Contemporary temperature changes at the ground surface and in the troposphere over Vojvodina, Serbia

Full text

Page 1

GEOGRAPHICA PANNONICA
Volume 12, Issue 2, 56-61 (2008)
56
Vladan Ducić1, Stevan Savić2, Jelena Luković1*
Received: January 2008 | Revised: November 2008 | Accepted: November 2008
Contemporary Temperature Changes
at the Ground Surface and in the
Troposphere Over Vojvodina, Serbia
Introduction
The problem of global air temperature changes has been the
priority issue of scientists and professionals in the field. Sig-
nificant number of papers has been written with various
solution proposals to the problem (Mears et al., 2003; Vin-
nikov, Grody, 2003; Christy, Norris, 2004; Vinnikov et al.,
2006; Radovanović et al., 2006). Intergovernmental Panel
on Climate Change (IPCC, 2007) supports the notion that
the major cause of air temperature increase in the late 20th
century is the anthropogenic factor, i.e. the increase of CO2
concentration in the atmosphere (http://ipcc-wg1.ucar.edu/
wg1/wg1-report.html).
On the other hand, some authors argue the domination
of natural factors in contemporary variability of the climate
system. For example, Robock emphasises that “the research
on the prospective influence of human-induced green-
house effect needs to be interrelated with the study of exter-
nal natural climatic factors (volcanic eruptions, changes in
the solar activity), which stay constant, as well as internal
dynamics of the climate system” (Robock, 2002).
Data and methodology
For the analysis of air temperature changes on the territo-
ry of Vojvodina, the temperature data series for the peri-
od 1979-2005 from the network of meteorological stations
(Palić, Sombor, Rimski Šančevi, Kikinda, Zrenjanin, Vršac,
Sremska Mitrovica) (figure 1), were collected to be com-
pared with the satellite measurements. Shown values of an-
nual mean temperatures for seven meteorological stations
were taken from Meteorological yearbooks of Republic Hy-
drometeorological Service of Serbia and Hydrometeorolog-
ical Service of the Province of Vojvodina. However, inho-
mogeneity of data at meteorological stations in Serbia was
detected (Smailagić and Jovanović, 1992), which demanded
the data homogenization procedure first.
Homogeneity of mean annual air temperatures, meas-
ured at those 7 meteorological stations on the territory of
Vojvodina, was tested by Alexandersson’s test for shifts (Al-
exandersson, 1986). The test is based upon the assumption
that the difference between temperature series at a can-
didate station (the one being tested) and the reference se-
ries is fairly constant in time. During the temperature se-
ries testing, the variations between candidate and reference
station were used. Simultaneously, instead of the values
obtained from meteorological stations, the successive in-
crements of the data were used for calculations of square
correlation coefficient (ΔT=Ti+1-Ti), with the aim of mini-
mizing the risk of making poor estimates of correlations
between the candidate and reference temperature series be-
cause of the existing inhomogeneities (Peterson and Easter-
Abstract
In this paper the near surface and tropospheric temperature trends are compared on the territory of Vojvodina were inves-
tigated, on the basis of four sets of data from ground-based and satellite observation, for the period 1979-2005. The trend
analysis has shown that there is no amplification of trend of air temperature increase with altitude. On the basis of our calcu-
lations, we obtained the results that the air temperature increase in Vojvodina, approximately 1.5 times higher in the near-sur-
face layer, compared to lower and middle layers of troposphere. This shows flat or decreasing amplification with altitude, in
contrary to some models.
Key words: observed climate change, temperature, tropospheric temperature, linear trend, Vojvodina, Serbia.
1 Faculty of Geography, University of Belgrade, Studentski trg 3/III, 11000 Belgrade, Serbia
2 Climatology and Hydrology Research Centre, Faculty of Science, University of Novi Sad, Trg Dositeja Obradovića 3, 21000 Novi Sad, Serbia
* Corresponding author: Jelena Luković, e-mail: jelenalu@yahoo.com; Co-authors: vladan@gef.ac.yu (Vladan Ducić); stevan.savic@ig.ns.ac.yu
(Stevan Savić)
ISSN 1820-7138 (online)

Page 2

57
Geographica Pannonica • Volume 12, Issue 2, 56-61 (2008)
Vladan Ducić, Stevan Savić,
Jelena Luković
ling, 1994). Between 2 and 5 stations were used for reference
series. Correlation coefficient was above 0.9 in all cases.
During the testing process of mean annual temperature
series, the results showed that at 6 meteorological stations
the temperature series were homogenous. The homogeneity
break point which exceeds the minimum threshold level of
importance α=0.05 (the value is 7.643) (Khaliq and Ouarda,
2007), was recorded at Rimski Šančevi station, and the
break point year was 1990. After performing series of sev-
eral homogeneity adjustments (Moberg and Alexandersson,
1997), the mean annual air temperatures at Rimski Šančevi
station were homogenised.
Mean value for homogenised series at all seven meteoro-
logical stations (MSV) was used in the analysis of contempo-
rary air temperature changes on the territory of Vojvodina.
Besides this set of data, the air temperature series were
also analysed from GHCN3 and HadCRUT34. The data ob-
tained from GHCN are the result of ground-based meteoro-
logical observations around the world, comprising also the
observations from this area. The data are available (http://
www.co2science.org/scripts/CO2ScienceB2C/data/temper-
3 GHCN-Global Historical Climatology Network
4 Hadley Centre and the Climatic Research Unit at the University of East An-
glia
atures/ghcn.jsp) for the period since 1880 as grids 5ºx5º lat-
itude and longitude. It is essential to this point to remark
that the distribution of meteorological stations over a grid
square varies; however, it can be eliminated by certain sta-
tistical tools. Thus, their comparison is enabled.
HadCRUT3 is a network of data from ground-based in-
strumental measurements which are expressed as temper-
ature anomalies from mean values. The data are available
(http://www.co2science.org/scripts/CO2ScienceB2C/data/
temperatures/hadley.jsp) in the form of mean annual val-
ues for the period since 1880 and organized in grids 5ºx5º.
This database is the result of cooperation between Met Of-
fice Hadley Centre and Climatic Research Unit at the Uni-
versity of East Anglia.
Ground-based measurements demonstrate objective de-
ficiencies. First, the measurements cover only certain spots
of the areas, the spots where meteorological stations are sit-
uated, but not the research area. Second, frequently at those
“spots” the phenomenon of urban heat island occurs, which
is significantly warmer than its surroundings. Thus, the
fake illustration of higher temperature values for the broad-
er area is formed.
With regards to objective deficiencies of ground-based
measurements which further reflect in the temperature
Figure 1 Positions of meteorological stations in Vojvodina and their geographical coordinates

Page 3

Contemporary Temperature Changes at the Ground Surface
and in the Troposphere Over Vojvodina, Serbia
58
Geographica Pannonica • Volume 12, Issue 2, 56-61 (2008)
trend, we have decided to include the series from satellite
observations. Despite the relatively short period of meas-
urements, satellite data offer reliable values deprived of lo-
cal influences (e.g. urban heat island).
The data from satellite observations for the period be-
tween 1979 and 2005 were used in the research. Satellite
data obtained from NASA were processed at the Univer-
sity of Alabama in Huntsville(UAHMSU)(http://www.co-
2science.org/scripts/CO2ScienceB2C/data/temperatures/
msu.jsp). Opposite to the ground-based observations data
at GHCN, these data refer to the 8 km thick bottom lay-
er of the atmosphere. As far as the covered area is con-
cerned, these data are global and available in the form of
grids 2.5ºx2.5º latitude and longitude. The time framework
for the data is set to the beginning of satellite observations
in 1979 (Christy et al., 2000).
Additionally, there are Remote Sensing Systems (RSS)
the network of satellite data in Santa Rosa, California.
These data are available as a geographical belt (from -70ºS
to 82.5ºN), unfortunately not in the form of grids, therefore
they were not used in the calculations of regional trends.
Our calculated correlation coefficient for those two sets
of satellite data (UAHMSU and RSS) for the belt 70ºS to
82.5ºN, for the series 1979-2005, had the values of 0.98, in-
dicating the high correspondence of the two series. Howev-
er, there is the difference between the values of linear trend
which is the result of data processing dissimilarity. Con-
cerning the fact that GHCN, MSU and HadCRUT3 are giv-
en as anomalies with reference to period from 1979 up to
1998, MSV data are also averaged using the same period.
Methodologically, the commonly mathematical-statisti-
cal methods (trend analysis, correlation coefficient, stand-
ard deviation) were used in the paper.
Results and discussion
Parallel data for different networks and satellite measure-
ments were given in the Table 1. According to the resolution
of available data, the coordinates 15-25ºE and 45-50ºN were
chosen, within which the larger portion of Vojvodina is sit-
uated. High resolution of satellite data enabled the obser-
vation of grid at 2.5ºx2.5º, i.e. in this case coordinates 17.5-
22.5ºE and 45-47.5ºN.
Correlation coefficient (R) between HadCRUT3 and
GHCN for the period 1979-2005 is 0.99. Homogenised
ground-based air temperature series for Vojvodina (MSV)
show the same correlation coefficient of 0.98 for both Had-
CRUT3 and GHCN. Firstly, it indicates that there is no ma-
jor discrepancy between HadCRUT3 and GHCN. On the
other hand, it justifies the procedure of data homogenisa-
tion for Vojvodina. In spite of some remained quality prob-
lems the used time series are suitable for examining cli-
matic variability. In case of near surface data the very high
correlations is a strong evidence for that.
However, correlation coefficients between ground and
satellite measurements are relatively low. Thus, R between
MSV and satellite data of high resolution 17.5-22.5ºE and
45-47.5ºN is 0.21, whereas similar values of R are calculat-
ed between HadCRUT3 (0.16) and GHCN (0.20) with satel-
lite data. Such value discrepancies of R could be explained
by the fact that air temperature data refer to different alti-
tudes. Ground measurements record the air temperature at
2 m height, whereas satellite data refer to the first 8 km lay-
er of troposphere. It is important to add that Johanson and
Fu (2006) have estimated the maximal error of calculat-
ed trends to be 0.01°K/decade. Also, the physical relevance
of estimated linear trends is affected by natural noise and
low frequency changes (Mills, 2006). This corruption is the
highest when the time series are short, as in our case.
All three groups of ground-based measurements record
minimum values in 1980, with the highest deviation record-
ed by MSV (-1.21ºС). Similarly, all ground measurements
record maximum values for 2000, but the deviations are
smaller, ranging from 1.73ºС for GHCN to 1.99ºС for MSV.
Satellite measurements for both resolutions recorded the
largest negative deviation in 1982 (-1.05ºС for 15-25ºE and
45-50ºN, -1.12ºС for 17.5-22.5ºE and 45-47.5ºN). Maximum
deviation was recorded in 1987 (ranging from 1.14ºС from
the first network to 1.05ºС for the second).
Comparative analysis of linear trend values has shown
the highest increase within MSV (0.0427ºС annually) (Fig-
ure 2). Slightly smaller values were observed within trends
for GHCN (0.0384ºС annually) and HadCRUT3 (0.0398ºС
annually). Explicitly, all three values of the ground-based
measurements trend are approximately similar and com-
parable. Compared to air temperature trends in Europe5
(annually GHCN 0.039ºС and HadCRUT3 0.040ºС), the
air temperature changes in Vojvodina are similar. How-
ever, the changes are higher than the global ones (annual-
ly GHCN 0.026ºС and HadCRUT3 0.020ºС). Relatively fast
warming in Vojvodina can be a sign of sensitivity demon-
strated by GCM scenarios (Tebaldi et al., 2005).
The data on the linear trend for the first layer of tropo-
sphere at altitude of 8 km based on satellite observations
show lower values compared to data of ground observations.
Thus, the trend of satellite data of low resolution is 0.0275ºС
annually, whereas the one for the high resolution (17.5-22.5ºE
and 45-47.5ºN) is 0.0288ºС annually. Both values are indis-
putably lower compared to those obtained by ground meas-
urements in all three networks. Compared to the data for Eu-
rope (0.030ºС), the linear trend of air temperature changes
in Vojvodina is slightly smaller. Considering the global lev-
el, the satellite data indicate a minor change (0.013ºС). It is
also essential to remark that all calculated values of the line-
ar trend of air temperature statistically significant with 0.05 –
0.025 risk of first type error in hypothesis testing.
Apparently, there is a discrepancy in the trend of air tem-
perature increase between the satellite and ground measure-
ments. The discrepancy was higher prior to the corrections
being made. Those corrections have been done 2005 due to
methodological processing (Maers and Wentz, 2005). Explic-
itly, UAHMSU measurements recorded lower values of global
temperature trend up to 2005 (0.07ºС). After the corrections
had been made in data processing, the discrepancy was low-
ered, but remained noticeable. Moreover, there were discrep-
ancies between the satellite data themselves, i.e. their process-
ing. On the basis of RSS data for the period 1979-2005, the
trend value6 was 0.19ºС per a decade (http://www.ssmi.com/
5 Between 35-70ºN and -10-45ºE
6 However, certain corrections were made at the beginning of 2008, and RSS
trend value is almost equal to MSU

Page 4

59
Geographica Pannonica • Volume 12, Issue 2, 56-61 (2008)
Vladan Ducić, Stevan Savić,
Jelena Luković
msu/msu_data_description.html#msu_decadal_trends),
whereas UAHMSU data show the value of 0.13ºС per decade
(http://vortex.nsstc.uah.edu/data/msu/t2lt/uahncdc.lt).
The researchers render dissimilar explanations of the di-
vergence between satellite and ground measurements, which
could be classified into three groups. The first group would
comprise the errors in the measurement procedure; the sec-
ond group would refer to various impacts of natural and an-
thropogenic factors on temperature at certain altitudes; and
the third group would be base on the divergence in spatial
distribution of observations (Santer et al., 2000). However,
the divergences are inevitable reality (Douglass et al., 2004b).
On the other hand, the analysis of radio sonde data in lower
and middle layers of troposphere match the UAHMSU data,
which supports the statement on the data reliability (http://
www.marshall.org/pdf/materials/415.pdf).
Considering the analysed data in the previous section, we
attempted to test out whether the divergence in the trends
possibly highlights the potential causes of climate change?
The scientific papers list that under the dominating condi-
tions of anthropogenic greenhouse effect; higher increase
of air temperature should be expected in middle and lower
layers of troposphere than in the near-surface layer (Santer
et al., 2000; NRC, 2000; NAS, 2001; IPCC - http://ipcc-wg1.
ucar.edu/wg1/wg1-report.html). The estimations stress the
fact that the increase should be 1.2 to 1.5 times higher, de-
pending on the latitude and applied model.
The data analysis of yearly trend divergence in air tem-
perature between satellite data for grid 45-47.5ºN 17.5-22.5ºE
and MSV (Figure 3) has shown the decreasing trend.
It complies with the fact that the increase trend of air tem-
perature in lower layer is higher. The largest negative devia-
tions were recorded in the last quarter of the observed period
(2000 and 2002). The obtained results do not contribute to the
hypothesis about altitude amplification of the increase trend
of air temperature. Conversely, on the basis of our calculations,
Table 1 Air temperature anomalies (averaged from 1979 through 1998) for different networks in Vojvodina
Years
GHCN
15-25º E
45-50º N
HADCRUT3
15-25º E
45-50º N
MSU
15-25º E
45-50º N
MSU
17.5-22.5º E
45-47.5º N
МSV
19790.25
-0.96
0.23
-1.03
-0.37-0.350.24
-1.21
19800.06 0.01
19810.11 0.20 0.33
-1.05
0.35
-1.12
-0.08
19820.39 0.25 0.09
19830.750.77 -0.48-0.46 0.40
1984 -0.15-0.17-0.75 -0.71 -0.40
1985-0.81 -0.87 -0.89-0.84 -1.02
1986 -0.04 -0.080.51
1.14
0.50
1.05
-0.35
1987 -0.53-0.43 -0.48
19880.19 0.370.77 0.81 0.12
19890.890.81-0.12-0.14 0.45
19900.97 0.95 0.120.230.79
1991 -0.13 -0.100.35 0.34-0.65
19920.95 1.000.78 0.850.99
19930.06 0.26-0.48 -0.73-0.02
19941.641.67 0.04 -0.011.42
19950.42 0.480.010.040.15
1996-0.46 -0.46 -0.70-0.45 -0.58
1997-0.020.060.150.16-0.29
19980.510.58 0.590.460.35
19990.81
1.73
0.89
1.76
0.38 0.510.61
1.99
20000.220.33
20010.690.670.420.540.59
2002 1.461.44-0.06-0.08 1.65
20030.720.72 0.71 0.800.54
20040.540.530.53 0.450.38
2005
Average
Standard deviation
Linear trend
0.12
0.37
0.68
0.11
0.39
0.68
0.51
0.10
0.56
0.46
0.11
0.56
-0.31
0.20
0.76
0.03840.03980.02750.0288 0.0427

Page 5

Contemporary Temperature Changes at the Ground Surface
and in the Troposphere Over Vojvodina, Serbia
60
Geographica Pannonica • Volume 12, Issue 2, 56-61 (2008)
we obtained the results that the air temperature increase in
Vojvodina, approximately 1.5 times higher in the near-surface
layer, compared to lower and middle layers of troposphere.
A panel convened by the National Research Council (2000)
found for the satellite era (since 1979) ‘apparently conflicting
surface and tropospheric temperature trends’ that could not
be reconciled, with the Earth’s surface warming faster than
the lower troposphere. The panel concluded, after consider-
ing possible systematic errors that ‘substantial disparity re-
mains.’ From a study of several independent observational
datasets (Douglass et al., 2004b) confirmed that the disparity
was real and arose mostly in the tropical zone. Also, Doug-
lass et al. (2004a) showed that three state-of-the-art General
Circulation Models (GCMs) predicted a temperature trend
that increased with altitude, reaching a maximum ratio to
the surface trend (‘amplification’ factor R) as much as 1.5–2.0
at a pressure (altitude) about 200–400 hPa. This was in disa-
greement with observations, which showed flat or decreasing
amplification factors with altitude.
Also the global data contribute to the conclusion. Name-
ly, in the observed period, similar results are obtained. The
trend value for ground-based measurements at GHCN is
0.026ºС annually, whereas the trend value obtained at Had-
CRUT3 is 0.020ºС annually. Both values are higher than the
satellite obtained trend value for air temperature, which are
0.013ºС and 0.019ºС at UAHMSU and RSS, respectively.
Conclusion
The data on air temperature measurements within differ-
ent observation networks over the territory of Vojvodina
were processed in this paper. The first set comprises ground
measurements within the global data network (GHCN and
Figure 2 Annual change of homogenised air temperature data in Vojvodina
Figure 3 Air temperature differences between MSU and MSV data
ºC
The annual changes of homogenous temperature data in Vojvodina
Linear trend
1997 200519992001 200319911993 19951985 1987
19891979 1981 1983
2.5
2.0
1.5
1.0
0.5
0.0
-0.5
-1.0
-1.5
y = 0.0427x - 0.3988
R2 = 0.1998
2.0
1.5
1.0
0.5
0.0
-0.5
-1.0
-1.5
-2.0
ºC
MSU - MSV
Linear trend
y = -0.0152x + 0.1147
R2 = 0.02
199720051999 20012003 199119931995 19851987
1989197919811983

Page 6

61
Geographica Pannonica • Volume 12, Issue 2, 56-61 (2008)
Vladan Ducić, Stevan Savić,
Jelena Luković
HadCRUT3) and at 7 meteorological stations in Vojvodina,
for which homogenization was performed by Alexanders-
son’s test. The second set of data comprises satellite obser-
vations in two resolutions.
The analysis of the data has shown the discrepancy in air
temperature trend between ground and satellite measure-
ments. Explicitly, the air temperature increase at the near-
surface layer has shown higher values (GHCN 0.0384ºС an-
nually, HadCRUT3 0.0398ºС annually, homogenised data
for Vojvodina - MSV 0.0425ºС annually) compared to sat-
ellite observations (UAHMSU from 0.0275ºС to 0.0288ºС
annually).
With reference to written resources it may be conclud-
ed that the variations in the air temperature trend between
measurements made at lower and middle layers of tropo-
sphere and those made in the near-surface layer stand for
physical reality. Besides, MSV data and satellite data may
be considered reliable.
The obtained results show the lack of amplification in the
increasing trend of air temperature both in Vojvodina and
global one, for the period from 1979 to 2005.
Acknowledgements
This study is supported by the Serbian Ministry of Science
under Grants 146005 and 146019. The authors would like to
thank the anonymous referees for their encouraging com-
ments.
References
Alexandersson, H. 1986. A homogeneity test applied to pre-
cipitation data. Journal of Climatology 6, 661-675.
Christy J.R., Spencer R.W., Braswell W.D. 2000. MSU Trop-
ospheric Temperatures: Dataset Construction and Radi-
osonde Comparisons. Journal of Atmospheric and Oce-
anic Research 17, 1153-1170.
Christy J.R., Norris, W.B. 2004. What may we conclude about
global tropospheric temperature trends? Geophysical Re-
search Letters 31, L06211, doi:10.1029/2003GL019361.
Douglass D.H, Pearson B.D, Singer S.F. 2004a. Altitude de-
pendence of atmospheric temperature trends: climate
models versus observation. Geophysical Research Letters
31, L13208, DOI:10.1029/2004GL020103.
Douglass, D.H., Pearson B.D., Singer S.F., Knappenberger
P.C., Michaels P.J. 2004b. Disparity of tropospheric and
surface temperature trends: new evidence. Geophysical
Research Letters 31, L13207, 1-5.
Intergovernmental Panel on Climate Change (2007), Cli-
mate Change (2007): The Scientific Basis. Contribution
of Working Group 1 to the Fourth Assessment Report of
the Intergovernmental Panel on Climate Change. Cam-
bridge
Johanson, C.M., Fu, Q. 2006: Robustness of tropospheric
temperature trends from MSU channels 2 and 4. Journal
Climate 19, 4234-4242.
Mills, T.C. 2006: Modelling current trends in northern
hemisphere temperatures. International Journal of Cli-
matology 26, 867-884.
Khaliq, M.N., Ouarda, T.B.M.J. 2007. On the critical values
of the standard normal homogeneity test (SNHT). Inter-
national Journal of Climatology 27, 681-687.
Mears C.A., Schabel, M.C., Wentz, F.J. 2003. A reanalysis
of the MSU channel 2 tropospheric temperature record.
Journal of Climatology 16, 3650–3664.
Mears, C.A., Wentz, F.J. 2005. The effect of diurnal correc-
tion on satellite-derived lower tropospheric temperature.
Science 309, 1548 –1551.
Moberg, A., Alexandersson, H. 1997. Homogenization of
Swedish temperature data. Part II: homogenized gred-
ded air temperature compared with a subset of global
gridded air temperature since 1861. International Jour-
nal of Climatology 17, 35-54.
NAS, 2001. Climate Change Science: An Analysis of Key
Questions. National Academy of Sciences, National
Academy Press, Washington DC, 1-42.
NRC, 2000. Reconciling Observations of Global Tempera-
ture Change. National Academy Press, Washington DC,
1-85.
Peterson, T.C., Easterling, D.R. 1994. Creation of homoge-
neous composite climatological reference series. Inter-
national Journal of Climatology 14, 671-679.
Radovanović, M., Vemić, M., Popović, I. 2006. Global Cli-
mate Changes – Antidogmatron. Geographica Pannoni-
ca 10, 9-13.
Robock, A. 2002. Volcanic eruptions. Encyclopedia of Glo-
bal Environmental Change, vol. 1, Ted Munn, Ed., (John
Wiley and Sons, London), 738-744.
Santer, B.D., Wigley, T.M.L., Gaffen, D.J., Bengtsson, L.,
Doutriaux, C., Boyle, J.S., Esch, M., Hnilo, J.J., Jones,
P.D., Meehl, G.A., Roeckner, E., Taylor, K.E., Wehn-
er, M.F. 2000. Interpreting Differential Temperature
Trends at Surface and in the Lower Troposphere. Science
287, 1227-1232.
Smailagić, J., Jovanović, O. 1992. Homogeneity of the an-
nual air temperature and precipitation on the major me-
teorological stations in Serbia. Researches Review of the
Hydrological and Meteorological Service of Serbia, Bel-
grade. (in Serbian)
Vinnikov, K.Y., Grody, N.C. 2003. Global warming trend of
mean tropospheric temperature observed by satellites.
Science 302, pp. 269– 272.
Vinnikov, K.Y., Grody, N.C., Robock, A., Stouffer, R.J., Jones,
P.D., Goldberg, M.D. 2006. Temperature trends at the
surface and in the troposphere. Journal of Geophysical
Research 111, D03106, doi:10.1029/2005JD006392.
Tebaldi, C., Smith, R.L., Nychka, D., Mearns, L.O. 2005.
Quantifying uncertainty in projections of regional cli-
mate hange: A Bayesian approach to the analysis of multi
model ensembles. Journal of Climate 18, 1524–1540.
http://ipcc-wg1.ucar.edu/wg1/wg1-report.html
http://www.co2science.org/scripts/CO2ScienceB2C/data/
temperatures/ghcn.jsp
http://www.co2science.org/scripts/CO2ScienceB2C/data/
temperatures/hadley.jsp
http://www.co2science.org/scripts/CO2ScienceB2C/data/
temperatures/msu.jsp
http://www.ssmi.com/msu/msu_data_description.
html#msu_decadal_trends
http://vortex.nsstc.uah.edu/data/msu/t2lt/uahncdc.lt
http://www.marshall.org/pdf/materials/415.pdf
IPCC - http://ipcc-wg1.ucar.edu/wg1/wg1-report.html