Evidence of climate warming from underground temperatures in NW Italy
Transcript of Evidence of climate warming from underground temperatures in NW Italy
www.elsevier.com/locate/gloplacha
Global and Planetary Cha
Temperature signal in the underground for climate history
reconstruction in Italy
V. Pasqualea,*, M. Verdoyaa, P. Chiozzia, L. Bodrib, S. Bellanic
aDipartimento per lo Studio del Territorio e delle Sue Risorse, Universita di Genova Viale Benedetto XV 5, I-16132 Genoa, ItalybGeophys. Res. Group, Hung. Acad. Sci., 1083 Budapest, Hungary
cCNR, Istituto Geoscienze e Georisorse Via Moruzzi 1, I-56124- Pisa, Italy
Received 22 December 2003; accepted 30 November 2004
Abstract
Underground temperature data from height boreholes logged between 1981 and 2000 were studied to infer the climate
change in central-northern Italy during the last 250 years. The ground surface temperature history was reconstructed by using
the functional space inversion method. A different inverse approach was also used for two temperature sets to obtain the fine
details of the most recent surface temperature change. The results were compared with the air temperature recorded since the
beginning of the 19th century at meteorological observatories. The analysis puts into evidence that the trend of the temperature
change in the western side of the Apennines chain differs from that of the eastern side. Since 1750 the western side shows
temperature lower than that of the 1990s, with minimum values in the period 1930–1960, followed by an almost linear increase
in the ground surface temperature. Along the eastern side the temperature is always larger than that inferred for the 1970s, with
maximum values in the period 1920–1940, which is followed by a sharp temperature decrease. Only since 1970–1980 a local
warming phase has started. By combining borehole temperature logs with meteorological surface air temperature records, the
pre-observational mean temperature was calculated. The results corroborate the difference of the climatic histories in both sides
of the Apennines concerning the ground surface temperatures. It also appears that the recent climatic changes have partly a local
origin and can obscure the changes forced by the regional surface air temperature influence.
D 2004 Elsevier B.V. All rights reserved.
Keywords: climate change; meteorological record; ground temperature history; global warming, Italy
1. Introduction
Analyses of surface air temperature (SAT) time
series, recorded over the period 1867–1995 at a
0921-8181/$ - see front matter D 2004 Elsevier B.V. All rights reserved.
doi:10.1016/j.gloplacha.2004.11.015
* Corresponding author. Fax: +39 10 352169.
E-mail address: [email protected] (V. Pasquale).
number of meteorological stations scattered on the
whole Italian territory, roughly delineated two
climatic sub-regions, northern and central-southern
Italy, both revealing increase in the temperatures
with average rates of 0.3–0.4 and 0.5–0.7 K/100
year, respectively (Maugeri and Nanni, 1998; Bru-
netti et al., 2000). These values are somewhat lower
nge 47 (2005) 36–50
V. Pasquale et al. / Global and Planetary Change 47 (2005) 36–50 37
to the north and coincide well to the south with the
estimates by Ghil and Vautard (1991) and Jones
et al. (1999) for the Northern Hemisphere. However,
the Italian orography and the position of the country
surrounded by the Mediterranean Sea imply strong
effects of the local air circulation that cause high
spatial and temporal climate variability (Pasquale
et al., 1996; Nanni et al., 1998).
To identify these differences by using only
meteorological data is problematic because of the
relative raw spatial resolution of the existing SAT
records and the possible presence of hidden inhomo-
geneities. Information from independent sources must
be considered to obtain more details on the climatic
pattern. The surface temperature changes inferred
from the inversion of underground temperature
measurements can supply an additional record of the
past climate signal. Pasquale et al. (1998, 2000) have
recently tested the applicability for northwestern Italy
of methods that utilise borehole temperature data to
infer the climatic history where meteorological data
are not available. On the basis of thermal information
from a selected geothermal borehole, they recon-
structed the climatic history since 1700, by employing
both inversion techniques and a forward modelling,
which combines air temperature time series from
conventional meteorological observatories with bore-
hole temperature profiles. The temperature signal
prior to the 1830s indicates a slightly warmer climate
compared to that of the 1990s and a sensibly
temperature rise during the last decade.
In this paper, we try to extend such an analysis to
outline the pattern of the recent climatic history at the
regional scale by studying the climatic signal stored in
geothermal boreholes scattered in central-northern
Italy. Inverse methods are used to infer the ground
surface temperature (GST) history from underground
temperatures recorded during the last 20 years, which
were partly available from the literature and partly
measured especially for this study. In particular, two
new datasets recorded in 1999–2000 are presented
together with results of thermal conductivity measure-
ments on borehole core samples.
Moreover, borehole temperature profiles were
combined with SAT records to obtain a baseline
temperature before the 19th century. Such linking
can give more reliable inferences about past temper-
ature variations than the meteorological time series
alone and can help to distinguish whether the
recently observed warming represents a natural
climatic variation or an induced anthropogenic
change of climate.
2. Underground temperatures
A number of boreholes have been drilled in the
last 20 years for surface heat-flux studies and
exploration of geothermal resources in central-north-
ern Italy. Underground temperatures can be affected
in principle by several kinds of thermal perturba-
tions, which deviates their shapes from that expected
for pure conduction heat transfer. Among the
available data, we have chosen only those temper-
ature–depth curves showing no or negligible thermal
perturbation, caused by groundwater circulation. The
selected dataset consists of eight temperature–depth
profiles, four of which were recorded in boreholes
located in the Tyrrhenian side of the Alps–Apennines
system and four in the Adriatic side (Fig. 1). The
boreholes are located in areas with subdued topog-
raphy and small elevation variations within a radius
of few kilometres, and are free of potential disturb-
ing effects of water masses. Their geographic
coordinates and logging time are shown in Table 1
while the temperature–depth distribution is shown in
Fig. 2.
2.1. Available data
Data of underground temperatures and thermal
conductivities for the borehole GH1, located in the
Tyrrhenian side of the northern Apennines, were
taken from a study by Verdoya et al. (1999). No
significant lithological variation exists along the
whole borehole, which is characterized by alternate
clay and marl. The recorded temperature profile
points to water flow in the 125–175 m depth range,
due to the presence of thin permeable layers. This
produced a slight perturbation on the background
thermal gradient. The temperature profile shown in
Fig. 2 includes a small correction within the
perturbed depth range, calculated according to the
approach of Ramey (1962). Laboratory results of
thermal conductivity measurements are available for
five marl samples collected near the borehole and
Fig. 1. Shaded relief map of central-northern Italy and location of the boreholes and the Meteorological Observatories whose thermal data were
used for the climatic study.
V. Pasquale et al. / Global and Planetary Change 47 (2005) 36–5038
four core specimens selected from the depth intervals
between 150 and 200 m. An average value of
2.2F0.1 W m�1 K�1 was determined (Chiozzi et al.,
1998).
Table 1
List of the investigated boreholes together with year of the
temperature measurements
Site Well Year Lat. N Long. E Alt.
(m)
Depth
(m)
Ref.a
Acqui Terme GH1 1996 448 40.6V 088 27.2V 152 200 1
Torniella GH2 1993 438 04.4V 118 11.6V 350 145 2
Fontalcinaldo GH3 2000 438 07.0V 118 29.4V 720 358 3
Murci GH4 1999 428 45.3V 118 24.3V 380 200 3
Imola GH5 1982 448 19.4V 118 38.1V 78 150 4
S. Marino GH6 1982 438 58.1V 128 28.2V 110 200 4
Rapagnano GH7 1981 438 10.1V 138 36.0V 120 190 4
Giulianova GH8 1982 428 42.5V 138 51.8V 105 200 4
a 1, Verdoya et al. (1999); 2, Baldi et al. (1994); 3, this paper; 4,
Mongelli and Zito (1988).
The lithological sequence observed in the bore-
hole GH2, located in the Tyrrhenian side south of
Florence, primarily consists of shales, and seconda-
rily of marly limestones and limestones both in the
uppermost 40 m and below 105 m. The rest of the
stratigraphy is characterized by prevailing limestones
and marly limestones, with minor layers of shales.
Several samples were collected for thermal conduc-
tivity determinations, and the resulting geometric
mean value is 2.2F0.2 W m�1 K�1 (Baldi et al.,
1994).
Boreholes located in the Adriatic side (GH5–GH8)
were drilled into clayey formations and the measured
underground temperatures showed that they were not
affected by relevant non-conductive disturbances
(Mongelli and Zito, 1988). The section consists of
clays and sandy clays in borehole GH5, located to the
south of Bologna, with the mean thermal conductivity
value of 1.8F0.3 W m�1 K�1. The hole GH6 is 200 m
Fig. 2. Temperature–depth distribution in the boreholes whose location is shown in Fig. 1.
V. Pasquale et al. / Global and Planetary Change 47 (2005) 36–50 39
deep. It crossed marly clays with silt intercalating very
thin sandy levels. Thermal conductivity determined
for core samples selected at 30–35 m one from each
other is 1.4F0.1 W m�1 K�1.
The boreholes GH7 and GH8 are located south of
Ancona. GH7 crossed silty clays down to 75 m and
subsequently sandy clays down to the bottom at 190
m. The hole GH8 has uniform lithology, consisting of
silty clays, excepting a thin lens of conglomerate at
the depth of 94 m. The mean thermal conductivity
determined on cores recovered every 25–30 m for the
Table 2
Underground temperatures observed in the borehole GH3
Depth (m) Temperature (8C) Depth (m) Temperature (8C) D
30 11.99 115 16.69 2
35 12.23 120 16.96 2
40 12.44 125 17.29 2
45 12.67 130 17.60 2
50 12.89 135 17.88 2
55 13.15 140 18.14 2
60 13.45 145 18.42 2
65 13.57 150 18.67 2
70 13.89 155 18.89 2
75 14.21 160 19.28 2
80 14.53 165 19.58 2
85 14.86 170 19.88 2
90 15.15 175 20.16 2
95 15.41 180 20.44 2
100 15.77 185 20.74 2
105 16.08 190 20.99 2
110 16.38 195 21.36 2
two holes is 1.7F0.1 and 1.8F0.1 W m�1 K�1,
respectively (Mongelli et al., 1983).
2.2. New thermal measurements
Two boreholes located in the Tyrrhenian side of the
Apennines chain (GH3 and GH4) were logged
specifically for this study. Temperature measurements
were carried out at 5-m intervals by means of a
precision temperature acquisition system with a 4-
wire shielded cable and equipped with a Pt-resistance
epth (m) Temperature (8C) Depth (m) Temperature (8C)
00 21.63 285 26.75
05 21.96 290 26.95
10 22.27 295 27.19
15 22.60 300 27.47
20 22.92 305 27.70
25 23.25 310 28.02
30 23.55 315 28.35
35 23.84 320 28.76
40 24.16 325 28.97
45 24.46 330 29.18
50 24.76 335 29.39
55 25.07 340 29.60
60 25.35 345 29.80
65 25.66 350 29.98
70 25.96 355 30.16
75 26.26 358 30.34
80 26.51
Fig. 3. Thermal gradient along the boreholes GH3 and GH4. Dots indicate average values for rock thermal conductivity and bars are standard
deviations.
V. Pasquale et al. / Global and Planetary Change 47 (2005) 36–5040
sensor. The total system uncertainty ranges from 0.01
to 0.03 8C.Table 2 shows the data recorded in the borehole
GH3, which crosses flysch-type sediments down to
260 m depth, consisting of shales and marls
alternating limestones, marly limestones and arena-
Table 3
Underground temperatures observed in the borehole GH4
Depth (m) Temperature (8C) Depth (m) Temperature (8C) D
30 18.07 75 23.75 1
35 18.97 80 24.54 1
40 19.73 85 25.20 1
45 20.38 90 25.88 1
50 20.94 95 26.63 1
55 21.47 100 27.47 1
60 22.00 105 28.34 1
65 22.54 110 29.11 1
70 23.06 115 29.97 1
ceous limestones, followed by breccias anhydrites
and phyllites down to the hole bottom (358 m). The
thermogram outlines an average geothermal gradient
of 58 mK m�1 in the uppermost 240 m. At larger
depths it varies from a minimum of 44 and a
maximum of 74 mK m�1 at the lithology change
epth (m) Temperature (8C) Depth (m) Temperature (8C)
20 30.72 165 36.54
25 31.41 170 37.03
30 32.10 175 37.50
35 32.78 180 37.99
40 33.42 185 38.56
45 34.09 190 39.16
50 34.75 195 39.78
55 35.40 200 40.45
60 36.02
V. Pasquale et al. / Global and Planetary Change 47 (2005) 36–50 41
which marks the transition between flysch and
breccias, at about 310 m depth. Thermal conduc-
tivity, determined on core samples recovered from
200–300 m depth, has an average value of 2.5F0.7
W m�1K�1, and it tends to be greater when the
geothermal gradient is lower (Fig. 3).
The borehole GH4 reached the depth of 200 m
and encountered shales alternating limestones and
flysch-type levels, more or less hard. Table 3 shows
temperature data, while Fig. 3 depicts the temper-
ature gradient and the thermal conductivity. Only
four cores collected at 100, 130, 160 and 190 m
depth were suitable for laboratory analysis, which
gave a weighted mean thermal conductivity of
2.5F0.15 W m�1 K�1. The average thermal
gradient is very high (130 mK m�1), and the
maximum value (171 mK m�1) is observed at 100
m depth, where the thermal conductivity is lower
(1.7F0.5 W m�1 K�1). The maximum thermal
conductivity (3.7F0.2 W m�1 K�1) corresponds to
Fig. 4. Ground surface temperature (GST) change from inversion of
the calcareous level showing the lowest thermal
gradient of 96 mK m�1.
3. Inversion of data and simulations
The functional space inversion (FSI) technique
was used to calculate the GST histories of the
selected boreholes (Shen and Beck, 1991, 1992).
Data processing was preceded by the evaluation of
different theoretical GST histories to find out the
optimal parameters of the inversion algorithm in
relation to the characteristics of the available bore-
holes, the amplitude and duration of the expected
climatic variations. To this purpose, different simple
models of climatic change have been constructed and
the effect on the underground temperature has been
simulated.
Similarly to other inversion techniques, the FSI
method assumes that heat transfer is only by
temperature–depth data in the boreholes GH1 and GH3–GH8.
Fig. 5. DT observed values (dots) of the borehole GH2. Dashed line
is the theoretical curve calculated from Eq. (1) for a sinusoida
ground temperature model fitting the residual temperature (bold
curve in the inset). For comparison, the GST obtained with FSI (thin
line) is also shown in the inset.
V. Pasquale et al. / Global and Planetary Change 47 (2005) 36–5042
conduction through a one-dimensional medium. The
method requires a priori estimates of the GST
variations (assumed to be zero in our calculations),
the thermophysical parameters and the heat flux,
which all are adjusted simultaneously in the course
of the inversion. Uncertainties in these parameters
are accounted for in a form of a priori standard
deviations (S.D.). This is especially important for
proper treating of the possibly unrecognised steady-
state noise in the measured temperature profile
arising from heterogeneity in the thermal conductiv-
ity, which is a crucial parameter in the GST
reconstruction. Geology at the investigated boreholes
shows for the layered structure negligible 3-D effects
on the subsurface temperature field. Direct thermal
conductivity determinations were available for the
investigated boreholes and a priori S.D. of 0.5 W
m�1 K�1 was assumed. Uniform thermal diffusivity
of 10�6 m2 s�1 and heat production of 1.0 AW m�3
were considered.
In general, inversions of underground temperatures
of the Tyrrhenian side (boreholes GH1, GH3 and
GH4) and of Adriatic side (GH5–GH8) gave con-
sistent results and clearly outline a different climatic
history for the western and eastern edge of the Italian
peninsula (Fig. 4). Results for borehole GH2, shown
in Fig. 5, deviate from the general trend inferred in the
neighbouring boreholes of the Tyrrhenian side, point-
ing to some local effect.
The relative shallow depth of GH2 implies that
the temperature data mainly contain information on
climate variation in the last 100 years. Moreover it
must be stressed that, due to diffusion, FSI can
recover only reasonable long-period features of the
GST history from underground temperature data,
whereas it is unable to resolve short wavelength
events in the distant past (Shen et al., 1992; Harris
and Chapman, 1998). Therefore, in order to gain a
finer detail of the climatic signal stored in this
borehole, a different inversion technique was
applied. We found that the temperature log of GH2
in the 20–80 m depth range matches a surface
temperature harmonic oscillation with a period
P=2p/x and amplitude DTo (Fig. 5). The under-
ground effect of such a variation is given by
DT z; tð Þ ¼ DTocoshexp � gð Þ ð1Þ
l
where h=(xt�g), g=z (x/(2v))1/2, the thermal
diffusivity is v=10�6 m2 s�1 and t is the time
since the variation is maximum. By differentiating
Eq. (1), one obtains the temperature gradient
distribution:
DC z; tð Þ ¼ DTogz
sinh � coshð Þexpð� gÞ ð2Þ
The wavelength k and the propagation velocity v
of the perturbation are given by k=2(pvP)1/2 and
v=2(pv/P)1/2. When z=k, DT(k, t)=0.002 DTo cos 2p[(t/P)�1] (Pasquale et al., 1996). From Eqs. (1) and
(2) it follows that
cosh ¼ 0 for DT z1; tÞ ¼ 0ð ð3Þ
and
cosh ¼ sinh for DC z2; tð Þ ¼ 0 ð4Þ
V. Pasquale et al. / Global and Planetary Change 47 (2005) 36–50 43
from which one obtains
xt � z1ðx= 2vð ÞÞ1=2 ¼ p=2 ð5Þ
and
xt � z2ðx= 2vð ÞÞ1=2 ¼ p=4 ð6Þ
where z1 is the first depth at which DT(z, t)=0 and z2is the first depth at which DC(z2, t)=0 (Beck, 1982;
Mongelli and Zito, 1988). Using the thermal diffu-
sivity, z1 and z2, being known from experimental data,
one determines x and t. The value of DTo is
calculated from Eq. (1) by using the DT(z, t) observed
values. By considering a linear pattern of the temper-
ature profile in the range 80–145 m, the background
temperature gradient is 115.6 mK m�1 and the
extrapolated surface temperature is 12.0 8C. The
climatic reconstruction results in a sinusoidal-type
change with amplitude 1.5 K, period 32 years and
time t=21 years.
This type of analysis was also applied to GH1
focussing on the most recent climatic history at this
site. The temperature profile observed in the upper-
most 80 m of the borehole is fitted by a linear change
Fig. 6. As in Fig. 5 for the borehole GH1. The ground temperature
change (in the inset) is of linear type (Eq. (7)).
in the surface temperature of the type DTo=bd t, where
t is the time and b is a constant (Fig. 6), whose
corresponding variation with depth DT(z, t) is given
by (Pasquale et al., 2000)
DT z; tð Þ ¼ bt
�erfc
z
2ffiffiffiffivt
p � zffiffiffiffivt
p ierfcz
2ffiffiffiffivt
p�
ð7Þ
and the integral of the error function is
ierfcz
2ffiffiffiffivt
p ¼ 1ffiffiffip
p exp� z2
4vt� z
2ffiffiffiffivt
p erfcz
2ffiffiffiffivt
p ð8Þ
If the lowermost (80–200 m) quasi-linear section of
the temperature log is considered to be unaffected by
the recent changes in surface air temperature, the least
squares fitting yields a background temperature
gradient of 92.6 mK m�1 and an extrapolated surface
temperature of 12.6 8C. The temperature increase
which fits better the disturbance, i.e. minimizing the
sum of the square differences between the theoretical
and observed borehole temperature data, is that for
DTo=1.6 K and t=12 years.
4. Pre-observational mean temperature
4.1. SAT records
A comparison with the surface air temperature
(SAT) records is useful to evaluate the reliability of
the foregoing interpretations and to which extent
inversion techniques yield meaningful results. There
is a number of historical temperature time series in
Italy. The homogenization of some well-known series,
such as Bologna, Genoa, Rome, etc., previously
stored on paper archives, began in the 1970s and still
continues (Brunetti et al., 2004). Temperature records
have been homogenized taking into account possible
changes in station location and instrumentation,
observational schedules, existing gaps, etc. Records
obtained in such a way were validated with different
statistical tests. Further efforts for widening this
database began in the second half of the 1990s by
including more series and improving the available
database, with the aim of a more precise partitioning
of Italy into different climatic sub-regions. At present
homogenized series of air temperature measurements
are available for central-northern Italy, in particular
Fig. 7. Annual surface air temperature change and its trend of the Tyrrhenian side from the Genoa and Florence meteorological time series in the
period 1814–2000. The curve (polynomial of degree 3) is the best estimate with respect to the average value of the whole recording period.
V. Pasquale et al. / Global and Planetary Change 47 (2005) 36–5044
since 1814 for the Florence Ximeniano Observatory
and for the Bologna University Observatory (Brunetti
et al., 2001, 2004), and since 1833 for the Genoa
University Observatory (Pasquale et al., 1998).
The climate history of the Tyrrhenian side can be
outlined by the surface air temperature change as
inferred from the Genoa and Florence data
expressed as departure from the average value over
the period 1814–2000 (Fig. 7). The trends of the
Genoa and Florence series are similar (Fisher’s
correlation coefficient of 0.87F0.08), thus, the
averaged SAT series in Fig. 7 can be regarded as
representative of the regional temperatures at least
for a 200-km long band on the Tyrrhenian side. The
best fitting polynomial curve shows a minimum
centred at the 1910s and larger temperature values
Fig. 8. As in Fig. 7 for the Adriatic side, from Bologna meteorological time
is the best estimate with respect to the average value of the whole record
at the present-day. Within this general trend, there
are two periods, about 10 year long, in which the
maximum positive departures are 2.5–3.0 K.
Fig. 8 shows the surface air temperature variation
recorded in Bologna (Adriatic side). The polynomial
curve fitting the mean departures over the period
1814–1982 shows a minimum value of 0.5 K centred
in 1860 and a maximum of about 0.25 K in 1945.
The maximum differences from the mean trend are
about 1.7 K in 1822 and �2.4 K in 1856. A
comparison of this series with the dendroclimato-
logical temperature estimates for NW Italy (Marti-
nelli, 2004) also shows significant coincidence,
implying that the SAT record of Bologna may be
representative of a wide region of the Po Basin and
the northern Adriatic Sea.
series in the period 1814–1982. The curve (polynomial of degree 3)
ing period.
Fig. 9. Reduced (dots) and calculated temperatures from combined
meteorological and geothermal data used to infer the POM
temperature for GH1 and GH5–GH8.
V. Pasquale et al. / Global and Planetary Change 47 (2005) 36–50 45
These SAT records represent the most reliable
regional information since the beginning of the 1800s.
However, it should be stressed that examined SAT
series show pronounced interdecadal variability,
which can hide the possible warming trends of the
20th century. Another bias can be due to the fact that
the homogenization of the time series was restricted
only to stations on the Italian territory. The main
problem of this procedure is that it does not detect the
inhomogeneities arising from changes in the national
standards of measurements. As discussed by Brunetti
et al. (2004), most of such inhomogeneities could
affect the end of the 19th century, when in many
observatories the standards of equipment position
were changed, and the period of World War II, which
may show gaps in records. According to Bohm et al.
(2001), the influence of these possible hidden
inhomogeneities can cause an underestimation of
temperature trends. Thus, borehole temperature logs,
containing independent non-meteorological data
source, jointly analysed with SAT records, may
provide not only useful information about the long-
term mean temperature for the period prior to the
onset of meteorological observations, but also the
basis for checking this kind of inhomogeneity.
4.2. Assessing POM
Chisholm and Chapman (1992) first tested the
consistency between meteorological and geothermal
data. Since then their joint analysis has been applied
to numerous paleoclimatic reconstructions, thus pro-
viding reliable inferences about temperature variations
over the 20th century (e.g. Pasquale et al., 1998; Bodri
et al., 2001). Pre-observational mean temperature
(POM) is the long-term mean surface temperature
against which 19th and/or 20th century warming can
be referenced. The POM values can be obtained by
comparing the measured borehole temperatures with
the synthetic temperature–depth profiles calculated
under a pure conductive thermal regime, when the
SAT series are used as a forcing function (surface
boundary condition).
The main assumption of this method is that the
ground and air temperatures track. Thus, the POM
value, calculated from the SAT and ground temper-
ature records, reflects the climate change in the
neighbourhood of the borehole. In presence of non-
climatic disturbances, the obtained POM can be
biased. Non-climatic sources for decoupling of the
air–ground temperatures were investigated in numer-
ous works (e.g. Chisholm and Chapman, 1992;
Harris and Chapman, 1998; Beltrami, 2001). In
view of the insignificance of topographic corrections
and groundwater flow in our boreholes, the air–
ground temperature differences might in principle
arise from temporary changes in the near surface
Table 5
As in Table 4, but POM is the long-term mean temperature before
the year 1900
Borehole POM (8C) D (K) ms (K)
GH1 15.58 0.24 0.128
GH5 13.81 0.09 0.041
GH6 13.93 �0.03 0.066
GH7 13.63 0.27 0.036
GH8 14.01 �0.11 0.095
V. Pasquale et al. / Global and Planetary Change 47 (2005) 36–5046
conductive properties. However, as the examined
boreholes were drilled in consolidated rocks, the
subsurface layers cannot be affected by significant
transient variations in moisture content and, corre-
spondingly, in conductive properties. Effect of soil
freezing (e.g. Beltrami, 2001) is similarly insignif-
icant, since it occurs rarely, is short lived and
changes randomly each year.
The change of vegetation and land use in the
neighbouring of the investigated boreholes could
instead produce significant temperature anomalies
(e.g. Majorowicz and Skinner, 1997), which do not
reflect regional air temperature variations. However,
the quantitative description of the nature and the
magnitude of such microclimatic effects is a difficult
problem. There are not many investigations on this
topic (see Bodri et al., 2001, and the references
therein), and in most cases historical changes of the
land use are poorly documented.
In order to calculate POM values the standard
least-square inversion analysis that minimizes the sum
of squared differences between reduced and synthetic
temperature–depth profiles was used (for details of
calculus see Bodri et al., 2001). The POM values were
calculated for GH1 and GH5–GH8. As a forcing
function, we applied the SAT of Genoa and Florence
to the borehole GH1 and Bologna to GH5–GH8. The
POM values were calculated for the period prior to the
beginning of SAT records as well as before 1900.
Boreholes GH3 and GH4 were measured between
1999 and 2000, while the homogeneous SAT records
exist only until 1994. Since the disagreement of the
end of SAT record and the date of borehole logging
can bias the POM estimates (Harris and Chapman,
1997), these boreholes were excluded from consid-
eration. As seen in Fig. 2, the bulk of the observed
Table 4
POM temperatures and rms misfit between the reduced temperatures
and the best fitting SAT–POM synthetic model
Borehole POM (8C) D (K) rms (K)
GH1 15.59 0.23 0.128
GH5 13.76 0.14 0.051
GH6 13.86 0.04 0.074
GH7 13.62 0.28 0.032
GH8 13.90 0.00 0.095
POM temperature is the long-term mean before the beginning of
SAT record. D is the amount of warming as a deviation of the POM
temperature from the 1930–1970 SAT mean.
borehole temperatures represents a quasi-steady-state
geothermal field. The POM values were calculated
from the reduced temperatures, which represent the
departure of the observed temperature–depth data
from the steady-state conditions. Results of reduced
and best-fit calculated temperatures are presented in
Fig. 9.
Results of POM calculations are summarized in
Tables 4 and 5, where we present the obtained values,
relative amounts of temperature change and the values
of the rms misfit. The latter value represents the
measure of deviations of the curve calculated with the
mean squares method from the measurements, and
thus characterizes the goodness of fit. Similarly to the
foregoing GST reconstructions, POM results indicate
difference in the climatic history between the Tyr-
rhenian and the Adriatic sides. In the Tyrrhenian side,
the POM temperature before the beginning of SAT
records and before 1900 is the same, indicating
warming by 0.23–0.24 K. All the boreholes of the
Adriatic side give generally coincident POM values,
which show, on average, slight warming by 0.12 and
0.06 K since 1814 and 1900, respectively.
5. Leftover temperatures
The calculated synthetic temperatures faithfully
match the reduced temperature pattern only for
borehole GH7 (Fig. 9). An rms misfit of ~0.03 K
can be regarded as a result of the noise in the
measurements. Larger values, like for boreholes GH6
and GH8, mean that at least a part of the transient
reduced temperatures cannot be explained by the
regional climatic changes, and may reflect other
disturbing factors. Excepting the foregoing possible
problems due to the hidden inhomogeneities in SAT,
local microclimatic effects could cause the incon-
sistency between the reduced and calculated temper-
V. Pasquale et al. / Global and Planetary Change 47 (2005) 36–50 47
atures. These can be due to a variety of factors, such
as variations of agricultural activities or vegetation
cover and urbanization in the surroundings of the
borehole site.
In order to move beyond the regional climatic
contribution and to identify additional local compo-
nents, causing the existing misfits, we investigated the
leftover reduced temperature (Ta), defined as
Ta(z)=Tr(z)�Tc(z), where Tr is the reduced temper-
ature and Tc is the calculated synthetic best fit POM–
SAT temperature (for details, see Cermak et al., 1992;
Cermak and Bodri, 2001). We performed numerous
trial runs with synthetic data, which showed that
inversion of the leftover temperatures can provide
information on the local climatic history.
The significance of the leftover temperatures can
be demonstrated with a simple, synthetically gener-
ated, example of noise-free temperature log for a
homogeneous half-space with thermal diffusivity
v=10�6 m2 s�1. Fig. 10 shows the synthetic reduced
temperature with an accuracy of 0.01 K (truncation
error) at 5 m intervals to the depth of 500 m
corresponding to a surface temperature increase by
0.5 8C during the last 50 years. Let us suppose an SAT
series to be available showing no temperature increase
in the last 100 years, i.e. not containing the recent 50
yearlong warming. In this case, the best-fit curve,
obtained with a constant forcing function, is for a
Fig. 10. (A) Reduced temperatures due to a surface temperature increase b
value of �0.42 K considering an SAT series with no change of surface tem
profile (for details see text).
POM value of �0.42 K. In other words, it would be
inferred warming diffused through the whole obser-
vational period and having somewhat weaker ampli-
tude, than that really occurred. However, despite of
smoothing and diffusion, the POM value gives a
meaningful estimate of the warming occurred during
the observational period.
The differences between the synthetic reduced
temperature and the best-fit curves (leftovers) are
within F0.08 K (Fig. 10 b). Negative temperature
indicates a cooling event, whose characteristics can be
roughly estimated with the formulae developed by
Lachenbruch (1994) and tested by Harris and Chap-
man (1998). The method allows calculating the
parameters of a step change of temperature, also
called bboxcarQ event, from the maximum leftover
temperature Tm, its depth zm and the depth of the
bottom anomalous temperature z2, corresponding to
1% of the surface temperature change. The onset time
of the bboxcarQ event is t2=z22/8v, its duration is D=z2
2/
4v�zm2 /v and the amplitude of the temperature change
is DT=8 Tmzm2 /(z2
2�4zm2 ). In our example zm=69 m,
z2=160 m, and Tm=0.056 K. One obtains that the
onset of cooling occurred 101.6 years ago, its duration
is 52 years and the amplitude �0.33 K. It appears that
the combination of the POM and leftover temper-
atures gives a quite real surface temperature trend. For
the field data, the FSI method can be formally applied
y 0.5 8C in the last 50 years (thick curve) and calculated for a POM
perature in the last 100 years (thin curve). (B) Leftover temperature
V. Pasquale et al. / Global and Planetary Change 47 (2005) 36–5048
on leftover temperature profiles to reconstruct more
complex GST histories.
The leftovers for the boreholes GH1 and GH5–
GH8 are presented in Fig. 11. The most noticeable
leftovers occur in the uppermost interval above 50 m
depth, and range from �0.65 to 0.53 K. The leftover
values decrease rapidly below this depth. In the 50–
200 m interval, leftover temperatures are generally
within F0.1 K and their average value is 0.05 K, thus
comparable with the precision of the temperature
logging. The occurrence of noticeable disturbances
mainly in the uppermost parts of the borehole logs can
be interpreted in terms of recent, local climatic
changes.
The rough estimate of the beginning and amount of
GST local changes can be obtained from inversion of
the Ta profiles by means of FSI. Due to the relatively
high bgeologicQ noise of the available temperature–
depth data and the shallow depth of the boreholes,
results of inversion were considered of significance
only for the last 60–80 years prior to the thermal
logging. All the boreholes indicate a temperature
decrease of 0.8–2.0 K from 1950 to 1960 and since
Fig. 11. Leftover reduced temperatures f
the end of the 1970s. The amount of cooling tends to
increase towards the south. Cooling was preceded by
warming at a rate of 0.3–0.4 K year�1 at boreholes
GH6 and GH8 and a relatively constant warmer
period at borehole GH5, with temperature 0.4–0.6 K
higher than that of the 1970s.
Since the local microclimatic influences have
opposite sign, they can obscure the large-scale
warming trend. There exist different more or less
effective techniques (e.g. averaging over all stations
into single time series) to soften the impact of local
factors and to obtain more general temperature
patterns for a region only from the SAT data. The
advantage of joint processing of SAT and borehole
temperature logs is its ability to distinguish local and
regional climatic changes, which cannot be achieved
when one uses these databases separately.
The local climatic influence is less significant for the
borehole GH1, in the Tyrrhenian side (Fig. 11). The
leftover temperature profile shows local warming of
about 0.4 K in the last 20 years, additional to that
inferred from the SAT records. The amount of warming
for borehole GH1 is almost identical to that of GH7
or boreholes GH1 and GH5–GH8.
V. Pasquale et al. / Global and Planetary Change 47 (2005) 36–50 49
(Tables 4 and 5), where the local influence has been
insignificant and more than 90% of observed transient
borehole temperature signal is accounted for by the
SAT history. In both cases POM values obtained for
periods before 1814–1833 and/or 1900 are identical,
which suggests that the whole amount of warming can
be attributed to the 20th century alone.
6. Discussion and conclusions
Our results argue for different patterns in the GST
histories since 1750 between the Tyrrhenian and
Adriatic sides. In the former, the four investigated
boreholes show the same trend. The temperature has
always been lower than that of the 1990s, with
minimum values of �0.5–0.8 K from 1930 to 1960.
It was preceded by a slow decrease of temperature
since 1840 and it has been followed by a warming
phase at a high rate up to now. These results of a rapid
recent warming confirm those inferred for northwest-
ern Italy by Pasquale et al. (1998, 2000).
The warming phase is put into better evidence by
examining in detail the temperature signal in the
uppermost section of the borehole GH1 (Fig. 6). The
climatic model consisting of a linear increase in the
surface temperature during the last 10 years at a rate
of about 0.1 K year�1 accounts for the shallower part
of the underground temperature profile.
The temperature disturbances observed in the
borehole GH2 are well fitted by a short period change
in the ground surface temperature, which shows a
harmonic oscillation during the last three decades.
This trend can hardly be resolved by the FSI method,
which gave quite unstable results for this case. Such
local periodic variations in the surface temperature
could be attributed to cycles of tree cutting followed
by slow re-growth of vegetation.
On the Adriatic side, the magnitude of the inferred
climatic change differs from that of the western side of
the Apennines chain. FSI results indicate that GST has
always been higher than that of the 1980s. After a
warmer period ending in 1940, there has been cooling
until the time of the temperature measurements.
During the latter phase, the temperature has decreased
at an average rate of �0.03 K year�1. Moreover,
inversion shows for boreholes GH5 and GH8 the
beginning of a warming phase between 1970 and
1980. This phase has been also recognised by
Mongelli and Zito (1988).
It appears that all reconstructed GST histories fit
those of the nearby meteorological observatories
records reasonably well, within the well known
limitation that a direct comparison of the borehole
temperature profiles and surface air temperature
records is difficult. Ground temperatures are generally
warmer than air temperatures by 0.5–2.0 K. In
particular, Murtha and Williams (1986) showed that
the vegetation cover as well as the moisture regime
may have a significant effect on the soil temperature,
which could be higher by as much as some degrees
than the mean annual temperature.
By usingminimum andmaximum daily temperature
from 1865 to 1995, Brunetti et al. (2000) obtained a
similar trend of the mean annual surface air temperature
in northern and southern Italy. On the contrary, our
results put into evidence different climatic histories
between the Adriatic and Tyrrhenian sides. The barrier
effect of the N–S trending Apennines ridge and the
destabilising effect of the Mediterranean Sea, which
tend to facilitate the genesis of cyclones west of the
mountain range, are the two main geographical factors
that may explain these different trends.
The results of the POM calculations show that the
warming obtained for the borehole GH1 at the
Tyrrhenian side coincides well with the amount of
warming obtained by Maugeri and Nanni (1998) for
northern Italy. The equality of POM temperatures
obtained for periods before 1833 and 1900 suggests
that all the observed warming occurred in the 20th
century. In the Adriatic side, approximately, half of the
obtained warming occurred already at 19th century.
Even if processing of data from more boreholes
would be recommendable for reaching definitive
conclusions, the available data hint that observed
difference in the course of the climatic history may
have been caused by local climatic changes in the
Adriatic side. The existence of such additional
climatic changes, which cannot be attributed to the
change of climate represented by the SAT observation,
deserves certain explanation. Due to the lack of
detailed information on the land-use processes in
Italy, the reasons for local climate changes can only be
speculated upon. Anyhow, since the long-term mon-
itoring and analysis of indicators of local environ-
mental change, which are sensitive to changes in
V. Pasquale et al. / Global and Planetary Change 47 (2005) 36–5050
climate for most regions of the world, are generally
insufficient and/or absent at all, the investigation of
the leftover temperatures can be regarded as an
additional independent tool to estimate the response
of the local environment to anthropogenically induced
changes, and to reveal the impact of these factors.
References
Baldi, P., Bellani, S., Ceccarelli, A., Fiordelisi, A., Squarci, P., Taffi,
L., 1994. Nuovi dati geotermici nelle aree a Sud-Est ed a Sud
del campo geotermico di Travale (Toscana). Atti del 138Convegno GNGTS, C.N.R, Roma, pp. 211–221.
Beck, A.E., 1982. Precision logging of temperature gradients and
the extraction of past climate. Tectonophysics 83, 1–11.
Beltrami, H., 2001. On the relationship between ground temperature
histories and meteorological records: a report on the Pomquet
station. Glob. Planet. Change 29, 327–348.
Bodri, L., Cermak, V., Kukkonen, I.T., 2001. Climate change of the
last 2000 years inferred from borehole temperatures: data from
Finland. Glob. Planet. Change 29, 189–200.
Bfhm, R., Auer, I., Brunetti, M., Maugeri, M., Nanni, T., Schfner,W., 2001. Regional temperature variability in the European Alps
1760–1998 from homogenised instrumental time series. Int. J.
Climatol. 21, 1779–1801.
Brunetti, M., Buffoni, L., Maugeri, M., Nanni, T., 2000. Trends of
minimum and maximum daily temperatures in Italy from 1865
to 1996. Theor. Appl. Climatol. 66, 49–60.
Brunetti, M., Buffoni, L., Lo Vecchio, G., Maugeri, M., Nanni, T.,
2001. Tre secoli di meteorologia a Bologna. Ed. CUSL, Milano.
Brunetti, M., Buffoni, L., Mangianti, F., Maugeri, M., Nanni, T.,
2004. Temperature, precipitation and extreme events during the
last century in Italy. Glob. Planet. Change 40, 141–149.
Cermak, V., Bodri, L., 2001. Climate reconstruction from subsur-
face temperatures demonstrated on examples of Cuba. Phys.
Earth Planet. Inter. 126, 295–310.
Cermak, V., Bodri, L., Safanda, J., 1992. Underground temperature
fields and changing climate: evidence from Cuba. Palaeogeogr.
Palaeoclimatol. Palaeoecol. 97, 325–337.
Chiozzi, P., Pasquale, V., Verdoya, M., 1998. Heat-flux anomaly in
the Piedmont Tertiary Basin (NW Italy). Proceedings of the
International Conference bThe Earth’s Thermal Field and
Related Research MethodsQ 19–21 May, 1998, Moscow, Russia,
pp. 59–62.
Chisholm, T.J., Chapman, D.S., 1992. Climate change inferred from
analysis of borehole temperatures: an example from western
Utah. J. Geophys. Res. 97, 14155–14175.
Ghil, M., Vautard, R., 1991. Interdecadal oscillations and the
warming trend in global temperature time series. Nature 350,
324–327.
Harris, R.N., Chapman, D.S., 1997. Borehole temperatures and a
baseline for 20th century global warming estimates. Science
275, 1618–1621.
Harris, R.N., Chapman, D.S., 1998. Geothermics and climate
change: Part 2. Joint analysis of borehole temperature and
meteorological data. J. Geophys. Res. 103, 7371–7383.
Jones, P.D., New, M., Parker, D.E., Martin, S., Rigor, I.G., 1999.
Surface air temperature and its change over the past 150 years.
Rev. Geophys. 37, 173–199.
Lachenbruch, A.M., 1994. Permafrost, the active layer, and
changing climate. USGS Open File Report 94–694, 43 pp.
Majorowicz, J.A., Skinner, W.R., 1997. Potential causes of the
differences between ground and surface air temperature warm-
ing across different ecozones in Alberta, Canada. Glob. Planet.
Change 115, 79–91.
Martinelli, N., 2004. Climate from dendrochronology: latest
developments and results. Glob. Planet. Change 40, 129–139.
Maugeri, M., Nanni, T., 1998. Surface air temperature variations in
Italy: recent trends and an update to 1993. Theor. Appl.
Climatol. 61, 191–196.
Mongelli, F., Zito, G., 1988. Effect of recent temperature change on
shallow geothermal measurements. Geothermics 17, 765–776.
Mongelli, F., Ciaranfi, N., Tramacere, A., Zito, G., Perusini, P.,
Squarci, P., 1983. Contributo alla mappa del flusso geotermico
in Italia: misure dalle Marche alla Puglia. Atti del 28 ConvegnoAnn. Gruppo Nazionale di geofisica della Terra Solida, Rome,
pp. 737–763.
Murtha, G.G., Williams, J., 1986. Measurements prediction and
interpretation of soil temperature for use in soil taxonomy:
tropical Australian experience. Geoderma 37, 189–206.
Nanni, T., Lo Vecchio, G., Cecchini, S., 1998. Variability of surface
air temperature in Italy 1870–1980. Theor. Appl. Climatol. 59,
231–235.
Pasquale, V., Verdoya, M., Chiozzi, P., 1996. Climatic signal from
underground temperatures. Proceedings of the 24th Intern. Conf.
Alpine Meteor., ICAM 96, Bled, Slovenia. Hydrometeorological
Institute of Slovenia, Ljubljana, pp. 201–208.
Pasquale, V., Verdoya, M., Chiozzi, P., 1998. Climate change from
meteorological observations and underground temperatures in
Northern Italy. Stud. Geophys. Geod. 42, 30–40.
Pasquale, V., Verdoya, M., Chiozzi, P., Safanda, J., 2000. Evidence
of climate warming from underground temperatures in NW
Italy. Glob. Planet. Change 25, 215–222.
Ramey, H.J., 1962. Well bore heat transmission. J. Pet. Technol. 14,
427–435.
Shen, P.Y., Beck, A.E., 1991. Least squares inversion of borehole
temperature measurements in functional space. J. Geophys. Res.
96, 19965–19979.
Shen, P.Y., Beck, A.E., 1992. Paleoclimatic change and heat flow
density inferred from temperature data in the Superior Province
of the Canadian Shield. Glob. Planet. Change 6, 143–165.
Shen, P.Y., Wang, K., Beltrami, H., Mareschal, J.C., 1992. A
comparative study of inverse methods for estimating climatic
history from borehole temperature data. Glob. Planet. Change
98, 113–127.
Verdoya, M., Pasquale, V., Chiozzi, P., 1999. Hydrothermal
circulation in the Acqui Terme district, Tertiary Piedmont Basin
(NW Italy). Bull. Hydrogeol. 17, 193–200.