of surface absorbed Rs. Fig. 2 shows,as expected from the above discussion, that the sensitivity ofDTR to Rs is higher in arid or semiarid areas than in humid areas.Surface aridity changes seasonally for most monsoon areas,

i.e., where it is wet only in a rainy season, but its interannual

variability is expected to be much less than such seasonalchanges. To reduce the impact of seasonality, we used monthlyand annual anomalies rather than absolute values of DTR and Rs.In the following discussion, we also divide a year into boreal warmseasons (May to October) and boreal cold seasons (Novemberto April).The correlations and sensitivity shown in Figs. 1 and 2 are the

lowest in coastal areas. Evidently the impact of Rs on DTR inthese areas is masked by the impact of energy advection withregular alteration between land breezes and ocean breezes. Thismasking can be substantially reduced by regional averaging ofDTR and Rs (6, 24).Figs. 3 and 4 compare Europe’s and China’s regional average

annual anomalies of DTR with those of Rs. These quantitiesagree quite well, partly because of their better data density anddata continuity (Fig. S3). The agreement between regional DTRand Rs over Europe has also been confirmed by both dataanalysis (10) and model simulation (24). In China, the decreaseof Rs is in good agreement with the reduction of DTR before1990 (11). However, Rs in China increased suddenly during theearly 1990s but not DTR and sunshine duration (25). The in-troduction of new pyranometers from 1990 to 1993 introducedthis inhomogeneity into the Rs observations (25, 26).Fig. 4 also shows that DTR has had a larger temporal vari-

ability than Rs, a consequence of the annual variability of pre-cipitation leading to variations in the partitioning of surfaceabsorbed Rs between λE and H. The Intergovernmental Panel onClimate Change (IPCC) Fourth Assessment Report (AR4) con-cluded that precipitation has had large annual variability duringthe last century, but that its long-term trend and thus its impacton the long-term trend of DTR has been negligible (1), as con-firmed by Figs. 3 and 4 and the following sections. The impact ofannual variability of precipitation is largely removed by using 5-ysmoothing of the anomalies of DTR as in the following.

Variability of DTR, a Proxy of Rs, from 1900 to 2010. This sectionestablishes what is available as a global record for DTR vari-ability. For estimation of DTR over land with optimum spatialand temporal coverage and the highest quality, we combinedthree data sources (27–29) (see SI Text for detailed information)

−150 −100 −50 0 50 100 150

−80

−60

−40

−20

0

20

40

60

80

−0.2 −0.1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8

Fig. 1. Correlation coefficients between monthly anomalies of DTR and Rs.The Rs observations are from the GEBA, and the homogenized maximumand minimum air temperature at 2 m are from the Global Historical Clima-tology Network (GHCN). Both datasets cover the period from 1950 to 2005.Each point in the figure represents a weather station where both Rs and DTRare available for more than 120 mo. There are 524 stations in total.

−150 −100 −50 0 50 100 150

−80

−60

−40

−20

0

20

40

60

80

−0.01 0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08

Fig. 2. The sensitivity of DTR to Rs (in °C per Wm−2) calculated from themonthly anomalies of Rs and DTR. The data used here are the same as in Fig. 1.

−20 −10 0 10 20−1

−0.5

0

0.5

1

1.5

Annual Rs Anomaly (Wm−2)

Ann

ual D

TR

Ano

mal

y (

C)

−20 −10 0 10 20−1

−0.5

0

0.5

1

1.5

Annual Rs Anomaly (Wm−2)

Ann

ual D

TR

Ano

mal

y (

C)

−20 −10 0 10 20−1

−0.5

0

0.5

1

1.5

Annual Rs Anomaly (Wm−2)

Ann

ual D

TR

Ano

mal

y (

C)

−20 −10 0 10 20−1

−0.5

0

0.5

1

1.5

Annual Rs Anomaly (Wm−2)

Ann

ual D

TR

Ano

mal

y (

C)

China, Warm Seasons Europe, Warm Seasons

China, Cold Seasons Europe, Cold Seasons

Fig. 3. Scatterplots of annual anomalies of regional DTR as a function ofannual anomaly of Rs during warm seasons (May to October) and cold sea-sons (November to April) from 1950 to 2005. The correlation coefficients are0.61 and 0.83 over China and 0.86 and 0.73 over Europe during the warmand cold seasons, respectively.

14878 | www.pnas.org/cgi/doi/10.1073/pnas.1311433110 Wang and Dickinson

Dow

nloa

ded

by g

uest

on

Aug

ust 5

, 202

0

for the past 110 y. Monthly anomalies of DTR were derived byremoving its seasonal cycle. Observations of DTR had the highestdensity in North America. To mitigate the impact of the differentdata densities, monthly anomalies of DTR were binned into 5° ×5° grids. Given the low correlation between DTR and Rs incoastal areas, we only selected the grids with more than 50% oftheir area over land, as shown in Fig. S4. The monthly anomaliesat each grid were averaged into regional monthly anomalies, andthen into annual values and 5-y average annual anomalies ateach region, as plotted in Fig. 5.Europe is the only region where measurements of both DTR

and Rs extend back to the 1920s (6). DTR generally increasedin Europe from the 1920s to the 1950s. After the late 1950s, itbegan to decrease until the 1980s, and since the 1990s increased.These variations of DTR are consistent with those of observed Rs(6, 24, 30). The better agreement of warm-season variability ofDTR with that of Rs is consistent with the larger Rs duringwarm seasons.Attempts have been made to correlate annual Rs and DTR

both at regional and global scales (6, 7). However, existingstudies have not recognized that DTR and Rs have differentseasonal cycles; Rs is largest in summertime as a result of highersolar elevation. However, DTR is relatively low in moist summersbecause of the small fraction of Rs that is partitioned into H.Therefore, annual variability of DTR is primarily determined byits variability during seasons other than summer. DTR and Rsagree well both for warm and cold seasons, and variability of Rsover warm seasons is more representative of its annual vari-ability. The reported annual variability of Rs, therefore, agreesbetter with DTR over warm seasons over Europe (and otherregions) than that over an entire year or cold seasons. Variabilityof DTR over warm and cold seasons is substantially different atboth the regional scale (Fig. 5) and the global scale (Fig. 6). Forthis reason, it is essential to consider these differences in recon-structing variability of Rs from DTR.In Asia, DTR substantially decreased from the 1950s to the

1980s, was stable until 2000, and then decreased again, consistentwith Rs derived from sunshine duration (25) and the dimming ofdirectly measured Rs between 1960 and 1990 in China (11, 31).As already mentioned, after the 1990s, direct observations of Rsbecame inconsistent with those of DTR and sunshine (31), a re-sult of the urban bias of Rs observations. When averaged overall stations (∼400 stations) rather than over the ∼50 urbanstations in China with direct observations of Rs, Rs derived from

sunshine duration was stable during the 1990s and decreasedafter 2000 (25).DTR substantially decreased in North America from 1900 to

2010, consistent with the increase of cloudiness, in particular, oflow clouds (32), and decrease of sunshine duration (33). Cloudcover alone accounted for up to 63% of the regional annual DTRvariability in the United States from 1902 to 2002, with cloud-cover trends especially driving DTR in northern United States(34). Aerosol loading over North America was relatively light(35) and rather stable during the past few decades (8). Obser-vations at six stations in the United States showed that Rs sig-nificantly increased from 1995 to 2007 (36, 37), primarily in the1990s (38).As there is a good agreement between Rs and DTR, changes of

Rs are expected to be similar to those of DTR, especially duringthe warm seasons. Fig. 5 shows the variability of DTR over landduring the past century, and hence provides qualitative estimatesof Rs variability over this period. However, it is difficult to re-construct Rs quantitatively using the variability of DTR becauseof the changes of their relationship with time (e.g., from wet todry seasons; Fig. 3) and region (e.g., from humid to arid regions)(Figs. 1 and 2). Below, we describe another approach for usingDTR to estimate the impact of Rs on Ta.

Estimation of the Impact of Rs on Ta from 1900 to 2010. Elevatedgreenhouse gases (GHG) have increased atmospheric downwardlongwave radiation (Ld) (39, 40) and Ta (41) during the twentiethcentury. However, variability in radiative forcing from aerosolsand clouds complicates the attribution of the observed climatechange to the elevated GHG. The previous sections haveestablished a more comprehensive climatology for DTR than thatavailable previously and its long-term variability is highly con-sistent with that of Rs. This climatology allows us to addressthe question of how much of the observed temperature changehas been a result of changes of Rs. For the following analysis,we assume: (i) Tmin is not changed by Rs; and (ii) DTR is onlychanged by changes of Rs (elaborated on in Discussion andConclusions).The globally averaged anomaly of DTR is calculated directly

from its grid values. Daily mean air temperature Ta is commonlyestimated by Ta = 0.5 × (Tmax + Tmin). As DTR = Tmax − Tmin, we

1950 1970 1990 2010−20

−10

0

10

20

Ann

ual R

s A

nom

aly

(Wm

−2 )

1950 1970 1990 2010−20

−10

0

10

20

1950 1970 1990 2010−20

−10

0

10

20

Ann

ual R

s A

nom

aly

(Wm

−2 )

1950 1970 1990 2010−20

−10

0

10

20

1950 1970 1990 2010−1

−0.5

0

0.5

1

1.5

1950 1970 1990 2010−1

−0.5

0

0.5

1

1.5

Ann

ual D

TR

Ano

mal

y (

C)

1950 1970 1990 2010−1

−0.5

0

0.5

1

1.5

1950 1970 1990 2010−1

−0.5

0

0.5

1

1.5

Ann

ual D

TR

Ano

mal

y (

C)

China, Cold Seasons Europe, Cold Seasons

Europe, Warm SeasonsChina, Warm Seasons

Fig. 4. Regionally averaged annual anomalies of Rs (in blue) and DTR (ingreen) during boreal warm seasons (May to October) and boreal cold sea-sons (November to April) from 1950 to 2005. Data used here are the same asin Fig. 3. Equivalent plots for the United States are given in Fig. S2.

1900 1950 2000

−0.5

0

0.5

1900 1950 2000

−0.5

0

0.5

1900 1950 2000

−0.5

0

0.5

Fiv

e−ye

ar A

vera

ge A

nom

aly

of D

iura

nl T

empe

ratu

re R

ange

(D

TR

,C

)

1900 1950 2000

−0.5

0

0.5

1900 1950 2000

−0.5

0

0.5

1900 1950 2000

−0.5

0

0.5

North America

Europe

Africa

Australia

Asia

South America

Fig. 5. Five-year average of annual anomaly (black) of regional DTR from1900 to 2010 averaged from the monthly anomalies at 5° × 5° grids (Fig. S4),which is calculated from weather stations. For comparison, anomalies duringboreal warm seasons (May to October, red) and boreal cold seasons (No-vember to April) are shown.

Wang and Dickinson PNAS | September 10, 2013 | vol. 110 | no. 37 | 14879

EART

H,A

TMOSP

HER

IC,

ANDPL

ANET

ARY

SCIENCE

S

Dow

nloa

ded

by g

uest

on

Aug

ust 5

, 202

0

obtain Ta = Tmin + 0.5 × DTR. For the given assumptions, theimpact of Rs on daily mean Ta is 0.5 × DTR (Fig. 6). Theseassumptions can be inaccurate for various reasons, e.g., changesof daytime radiation can be stored and released to changenighttime temperature. Observations from global flux networksshow that storage fraction is less than 10% of Rs at most surfaces(42, 43). Allowing for this effect would likely amplify our esti-mate of the impact of Rs on Ta over global land by a factorof ∼1.1.We calculate the impact of Rs on mean Ta during the three

time periods: (i) 1900–2010 (the whole time period when dataare available); (ii) 1940–1984 (the global dimming period) and(iii) 1985–2010 (the global brightening period). The results aresummarized in Table 1.Table 1 and Fig. 6 indicate that a reduction in Rs has reduced

Ta and that it decreased most rapidly during the dimming periodof 1940–1984. The rate of temperature increase during the cold

seasons has been reported to be much higher than that duringthe boreal warm seasons (May to October) (44). Fig. 6 showsthat warm-season Rs substantially decreased from the 1940s toearly 1950s and during the 1970s, resulting in a reduction ofmore than 0.2 °C in Ta. Similarly, cold-season Rs substantiallydecreased from the 1960s through the 1970s, resulting in a de-crease of nearly 0.2 °C in Ta. A subsequent increase of Rs wasonly significant over Europe. In conclusion, the variations ofRs partly accounted for the near absence of warming frommidcentury through the 1970s. The maximum cooling seen inthe early 1980s and early 1990s were consistent with theeffects expected from the El Chichón and Pinatubo volcanoes,respectively. Fig. 6 also shows that the results are substantial-ly different for warm seasons, cold seasons, and the entireyear when using DTR to quantify the impact of Rs on air tem-perature (7).

Discussion and ConclusionsThis paper shows, using direct Rs observations (6, 45) and sun-shine duration observations (25), that the interannual variabilityof DTR can be used as a proxy for the long-term variability of Rs.In principle, this relationship should also be applicable to modelsimulations. AR4 climate models (46) show a weak monotonicincrease of DTR from 1950 (44), compare with their figure 5,suggesting that many of the models examined applied a slowconstant ramp-up of aerosol forcing rather than concentratedincreases before 1980 as indicated here. Changes of DTR areexpected to be directly related to H from surface to overlying airbut the magnitudes of these turbulent fluxes are not readilyestimated (22). Many parameters affect the relationship be-tween Rs and H, and consequently, the relationship between Rsand DTR.Impacts of land-cover and land-use change (i.e., urbanization

and irrigation) have been ignored here. In developing countries,such as China and India (47), there has been substantial ur-banization and increased irrigation activity (48) since 1900 withopposing and possibly largely cancelling effects on DTR (16, 49,50), so with impacts likely to be important locally, but likely to besmall at a regional scale (1). Precipitation had a large annualvariability but its long-term trend was negligible during the lastcentury (1), and so likely also its impact on the long-term trendof DTR. At annual timescale or station-scale changes of pre-cipitation and land-cover/land introduce substantial uncertain-ties. Therefore, use of DTR for estimates of variability of Rs and

1900 1920 1940 1960 1980 2000−0.2

−0.1

0

0.1

0.2Im

pact

of R

s on

Ta(

C)

1900 1920 1940 1960 1980 2000

−1

−0.5

0

0.5

1

Mea

n A

ir T

empe

ratu

re (

Ta)

( C

)

Year

May−OctoberNovember−AprilEntire Year

A

B

Fig. 6. (A) The 5-y smoothed impact of Rs on mean air temperature (Ta) overglobal land during boreal warm seasons (May to October, red), boreal coldseasons (November to April, blue), and an entire year (black). The mean airtemperatures over global land are also shown in B.

Table 1. The impact of Rs on daily mean air temperature (Ta) during three periods, 1900–2010,1985–2010, and 1940–1984 (in °C per 100 y)

Time periods Global land North America South America Europe Africa Asia Australia

Yearly1900–2010a −0.11* −0.11* −0.68* 0.01 −0.04 −0.50* −0.0841985–2010b 0.07 −0.21 −0.23 0.46* 0.34 −0.57* 0.801940–1984 −0.36* -0.46* −1.04* −0.24* −0.29* −0.53* −0.08

Warm seasons1900–2010a −0.11* −0.15* −0.47 −0.01 0.03 −0.43* −0.061985–2010b 0.19 −0.31 −0.03 0.75* 1.10* −0.22 1.96*1940–1984 -0.45* −0.62* −0.82 −0.29* −0.43* −0.54* −0.08

Cold seasons1900–2010a −0.12* −0.07 −0.82* 0.03 −0.14 −0.52* −0.091985–2010b −0.09 −0.02 −0.28 0.27 0.49 −0.50 −0.411940–1984 −0.29* −0.22 −1.58* −0.20* −0.28 −0.55* 0.03

Negative values indicate that Rs reduced the rate of warming caused by the elevated GHG, and positive valuesmean that Rs amplified the warming rate by GHG. We also divide the data into boreal warm seasons (May toOctober) and cold seasons (November to April). The asterisk represents impact of Rs is statistically significant (i.e.,pass the Student’s t confidence test at α = 0.05).aTime periods for different regions are different and may cover only a fraction of 1900–2010 (Fig. 5).bTime periods for different regions are different and may cover only a fraction of 1985–2010 (Fig. 5).

14880 | www.pnas.org/cgi/doi/10.1073/pnas.1311433110 Wang and Dickinson

Dow

nloa

ded

by g

uest

on

Aug

ust 5

, 202

0

its impact on Ta should be confined to decadal timescale andregional space scale.Because of the sparse distribution of measurement stations (1)

and changes in measurement methods (38) and instruments (25,51), direct observations cannot provide a reliable estimate of Rsover land during the past century, nor do current climate modelsgenerate long-term variability of Rs (52). This study qualitativelyreconstructs Rs over land from 1990 to 2010 using the latesthomogenized DTR observations at globally distributed weatherstations. It infers that Rs over land globally peaked in the late1930s, substantially decreased from the 1940s to the 1970s, andchanged little after that. These estimates are consistent withobservations of Rs and sunshine duration where these observa-tions are available.More importantly, the DTR observations allow us to estimate

the impact of Rs on the observed changes of Ta. Only changesbefore 1984 appear related to the observed temperature trendsand DTR variability after 1995 indicates a negligible global im-pact of Rs variability. The small impact of Rs on Ta may be partlya result of the low sensitivity of Ta to Rs, much lower than thesensitivity of Ta to longwave radiation caused by greenhousegases (53).The surface energy budget directly determines the Earth’s

surface climate and its changes, but on more local scales stronglyinteracts with transport processes. In consequence, most existingstudies have focused on the energy balance at the top of theatmosphere (5), which is indirectly related to surface Ta, dependingon how clouds (54, 55), aerosols, and other feedbacks work.This paper provides a direct and simple method to estimatethe variability of Rs over land, which is applied from 1900 to

2010 and estimates the impact of this variability on surfacetemperature change.Changes of Rs are primarily determined by changes of clouds

and aerosols. Aerosols are known to have accounted for vari-ability of Rs in Europe and China (25, 56), while clouds havebeen used to explain changes of Rs in the United States (36, 37)during the last two decades. Natural variability from clouds isexpected to be more regional and of shorter timescale than thetrends from aerosols, but otherwise we are not able to separatetheir effects. This paper also does not address the mechanismsthrough which clouds and aerosols respond to climate change(57), i.e., through changes of cloud-cover fraction or cloudheight (58).Our analysis of impact of Rs on Ta does not account for

warming effect of solar radiation absorbed by aerosols, i.e., fromblack carbon (59–61). To zeroth order, aerosol absorption withinthe daytime boundary layer will return the solar energy removedfrom the surface, so will not change DTR but will contribute towarming Ta. Our analysis, in principle, cannot include thewarming of absorbing aerosols in the aerosol layer although theirscattering and absorption effects on surface Rs are included.

ACKNOWLEDGMENTS. Chinese homogenized daily maximum and minimumtemperature at 549 stations were provided by Prof. Zhongwei Yan. GEBAsurface incident solar radiation data were kindly provided by Prof. MartinWild. We thank Dr. Qian Ma for processing some data for this study. Thisstudy was supported by the National Basic Research Program of China(2012CB955302), the National Natural Science Foundation of China(41175126), and the US Department of Energy (BER) Grant DE-FG02-09ER64746.

1. Trenberth KE, et al. (2007) Observations: Surface and atmospheric climate change.

Climate Change 2007: The Physical Science Basis. Contribution of Working Group I to

the Fourth Assessment Report of the Intergovernmental Panel on Climate Change,

eds Solomon S, et al. (Cambridge Univ Press), pp 236–247.2. Booth BBB, Dunstone NJ, Halloran PR, Andrews T, Bellouin N (2012) Aerosols impli-

cated as a prime driver of twentieth-century North Atlantic climate variability. Nature

484(7393):228–232.3. Rasool SI, Schneider SH (1971) Atmospheric carbon dioxide and aerosols: Effects of

large increases on global climate. Science 173(3992):138–141.4. Wei T, et al. (2012) Developed and developing world responsibilities for historical

climate change and CO2 mitigation. Proc Natl Acad Sci USA 109(32):12911–12915.5. Kaufmann RK, Kauppi H, Mann ML, Stock JH (2011) Reconciling anthropogenic cli-

mate change with observed temperature 1998-2008. Proc Natl Acad Sci USA 108(29):

11790–11793.6. Wild M (2009) Global dimming and brightening: A review. J Geophys Res 114(D10):

D00D16.7. Wild M, Ohmura A, Makowski K (2007) Impact of global dimming and brightening on

global warming. Geophys Res Lett 34(4):L04702.8. Wang KC, Dickinson RE, Liang S (2009) Clear sky visibility has decreased over land

globally from 1973 to 2007. Science 323(5920):1468–1470.9. Streets DG, et al. (2009) Anthropogenic and natural contributions to regional trends

in aerosol optical depth, 1980-2006. J Geophys Res 114(D10):D00D18.10. Makowski K, Wild M, Ohmura A (2008) Diurnal temperature range over Europe be-

tween 1950 and 2005. Atmos Chem Phys 8(21):6483–6498.11. Liu B, Xu M, Henderson M, Qi Y, Li Y (2004) Taking China’s temperature: Daily range,

warming trends, and regional variations, 1955–2000. J Clim 17(22):4453–4462.12. Dai A, DelGenio AD, Fung IY (1997) Clouds, precipitation and temperature range.

Nature 386(6626):665–666.13. Dai A, Trenberth KE, Karl TR (1999) Effects of clouds, soil moisture, precipitation, and

water vapor on diurnal temperature range. J Clim 12(8):2451–2473.14. Gilgen H, Wild M, Ohmura A (1998) Means and trends of shortwave irradiance at the

surface estimated from global energy balance archive data. J Clim 11(8):2042–2061.15. Zhou LM, et al. (2009) Spatial dependence of diurnal temperature range trends on

precipitation from 1950 to 2004. Clim Dyn 32(2-3):429–440.16. Zhou LM, Dickinson RE, Tian YH, Vose RS, Dai YJ (2007) Impact of vegetation removal

and soil aridation on diurnal temperature range in a semiarid region: Application to

the Sahel. Proc Natl Acad Sci USA 104(46):17937–17942.17. Wang KC, Liang S (2008) An improved method for estimating global evapotranspi-

ration based on satellite estimation of surface net radiation, vegetation index, tem-

perature, and soil moisture. J Hydrometeorol 9(4):712–727.18. Wang KC, et al. (2007) Influences of urbanization on surface characteristics as derived

from the Moderate-Resolution Imaging Spectroradiometer: A case study for the

Beijing metropolitan area. J Geophys Res 112(D22):D22S06.

19. Wang KC, Wang P, Li ZQ, Cribb M, Sparrow M (2007) A simple method to estimateactual evapotranspiration from a combination of net radiation, vegetation index, andtemperature. J Geophys Res 112(D15):D15107.

20. Wang KC, Dickinson RE, Liang S (2012) Global atmospheric evaporative demand overland from 1973 to 2008. J Clim 25(23):8353–8361.

21. Wang KC, Dickinson RE, Wild M, Liang S (2010) Evidence for decadal variation inglobal terrestrial evapotranspiration between 1982 and 2002: 2. Results. J GeophysRes 115(D20):D20113.

22. Wang KC, Dickinson RE (2012) A review of global terrestrial evapotranspiration:Observation, modeling, climatology, and climatic variability. Rev Geophys 50(2):RG2005.

23. Wang KC, Li ZQ, Cribb M (2006) Estimation of evaporative fraction from a combina-tion of day and night land surface temperatures and NDVI: A new method to de-termine the Priestley-Taylor parameter. Remote Sens Environ 102(3-4):293–305.

24. Makowski K, et al. (2009) On the relationship between diurnal temperature rangeand surface solar radiation in Europe. J Geophys Res 114(D10):D00D07.

25. Wang KC, Dickinson RE, Wild M, Liang S (2012) Atmospheric impacts on climaticvariability of surface incident solar radiation. Atmos Chem Phys 12(20):9581–9592.

26. Tang WJ, Yang K, Qin J, Cheng CCK, He J (2011) Solar radiation trend across China inrecent decades: A revisit with quality-controlled data. Atmos Chem Phys 11(1):393–406.

27. Lawrimore JH, et al. (2011) An overview of the Global Historical Climatology Networkmonthly mean temperature data set, version 3. J Geophys Res 116(D19):D19121.

28. Menne MJ, Williams CN, Vose RS (2009) The U.S. historical climatology networkmonthly temperature data, version 2. Bull Am Meteorol Soc 90(7):993–1007.

29. Li Z, Yan Z (2009) Homogenized daily mean/maximum/minimum temperature seriesfor China from 1960-2008. Atmos Ocean Sci Lett 2(4):237–243.

30. Ohmura A (2006) Observed long-term variations of solar irradiance at the earth’ssurface. Space Sci Rev 125(1-4):111–128.

31. Shi GY, et al. (2008) Data quality assessment and the long-term trend of ground solarradiation in China. J Appl Meteorol Climatol 47(4):1006–1016.

32. Sun B, Karl TR, Seidel DJ (2007) Changes in cloud-ceiling heights and frequencies overthe United States since the early 1950s. J Clim 20(15):3956–3970.

33. Angell JK (1990) Variation in United States cloudiness and sunshine duration between1950 and the drought year of 1988. J Clim 3(2):296–308.

34. Lauritsen RG, Rogers JC (2012) U.S. diurnal temperature range variability and regionalcausal mechanisms, 1901–2002. J Clim 25(20):7216–7231.

35. Wang KC, Dickinson RE, Su L, Trenberth KE (2012) Contrasting trends of mass andoptical properties of aerosols over the Northern Hemisphere from 1992 to 2011.Atmos Chem Phys 12(19):9387–9398.

36. Long CN, et al. (2009) Significant decadal brightening of downwelling shortwave inthe continental United States. J Geophys Res 114(D10):D00D06.

37. Augustine JA, Dutton EG (2013) Variability of the surface radiation budget over theUnited States from 1996 through 2011 from high-quality measurements. J GeophysRes 118(1):43–53.

Wang and Dickinson PNAS | September 10, 2013 | vol. 110 | no. 37 | 14881

EART

H,A

TMOSP

HER

IC,

ANDPL

ANET

ARY

SCIENCE

S

Dow

nloa

ded

by g

uest

on

Aug

ust 5

, 202

0

38. Wang KC, Dickinson KE, Ma Q, Augustine JA, Wild M (2013) Measurement methodsaffect the observed global dimming and brightening. J Clim 26(12):4112–4120.

39. Wang KC, Liang S (2009) Global atmospheric downward longwave radiation overland surface under all-sky conditions from 1973 to 2008. J Geophys Res 114(D19):D19101.

40. Wang KC, Dickinson RE (2013) Global atmospheric downward longwave radiation atthe surface from ground-based observations, satellite retrievals and reanalyses. RevGeophys 51(2):150–185.

41. Dickinson RE, Cicerone RJ (1986) Future global warming from atmospheric tracegases. Nature 319(6049):109–115.

42. Jacobsen A (1999) Estimation of the soil heat flux/net radiation ratio based onspectral vegetation indexes in high-latitude Arctic areas. Int J Remote Sens 20(2):445–461.

43. Clothier BE, et al. (1986) Estimation of soil heat flux from net radiation during thegrowth of alfalfa. Agric For Meteorol 37(4):319–329.

44. Wallace JM, Fu Q, Smoliak BV, Lin P, Johanson CM (2012) Simulated versus observedpatterns of warming over the extratropical Northern Hemisphere continents duringthe cold season. Proc Natl Acad Sci USA 109(36):14337–14342.

45. Wild M, et al. (2009) Global dimming and brightening: An update beyond 2000.J Geophys Res 114(D10):D00D13.

46. Zhou LM, Dickinson RE, Dai AG, Dirmeyer P (2010) Detection and attribution of an-thropogenic forcing to diurnal temperature range changes from 1950 to 1999:Comparing multi-model simulations with observations. Clim Dyn 35(7-8):1289–1307.

47. Zhou LM, et al. (2004) Evidence for a significant urbanization effect on climate inChina. Proc Natl Acad Sci USA 101(26):9540–9544.

48. Mukherji A, et al. (2009) Revitalizing Asia’s Irrigation: To Sustainably Meet To-morrow’s Food Needs (Food and Agricultural Organization of the United Nations,Rome)Rome,.

49. Geerts B (2002) On the effects of irrigation and urbanisation on the annual range ofmonthly-mean temperatures. Theor Appl Climatol 72(3-4):157–163.

50. Sacks WJ, Cook BI, Buenning N, Levis S, Helkowski JH (2009) Effects of global irriga-tion on the near-surface climate. Clim Dyn 33(2-3):159–175.

51. Wang KC, Augustine J, Dickinson RE (2012) Critical assessment of surface incidentsolar radiation observations collected by SURFRAD, USCRN and AmeriFlux networksfrom 1995 to 2011. J Geophys Res 117(D23):D23105.

52. Wild M, Schmucki E (2011) Assessment of global dimming and brightening in IPCC-AR4/CMIP3 models and ERA40. Clim Dyn 37(7):1671–1688.

53. Stanhill G (2011) The role of water vapor and solar radiation in determiningtemperature changes and trends measured at Armagh, 1881-2000. J Geophys Res116(D3):D03105.

54. Spencer RW, Braswell WD (2010) On the diagnosis of radiative feedback in the pres-ence of unknown radiative forcing. J Geophys Res 115(D16):D16109.

55. Dessler AE (2010) A determination of the cloud feedback from climate variations overthe past decade. Science 330(6010):1523–1527.

56. Streets DG, Wu Y, Chin M (2006) Two-decadal aerosol trends as a likely explanation ofthe global dimming/brightening transition. Geophys Res Lett 33(15):L15806.

57. Kerr RA (2010) Climate change. El Niño lends more confidence to strong globalwarming. Science 330(6010):1465.

58. Davies R, Molloy M (2012) Global cloud height fluctuations measured by MISR onTerra from 2000 to 2010. Geophys Res Lett 39(3):L03701.

59. Jacobson MZ (2012) Investigating cloud absorption effects: Global absorption prop-erties of black carbon, tar balls, and soil dust in clouds and aerosols. J Geophys Res117(D6):D06205.

60. Ramanathan V, et al. (2005) Atmospheric brown clouds: Impacts on South Asian cli-mate and hydrological cycle. Proc Natl Acad Sci USA 102(15):5326–5333.

61. Ramanathan V, et al. (2007) Warming trends in Asia amplified by brown cloudsolar absorption. Nature 448(7153):575–578.

14882 | www.pnas.org/cgi/doi/10.1073/pnas.1311433110 Wang and Dickinson

Dow

nloa

ded

by g

uest

on

Aug

ust 5

, 202

0