CLIMATE RESEARCH
Clim Res
Vol. 64: 123–140, 2015
doi: 10.3354/cr01304
Published online July 7
Transient high-resolution regional climate
simulation for Greece over the period 1960−2100:
evaluation and future projections
P. Zanis1,*, E. Katragkou1, C. Ntogras1, G. Marougianni2, A. Tsikerdekis1, H. Feidas1,
E. Anadranistakis3, D. Melas2
1
Department of Meteorology and Climatology, School of Geology, Aristotle University of Thessaloniki, Thessaloniki, Greece
2
Laboratory of Atmospheric Physics, School of Physics, Aristotle University of Thessaloniki, Thessaloniki, Greece
3
Hellenic National Meteorological Service, Ellinikon, Greece
ABSTRACT: A transient regional climate model simulation with a spatial grid resolution of 10 km
(RCM10), nested to a regional simulation with 25 km resolution (RCM25), was carried out over
Greece with RegCM3 for the period 1960−2100 under the IPCC A1B scenario. RCM10 precipitation and temperature fields depict the finer regional characteristics over the complex Greek terrain compared to RCM25, but a station-based evaluation for the period 1975−2000 does not reveal
a considerable improvement in RCM10 compared to RCM25. Future projections for the earlyfuture period 2021−2050 indicate small changes, with annual temperature increasing mostly over
land by less than 1.8°C and precipitation changing by ±15%, being mostly negative in the southern part of the domain. At the end of the century (2071−2100), the projected changes become
larger, with mean annual temperature increasing by about 3.4 to 4.2°C over land and by 2.6 to
3.4°C over the sea and precipitation decreasing by 10 to 40%, with a positive gradient from the
north to the south. Summer presents the largest future increase in mean near-surface temperature
over the Greek mainland, while winter and spring show the largest decreases in precipitation rate.
The number of hot days, warm nights, night frosts and continuous dry spell days and length of the
growing season are projected to increase slightly in the near-future period, but markedly and consistently in the late 21st century future period in accordance with the generally warmer and drier
climate projected from the RCM10 simulation.
KEY WORDS: Regional climate models · Greece · High resolution · Evaluation · Future projections
Resale or republication not permitted without written consent of the publisher
Policy and decision makers in governmental and
non-governmental organizations, as well as endusers in the private sector and the general public,
require detailed regional information on future climate to assess the risks of anticipated climate
changes due to the anthropogenic forcing of the
greenhouse effect. Although high-resolution global
climate models (GCMs) have been increasingly utilized over the last decade (Held & Zhao 2011), the
recent simulations with GCMs used in the IPCC
fifth assessment report (AR5) still have a coarse horizontal resolution to resolve the effects on regional
climate of local- and regional-scale forcings, such as
topographic characteristics with complex mountain
ranges, coastlines, peninsulas, small islands and
lakes, as well as land-use characteristics and chemical composition of short-lived species (e.g. aerosols,
tropospheric ozone). Although planetary- and synoptic-scale forcings and circulations determine the
statistics of weather events that characterize the cli-
*Corresponding author: zanis@geo.auth.gr
© Inter-Research 2015 · www.int-res.com
1. INTRODUCTION
124
Clim Res 64: 123–140, 2015
mate of a region, the above-mentioned regionaland local-scale forcings and circulations modulate
the regional climate and can possibly provide feedback to the large-scale circulation (Giorgi & Mearns
1999, IPCC 2007).
Hence, regional climate models (RCMs) have been
developed since the late 1980s for the application of
dynamical downscaling methods to enhance the regional information provided by the GCMs for past
and future climate simulations or by the large-scale
reanalysis fields in hindcast simulations (Dickinson
et al. 1989, Giorgi et al. 1990, Giorgi & Mearns 1999).
Extensive application of the dynamical downscaling
methodology with the use of RCMs has taken place
over the last decade (e.g. Giorgi et al. 2004, Gao et al.
2006, Christensen & Christensen 2007, Déqué et al.
2007, Jacob et al. 2007, Sanchez-Gomez et al. 2009,
Rauscher et al. 2010), and the capabilities and limitations of the methodology have been investigated
(Giorgi & Mearns 1999, Laprise et al. 2008, Xue et al.
2014 and references therein).
The use of RCMs is necessary in regions with multiple topographic characteristics; Greece is a Mediterranean country characterized by complex topography, with steep orography from the mountainous
regions to the coast, a long and convoluted coastline
and many small islands in the Aegean and Ionian
seas. Hence, the use of downscaling to higher resolution is necessary to assess the regional and subregional climate of the complex topographical area of
Greece (Kostopoulou et al. 2007, Tolika et al. 2008,
2012, Zanis et al. 2009a). The highest commonly used
horizontal resolution in multidecadal to centennial
simulations with RCMs ranges between 50 and 25
km. But for mountainous regions, even 10 km, which
is roughly the highest horizontal resolution limit regularly used in hydrostatic models, can be considered
a coarse resolution (Im et al. 2010). Recently, a new
high-resolution regional climate change ensemble
with a horizontal resolution of 0.11° has been established for Europe within the World Climate Research
Program Coordinated Regional Downscaling Experiment (EURO-CORDEX) initiative (Jacob et al. 2014).
Over the eastern Mediterranean (EM) region, climate conditions are characterized by moderate air
temperatures and variable rainy weather during the
winter season, often related to eastward movement
of frontal systems (Maheras et al. 2001, Xoplaki et al.
2004). During spring, the mid-latitude baroclinicity
zone moves northwards, and in summer, the region is
subject to tropical influences leading to enhanced
subsidence, which suppresses clouds and rain. This
subsidence is primarily related to the interaction with
the mid-latitude westerlies of an equatorially trapped
Rossby wave to its west, induced by the South Asian
monsoon heating, as well as an enhancement of the
descent due to diabatic radiative cooling under clear
sky conditions (Rodwell & Hoskins 1996, Tyrlis et
al. 2013). Moreover, the geographic location of the
Greek peninsula, with the contrasts of maritime and
continental air masses, as well as its orographic features results in a distinctive climate variation from
west to east and from north to south (Maheras &
Anagnostopoulou 2003).
Furthermore, the whole Mediterranean is recognized among the most responsive regions to climate
change (Giorgi 2006). Model projections based on
GCMs and RCMs indicate a consistent warming and
drying of the Mediterranean region mainly over the
last decades of the 21st century for various emission
scenarios (e.g. Gibelin & Déqué 2003, Gao et al.
2006, Diffenbaugh et al. 2007, Goubanova & Li 2007,
Giorgi & Lionello 2008, Krichak et al. 2011, Lelieveld
et al. 2013).
Most of the previous RCM studies for the future
climate change of the EM or of specific countries of
the EM were based on simulations with horizontal
resolution down to 25 km. Projected climate change
over the EM and Turkey has been analyzed by
using a time-slice approach, with the reference
(1961−1990) and future (2071−2100) climate simulations produced by RegCM3 with a horizontal resolution of 30 km under the IPCC A2 scenario (Önol &
Semazzi 2009). They provided compelling evidence
of the added value of RCM downscaling in the
region based on countrywide averages of RegCM3
simulations being consistent with corresponding
averaged station data. Krichak et al. (2011) carried
out a double-resolution transient RCM climate
change simulation experiment with RegCM3 over
the period 1960−2060 under the IPCC A1B scenario
for a near-coastal eastern zone of the EM region,
with the inner domain having a horizontal resolution
of 25 km. They revealed a notable sensitivity of the
projected larger-scale climate change signals to the
smaller-scale effects. Hadjinicolaou et al. (2011)
assessed mid-21st century climate and weather
extremes in Cyprus as projected by 6 RCMs with a
horizontal resolution of 25 km under the IPCC A1B
scenario, which indicated a shift in the mean
climate to a warmer state, with a relatively strong
increase in warm extremes. Lelieveld et al. (2013)
presented similar model results from a transient
simulation over the period 1950−2099 under the
A1B scenario based on the RCM PRECIS with a horizontal resolution of ~25 km. Recently, Zittis et al.
Zanis et al.: High-resolution climate simulation for Greece
(2014) studied the role of soil moisture in the amplification of climate warming in the EM based on a
transient simulation with the Hadley Centre
HadRM3P RCM covering the period 1951−2099
under the A1B scenario with a horizontal resolution
of 25 km.
For Greece in particular, there are only a few
studies so far on future climate change based on
RCM simulations with spatial resolution down to
25 km. Zanis et al. (2009a) investigated the simulated changes in temperature and precipitation over
Greece for the period 2071−2100 under the A2
emission scenario based on time-slice projections
from 9 RCMs with a spatial grid resolution of 50 km
carried out within the PRUDEN CE project (http://
prudence.dmi.dk/). An assessment of regional climate change for Greece based on available RCM
simulations was presented by the Climate Change
Impacts Study Committee (Zerefos et al. 2011). Tolika et al. (2012) updated the assessment of future
climate change over Greece for summer and winter
seasons using 22 simulations from various RCMs
under the A2, A1B and B2 future emission scenarios
of the IPCC. The study was based on time-slice
projections with a spatial grid resolution of 50 km
carried out within the PRUDENCE project and transient 21st century projections with a spatial resolution of 25 km carried out within the ENSEMBLES
project. N astos et al. (2013) studied future changes
of aridity in Greece based on transient 21st century
projections from 8 RCMs with a spatial resolution of
~25 km carried out within the ENSEMBLES project
(www.ensembles-eu.org/).
Tselioudis et al. (2012) investigated whether the
increased resolution of an RCM from 50 to 11 km
introduces novel information in future precipitation
change over the mountainous region of Greece. So
far, this is the first study showing results of future
climate change with such a high spatial resolution
(of 11 km) for the region of Greece. However, the
study was based only on two 5 yr runs, a control run
(1991−1995) and a future climate run based on the
IPCC A1B scenario (2091−2095). Furthermore, Önol
(2012) presented for the first time a long-term highresolution (of 10 km) regional climate simulation
using RegCM3 for the EM including Greece, but
this simulation was a hindcast simulation driven by
N CEP reanalysis over the period 1961−2008. Here,
we present results of a high-resolution simulation
over the period 1960−2100 with RegCM3 that constitutes a transient regional projection for future climate change, for the first time to our knowledge,
with 10 km spatial resolution for Greece.
125
2. DATA AND MODEL SETUP
The high-resolution (10 km grid resolution) transient regional climate simulation (1960−2100) was
performed with the RCM RegCM3 over Greece
(hereafter referred to as RCM10) with 143 × 109 grid
cells horizontally, 18 vertical levels and the model top
at 50 mb. The domain and the model topography are
shown in Fig. 1. The RCM10 simulation was initialized in December 1958, with the first 2 yr of the simulation considered as the spin-up time. This high-resolution simulation was a nested run of the RegCM3
simulation with 25 km spatial resolution over an extended European domain carried out in the framework of the European Union project EN SEMBLES
(hereafter referred to as RCM25). In our centennial
RCM10 simulation case, the dynamical downscaling
ratio was 2.5 to 1, down to the highest horizontal resolution limit regularly used in hydrostatic RCMs, thus
reducing the gridbox surface from 625 to 100 km2. In
turn, the RCM25 simulation used lateral boundary
conditions from the T63 GCM ECHAM5 coupled to
the ocean model MPI-OM (run 3) with 1.875° resolution for the 20th century (Roeckner 2006) and the 21st
century under the IPCC Special Report on Emissions
Scenarios (SRES) A1B scenario (Roeckner 2008). The
20th century simulation (including year 2000) used
observed anthropogenic forcings of carbon dioxide
(CO2), methane (CH4), nitrous oxide (N2O), chlorofluorocarbons (CFCs), ozone (O3) and sulfate (SO4). The
A1B scenario is part of the A1 family, which describes
a balance across all energy sources. The ECHAM5
experiment was initialized in year 2000 of the 20th
44°N
43°
42°
Pindus
mountains
Macedonia
Thrace
41°
Epirus
40°
Thessaly
Central Aegean
sea
Greece
Ionian
sea
39°
38°
Peloponnese
37°
Crete
Rhodes
36°
35°
34°
16°E
0
18°
400
20°
800
22°
24°
1200
26°
1600
28°
30°
2100
Elevation (m a.s.l.)
Fig. 1. Topography of the Greek domain with a grid horizontal
resolution of 10 km. The regions of Greece are also indicated
126
Clim Res 64: 123–140, 2015
century run and continues until year 2100 with
anthropogenic forcings (CO2, CH4, N 2O, CFCs, O3
and SO42–) according to A1B. The A1B scenario predicts a medium-high increase of CO2 concentration
to about 700 ppm by 2100 (IPCC 2007). SRES scenarios have been used for the climate projections of the
IPCC fourth assessment report (AR4). However, the
recent climate projections in IPCC AR5 are based on
the most recent representative concentration pathways (RCPs). Comparison of CO2 concentrations for
the 21st century from the RCPs and SRES scenarios
shows that A1B lies between RCP6 and RCP8.5 but is
closest to RCP6 (Meinshausen et al. 2011). Rogelj et
al. (2012) also showed that the median of the projected warming for the end of the 21st century under
the A1B scenario is found between the values of the
RCP6 and RCP8.5 scenarios but nearer to RCP6.
RegCM3 was initially developed at the N ational
Center for Atmospheric Research (N CAR), and has
been used in a number of RCM in studies of regional
climate and seasonal predictability around the world
(Giorgi et al. 2006, Pal et al. 2007). The dynamical
core is based on the hydrostatic version of the Pennsylvania State University/N CAR mesoscale model
version 5 (Grell et al. 1994). The radiative transfer
package is taken from the community climate model
version 3 (Kiehl et al. 1996). The large-scale cloud
and precipitation computations are performed by the
subgrid explicit moisture scheme (Pal et al. 2000).
The biosphere−atmosphere transfer scheme (Dickinson et al. 1993) is used to represent surface processes, while boundary layer physics is described via
the non-local vertical diffusion scheme of Holtslag et
al. (1990).
The RCM25 simulation over Europe, which provided the lateral boundary conditions for the RCM10
simulation, used the Grell scheme (Grell 1993) for
convection, implementing the Fritsch & Chappell
(1980) closure assumption (hereafter referred to as
Grell-FC). However, a sensitivity study of RegCM3 to
the convective scheme for southeastern Europe
showed that a cold bias with Grell is significantly reduced when the Emanuel convective scheme (Emanuel & Živković-Rothman 1999) is used, even during
months of low convective activity (Zanis et al. 2009b).
Furthermore, in a previous study for central Europe
with RegCM3, high-resolution simulations with the
default Grell-FC scheme tend to significantly overestimate precipitation for the mountainous area of
the Carpathians (Torma et al. 2011). Torma et al.
(2011) thus suggested an adjustment of some parameters in the Grell-FC scheme to reduce precipitation
in the nested simulation such as the cloud-to-rain
autoconversion rate, raindrop evaporation rate coefficient and raindrop accretion rate. Furthermore,
Önol (2012) carried out a high-resolution regional climate simulation (10 km) nested in a 50 km RegCM3
simulation with the default Grell-FC scheme for the
EM, driven by N CEP reanalysis over the period
1961−2008. In this study, the nested high-resolution
simulation reveals a clear overestimation of precipitation over the mountainous regions of Greece during winter and spring in comparison to gridded
datasets (see Fig. 6).
Hence, it was necessary to configure the convection scheme to optimise RegCM3 for our high-resolution simulations over Greece. The model configuration was chosen after evaluating 6 yearly sensitivity
simulations with different convection schemes, as
described in more detail by Mystakidis et al. (2012).
Specifically, the simulations comprised 2 experiments using the Grell-FC scheme and 4 experiments
using the Emanuel convective scheme. The 2 simulations with the Grell-FC scheme included one simulation with the default version and the other simulation
with the modification suggested by Torma et al.
(2011). The 4 experiments with the Emanuel scheme
include changes in the relaxation rate a (kg m−2 s–1
K–1), which determines the rate at which the cloudbase upward mass flux is relaxed to steady state, and
in the warm cloud autoconversion threshold l0 (kg
kg−1), which determines the amount of cloud water
available for precipitation conversion, following the
study by Segele et al. (2009).
Simulations using the modified Emanuel convective scheme with l0 = 0.01 and a = 0.1 showed the best
model performance, reducing the mean bias in temperature over 25%, in cloudiness over 20% and in
precipitation over 70% when compared with the
respective observed values from the network of the
Hellenic N ational Meteorological Service (HN MS)
(Mystakidis et al. 2012). Following these results, the
high-resolution transient regional climate simulation
(1960−2100) was performed using the Emanuel
scheme with l0 = 0.01 and a = 0.1.
The transient high-resolution simulation was evaluated over the period 1975−2000 using temperature
and precipitation measurements from 79 stations of
the HN MS observational network. Selection of the
period 1975−2000 was based on optimum spatial coverage of stations over this period, meaning a compromise between the longest period and the highest density of stations. The data were subjected to a series of
quality control tests including internal consistency
checks and identification of outliers in the monthly
values. An algorithm to correct for missing values,
Zanis et al.: High-resolution climate simulation for Greece
127
The metrics used for model evaluation are (1) the
mean annual bias, providing an estimation of the
over- or underestimation of the selected meteorological variable; (2) the correlation coefficient R of the
observed and modeled time-series over the HN MS
stations to identify their temporal agreement; and (3)
the normalized standard deviation (N SD), which is
the ratio of the standard deviation of the modeled
monthly values to the standard deviation of the
observed values. This measure ideally equals unity
and becomes either >1 or <1 depending on whether
the model over- or underestimates the
E-obs 0.22° × 0.22°
amplitude of variability of the evaluated variable.
Furthermore, in our analysis, we
used monthly temperature and precipitation from (1) the forcing T63
general circulation model (GCM)
ECHAM5 coupled to MPI-OM (run 3)
with 1.875° resolution for the 20th century (Roeckner 2006) downloaded
from the Climate and Environmental
Retrieval and Archive (CERA) datab)
base (http://cera-www.dkrz.de/CERA/
index.html), (2) the RCM25 simulation
15°E
20°
25°
30°
over Europe and (3) the E-OBS gridRegCM 25 × 25 km
ded dataset based on observational
data (Haylock et al. 2008) with a spatial resolution of 0.22° on a rotated
grid (www.ecad.eu/).
based on current WMO guidelines, was also applied
using the criterion that a year is considered to be complete when at least 8 monthly values are recorded
(Marougianni et al. 2012). The evaluation analysis
was done on a monthly basis; hence, monthly mean
time-series of surface temperature and precipitation
were extracted from model fields to be compared with
the station observations. Because of differences between station altitude and model grid altitude, a correction was applied in modelled temperature, with a
standard temperature vertical lapse rate of 6°C km−1.
HNMS stations
42°N
44°N
42°
40°
40°
38°
38°
36°
36°
a)
34°
34°
20°E
25°
30°
RegCM 10 × 10 km
44°N
44°N
42°
42°
40°
40°
38°
38°
3. RESULTS
36°
3.1. Evaluation over the period
1975−2000
36°
c)
d)
34°
34°
15°E
20°
25°
30°
15°E
20°
25°
30°
ECHAM 1.9° × 1.9°
44°N
20
15
40°
10
38°
36°
5
e)
34°
15°E
20°
25°
30°
Temperature (°C)
42°
Fig. 2. Mean annual near-surface temperature (1975−2000)
for (a) Hellenic National Meteorological Service (HNMS) stations, (b) E-OBS, (c) RCM10,
(d) RCM25 and (e) the forcing
GCM ECHAM5
The average annual near-surface
temperatures over the period 1975−
2000 for the HNMS stations are shown
as points in Fig. 2a, while the respective gridded E-OBS field is illustrated in Fig. 2b. Visual comparison of
Fig. 2a,b indicates rather colder annual temperatures in E-OBS than at
the HNMS stations. The annual nearsurface temperature average fields for
RCM10, RCM25 and ECHAM are also
shown in Fig. 2 with their respective
grid resolution. Visually, we note a
rather good comparison of the RCM10
temperature field with RCM25 and EOBS, with a tendency for colder tem-
Clim Res 64: 123–140, 2015
128
pared to RCM25, with RCM10 depicting the finer
regional characteristics over the mountain regions.
Both RCM10 and RCM25 capture the precipitation
pattern shown in E-OBS, but there is a tendency for
higher precipitation over western Greece and the
Pindus mountain range in comparison to the E-OBS
and HN MS stations. Fig. 3 also indicates the improvement in depicting spatially the topographic
characteristics as we move from the global model
with about 190 km resolution towards RCM25 and
the even finer characteristics in RCM10.
Considering that the actual station
E-obs 0.22° × 0.22°
data provide a more realistic basis for
44°N
the evaluation of the RCM10 simulation, the statistical evaluation meas42°
ures were calculated with regard to
the HN MS station data. Fig. 4 and
40°
Table 1 provide a quantitative outlook
for the evaluation of RCM10 and
RCM25 with respect to HN MS stati38°
ons on an annual basis. Specifically,
Fig. 4 illustrates the mean annual sur36°
face temperature (top) and precipitab)
tion (bottom) bias of RCM10 and
34°
RCM25 for the time period 1975−2000.
15°E
20°
25°
30°
Furthermore, Table 1 shows summary
RegCM 25 × 25 km
statistics of the 3 evaluation metrics
44°N
(bias, R, N SD) for near-surface temperature and precipitation. The results
42°
for near-surface temperature indicate
negative biases for both RCM25 and
40°
RCM10. The median annual bias for
all HNMS stations is slightly reduced
in RCM10 (median bias −2.2°C) by
38°
about 0.3°C compared to RCM25
(median bias −2.5°C). The percentile
36°
analysis shows that the P75 (3rd quard)
tile) of the bias is also reduced in
34°
RCM10 but that in the lower part of the
15°E
20°
25°
30°
distribution, the P25 (1st quartile) indicates slightly larger negative biases for
RCM10 compared to RCM25. FurtherFig. 3. Annual precipitation
more, the maximum and minimum valsum (1975−2000) for (a) Hel2000
ues of the biases are larger in RCM10
lenic N ational Meteorological Service (HNMS) stations,
than in RCM25. A question that arises
1500
(b) E-OBS, (c) RCM10, (d)
is to what extent the differences in
RCM25 and (e) the forcing
the temperature biases of RCM10 and
GCM ECHAM5. The white
1000
RCM25 are related to the different reareas in (c) RCM10 and (d)
RCM25 denote areas with
solution or the different convective
500
annual precipitation beyond
schemes. To address this issue, we carthe selected colour scale
ried out a sensitivity study running
(>2400 mm). The white areas
0
an additional annual simulation of
in (b) E-OBS are areas with
lack of observational data
RCM10 for 1 yr (year 1995) but using
peratures and more structural topographic characteristics over the mountainous regions (Pindus mountain range) due to the higher resolution. Fig. 2e also
shows the limited regional information of the temperature field of the GCM in comparison to the E-OBS,
RCM10 and RCM25 fields.
Similarly, Fig. 3 shows the annual precipitation
sum over the period 1975−2000 for the HN MS stations, E-OBS gridded data, RCM10, RCM25 and the
forcing GCM ECHAM5. Visually, we note a good
comparison of the RCM10 precipitation field com-
HNMS stations
42°N
40°
38°
36°
a)
34°
20°E
25°
30°
RegCM 10 × 10 km
44°N
42°
40°
38°
36°
c)
34°
15°E
20°
25°
30°
ECHAM 1.9° × 1.9°
44°N
Precipitation (mm)
42°
40°
38°
36°
e)
34°
15°E
20°
25°
30°
Zanis et al.: High-resolution climate simulation for Greece
42°N
RCM10–HNMS
42°N
RCM25–HNMS
129
4
3
2
1
0
–1
–2
–3
–4
Bias (mm)
41°
Bias (°C)
c064 p123 _ supp. pdf). This re sult in di cates that the reduction of the median
40°
40°
annual bias in near-surface tempera39°
39°
ture by about 0.3°C for RCM10 com38°
38°
pared to RCM25 is rather linked to
37°
37°
36°
36°
the different convective schemes im35° a)
35° b)
plemented through a feedback mechanism, as has been also discussed by
19°
21°
23°
25°
27°
29°
19°
21°
23°
25°
27°
29°
Zanis et al. (2009b). Use of the Emanuel convective scheme imposes a
RCM10–HNMS
RCM25–HNMS
42°N
42°N
80
stronger convective activity than the
41°
41°
60
Grell-FC scheme, stronger vertical
40°
40°
40
redistribution of humidity, fewer low39°
39°
20
38°
38°
level clouds, more short-wave solar
0
37°
37°
–20
radiation absorbed from the ground
–40
36°
36°
and, hence, warmer low-level tem–60
35° c)
35° d)
peratures. However, the different res–80
olutions in RCM10 and RCM25 play a
19°
21°
23°
25°
27°
29°
19°
21°
23°
25°
27°
29°
role in the distribution of the biases.
Fig. 4. Mean annual bias in near-surface temperature and precipitation over
In the case of precipitation, there is
the period 1975−2000 for (a,c) RCM10 and (b,d) RCM25
a larger negative bias in RCM10 than
in RCM25, with the median bias beTable 1. Summary statistics of the major evaluation metrics on an annual baing −9.8 mm for RCM10 and −3.3 mm
sis (25th, 50th and 75th percentiles [P25, P50 and P75, respectively] and
for RCM25, but the distribution of
mean, minimum and maximum values) based on the comparison of RCM10
biases is wider in RCM10 than in
and RCM25 near-surface temperature and annual precipitation values with
the respective values from the 79 Hellenic National Meteorological Service
RCM25 (Table 1) based on the P25
stations. The statistical evaluation metrics include mean bias (°C for temperaand P75 values (1st and 3rd quartiles,
ture, mm for precipitation), temporal correlation coefficient and normalised
respectively). However, RCM25 prostandard deviation (NSD)
duces slightly more precipitation at
the HNMS stations over the Pindus
Bias
R
NSD
mountain range compared to RCM10.
RCM10 RCM25
RCM10 RCM25
RCM10 RCM25
The Pindus mountain range stretches
Near-surface temperature
from near the Greek−Albanian borP25
−3.23
−2.99
0.94
0.94
0.77
0.89
ders in northern Epirus, entering the
P50
−2.18
−2.46
0.95
0.95
0.84
0.95
Epirus region and Macedonia region
P75
−1.20
−1.70
0.95
0.95
0.91
1.02
in northern Greece, down to the north
Mean
−2.49
−2.42
0.95
0.94
0.84
0.97
of the Peloponnese. For 5 HNMS staMinimum
−11.61
−8.88
0.92
0.93
0.56
0.72
Maximum
2.87
1.43
0.96
0.95
1.15
1.33
tions located in the greater area of the
Pindus of northwestern Greece (IoanAnnual precipitation
nina, Kastoria, Konitsa, Kozani and
P25
−24.05 −15.39
0.29
0.24
0.68
0.63
Florina), the mean annual precipitaP50
−9.83
−3.26
0.35
0.33
0.92
0.80
P75
7.35
4.22
0.43
0.41
1.32
0.96
tion bias in RCM10 is −19.9 ± 82.1
Mean
−8.27
−2.83
0.34
0.32
0.99
0.91
mm, while in RCM25, it is 29 ± 86
Minimum
−79.29 −54.03
0.09
0.07
0.12
0.35
mm. A direct comparison of RCM10
Maximum
70.07
86.86
0.52
0.52
3.48
3.73
using Emanuel with RCM10G using
Grell-FC indicated a wet bias of
RCM10G over western Greece and
the Pindus mountain range from 50 to 300 mm annuthe same convective scheme (Grell-FC) as in RCM25
ally (see Fig. S1 in the Supplement). These biases are
(hereafter denoted RCM10G). A direct comparison of
higher than the difference of the annual precipitation
RCM10 using Emanuel with RCM10G using Grellbias between RCM10 and RCM25. This indicates
FC indicated a cold bias of RCM10G over the Greek
that the use of Grell-FC in RCM25 leads to a slightly
domain ranging from −0.2 to −0.5°C (see Fig. S1 in
higher precipitation rate and bias than RCM10 (using
the Supplement at www.int-res.com/articles/suppl/
41°
130
Clim Res 64: 123–140, 2015
the modified Emanuel scheme), but these wet biases
accentuate at higher resolution, as they become
larger with the use of the Grell-FC scheme in
RCM10G. A seasonal analysis of the biases in temperature and precipitation for the RCM10 simulation
is provided in Table S1 in the Supplement at www.
int-res.com/articles/suppl/c064p123_supp.pdf.
The temporal correlation of RCM10 with observations is high for temperature, ranging from 0.92 to
0.96 at all stations, but is considerably lower for precipitation, with values ranging from 0.09 to 0.52. Generally, climate models have difficulties successfully
representing the amount of precipitation and its spatial and temporal characteristics since it is controlled
by several factors including the convection scheme,
the energy and water budget and topography. Table 1
indicates that the higher spatial resolution in RCM10
compared to RCM25 has no impact on the temporal
correlations for temperature and precipitation. Furthermore, the spatial correlation is rather low for precipitation (ranging from 0.2 to 0.3) and higher for the
homogeneous field of temperature (around 0.8) for
both RCM10 and RCM25. This result indicates no
considerable improvement in spatial correlation, despite the higher spatial resolution in RCM10 compared
to RCM25, presumably because of the limited number
of stations over the mountainous regions.
The NSD shows a median value of 0.84 in temperature and 0.92 in precipitation for RCM10, while the
respective values for RCM25 are 0.95 and 0.8. This
indicates that the amplitude of variability is slightly
more underestimated in RCM10 than in RCM25 for
temperature but reverses for precipitation. The range
of the distribution in NSD values (difference between
P25 and P75 in Table 1) are greater in RCM10 than in
RCM25, which could be related to the different resolution.
As mentioned in Section 2, selection of the period
1975−2000 for our evaluation was based on optimum
spatial coverage of stations over this period. Nevertheless, we carried out a similar analysis with fewer
HNMS stations for the control period 1961−1990,
revealing similar results with those discussed for the
period 1975−2000.
3.2. Future projections in annual near-surface
temperature and precipitation
The simulated RCM10 projected changes on an
annual basis in near-surface temperature and precipitation between the 2 future time slices (2021−2050
and 2071−2100; hereafter FUT1 and FUT2, respec-
tively) and the control time slice (1961−1990) are
shown in Figs. 5 & 6. In the first half of the century,
small changes are shown for temperature and precipitation. Mean annual temperature increases mostly by up to 1.4°C over most of Greece and by up to
1.8°C over parts of Peloponnese and Crete as well as
islands of the Aegean Sea (Fig. 5a). The change in
maximum annual temperature is slightly higher, with
an increase ranging from 1.4 to 1.8°C over land for
almost the whole of Greece (Fig. 5c). The change in
minimum annual temperature is slightly lower, with
an increase ranging from 1.0 to 1.4°C over the whole
of Greece except the island of Rhodes, where the increase ranges from 1.4 to 1.8°C (Fig. 5e). All of these
changes in temperature are statistically significant at
the 95% confidence level. Concerning precipitation,
the near-future change is ±15%, being mostly negative in the southern part of the domain, indicating
drier conditions, but these changes are non-statistically significant at the 95% confidence level (Fig. 6a).
The largest part of the changes in total precipitation
are related to changes in non-convective precipitation (Fig. 6c,e).
At the end of the 21st century, larger changes in
both near-surface temperature and precipitation are
projected under the A1B scenario. Mean annual temperature increases by about 3.4 to 4.2°C over land
and by 2.6 to 3.4°C over the sea (Fig. 5b). The change
in maximum annual temperature is slightly higher,
reaching values of about 4.2 to 4.6°C over the mainland (Fig. 5d), while the change in minimum annual
temperature is slightly lower, with an increase up to
3.8°C over land (Fig. 5f). Precipitation is projected to
decrease between 10 and 40%, with the stronger decrease shown over southern Greece (Fig. 6b). Here,
we also note that the largest part of the changes in
total precipitation are related to non-convective precipitation (Fig. 6d,f), namely decreasing significantly
(at the 95% confidence level) all over Greece with
stronger decreasing rates towards the southern parts
(Fig. 6f).
Notably, RCM25 and RCM10 projected changes in
near-surface temperature and total and convective
precipitation (not shown) are comparable in magnitude and spatial distribution. This indicates the robustness of the climate change signal in the future
projections, despite the different resolution and the
different convective schemes implemented. Furthermore, this robustness, especially for the end of the
21st century, implies that the physical signal of climate change in both simulations is presumably
greater than the important issue of the model’s internal variability resulting from chaotic processes intrin-
Zanis et al.: High-resolution climate simulation for Greece
Mean
2021–2050 – 1961–1990
44°N
42°
40°
40°
38°
38°
36°
36°
34°
a)
16°E
34°
18°
20°
22°
24°
26°
28°
2071–2100 – 1961–1990
44°N
42°
30°
131
b)
16°E
18°
20°
22°
24°
26°
28°
30°
18°
20°
22°
24°
26°
28°
30°
18°
20°
22°
24°
26°
28°
30°
Max
44°N
44°N
42°
42°
40°
40°
38°
38°
36°
36°
c)
d)
34°
34°
16°E
18°
20°
22°
24°
26°
28°
16°E
30°
Min
44°N
44°N
42°
42°
40°
40°
38°
38°
36°
36°
e)
f)
34°
34°
16°E
1
18°
20°
22°
1.4 1.8 2.2 2.6
24°
3
26°
28°
3.4 3.8 4.2 4.6
30°
16°E
5
Temperature (°C)
Fig. 5. Change between 2021−2050 and 1961−1990 as well as between 2071−2100 and 1961−1990 for (a,b) annual mean, (c,d)
maximum near-surface temperature and (e,f) minimum near-surface temperature based on the high-resolution simulation
RCM10. All differences are statistically significant at the 95% confidence level according to a 2-tailed paired t-test
sic to the atmosphere (Giorgi & Bi 2000, Crétat & Pohl
2012). The issue of model internal error is more of a
concern for the early decades of the 21st century,
when the physical signal of the anthropogenic climate change is low. However, this issue can only be
addressed quantitatively by generating ensembles of
simulations since, owing to its random nature, the
variability should tend towards zero as the number of
ensemble members increases (O’Brien et al. 2011).
3.3. Future projections in seasonal near-surface
temperature and precipitation
The RCM10 projected changes on a seasonal basis
in mean near-surface temperature and precipitation
between 2071−2100 and the control period 1961−
1990 are illustrated in Figs. 7 & 8. In winter, we note
the lowest changes in mean near-surface temperature, with values ranging between 3.0 and 3.4°C over
Clim Res 64: 123–140, 2015
132
Total
2021–2050 – 1961–1990
44°N
42°
42°
40°
40°
38°
38°
36°
36°
34°
a)
16°E
34°
18°
20°
22°
24°
26°
28°
2071–2100 – 1961–1990
44°N
b)
30°
16°E
18°
20°
22°
24°
26°
28°
30°
18°
20°
22°
24°
26°
28°
30°
18°
20°
22°
24°
26°
28°
30°
Convective
44°N
44°N
42°
42°
40°
40°
38°
38°
36°
36°
c)
d)
34°
34°
16°E
18°
20°
22°
24°
26°
28°
16°E
30°
Non-convective
44°N
44°N
42°
42°
40°
40°
38°
38°
36°
36°
e)
f)
34°
34°
16°E
18°
20°
22°
–50 –40 –30 –20 –10
24°
0
26°
28°
30°
16°E
10 20 30 40 50
Precipitation (%)
Fig. 6. Percentage change between 2021−2050 and 1961−1990 as well as between 2071−2100 and 1961−1990 for (a,b) annual total precipitation, (c,d) convective precipitation and (e,f) non-convective precipitation based on the high-resolution simulation
RCM10. The line shading indicates areas in which the differences between future runs and the control run are not statistically
significant at the 95% confidence level according to a 2-tailed paired t-test
land and the northern Aegean Sea (Fig. 7a). In
spring, the changes over the largest part of the Greek
mainland range between 3.4 and 3.8°C (Fig. 7b). In
autumn, the changes are slightly higher, especially
over northern Greece, the Peloponnese and Crete,
with values ranging between 3.8 and 4.2°C (Fig. 7d).
The largest changes in mean near-surface temperature are seen during summer over the Greek main-
land, with values between 4.2 and 4.6°C for southern
Greece (e.g. the Peloponnese, Crete and east central
Greece), between 4.6 and 5°C for northeastern
Greece and more than 5°C for northwestern Greece.
Changes in maximum near-surface temperature in
summer for FUT2, with values higher than 5°C, cover
a larger part of northern Greece (not shown) in comparison to the mean summer temperature. For FUT1,
Zanis et al.: High-resolution climate simulation for Greece
Win (DJF)
44°N
42°
40°
40°
38°
38°
36°
36°
34°
a)
16°E
34°
18°
20°
22°
24°
Spr (MAM)
44°N
42°
26°
28°
30°
133
b)
16°E
18°
20°
Sum (JJA)
22°
24°
26°
28°
30°
26°
28°
30°
Aut (SON)
44°N
44°N
42°
42°
40°
40°
38°
38°
36°
36°
c)
d)
34°
34°
16°E
1
18°
20°
22°
1.4 1.8 2.2 2.6
24°
3
26°
28°
3.4 3.8 4.2 4.6
Temperature (°C)
30°
16°E
18°
20°
22°
24°
5
Fig. 7. Seasonal mean near-surface temperature change between 2071−2100 and 1961−1990 for (a) winter, (b) spring, (c) summer and (d) autumn based on the high-resolution RCM10. All differences are statistically significant at the 95% confidence level
according to a 2-tailed paired t-test
the summer period shows the largest changes
throughout the year in mean near-surface temperature, with values ranging between 1.8 and 2.2°C over
the land, while the winter period shows the lowest,
with values ranging between 0.8 and 1.4°C (not
shown).
Concerning precipitation for the FUT2 period,
summer presents the lowest changes in the mean
daily precipitation rate because of the very low
amounts of precipitation during this season, with
these changes being non-statistically significant at
the 95% confidence level (Fig. 8d). Winter and spring
show the largest decreases in future precipitation
daily rate, with values ranging from –0.15 mm d−1 to
more than –0.75 mm d−1 over the Pindus mountain
range. Winter shows the largest decreases in precipitation rate in western and southern Greece, while
spring shows the largest decrease in northern
Greece (Fig. 8a,b). The percentage changes in total
precipitation reach a decrease up to roughly 40 to
45% in certain regions during winter and spring (see
Fig. S2 in the Supplement). Autumn also shows decreasing daily precipitation rates for the whole of
Greece (Fig. 8d), but the changes are smaller compared to spring and winter. The largest part of the
changes in total precipitation are related to changes
in non-convective precipitation for all seasons. The
changes in convective precipitation for the FUT2
period are provided in Fig. S3 in the Supplement.
In the FUT1 period, the seasonal changes in precipitation daily rate are either slightly positive or
negative but generally non-statistically significant
at the 95% confidence level (not shown). For example, in winter, the daily precipitation rate decreases
in southern Greece by about −0.15 to −0.45 mm d−1
but increases over northern Greece by 0.15 to 0.30
mm d−1. In spring, the precipitation daily rate
slightly decreases by about −0.15 mm d−1 for the
whole of Greece, while in autumn, it increases by a
similar amount for the whole of Greece.
Clim Res 64: 123–140, 2015
134
Win (DJF)
44°N
42°
42°
40°
40°
38°
38°
36°
36°
34°
a)
16°E
34°
18°
20°
22°
24°
Spr (MAM)
44°N
26°
28°
30°
b)
16°E
18°
20°
Sum (JJA)
22°
24°
26°
28°
30°
26°
28°
30°
Aut (SON)
44°N
44°N
42°
42°
40°
40°
38°
38°
36°
36°
c)
d)
34°
34°
16°E
18°
20°
22°
24°
–1.5 –1.2 –0.9 –0.6 –0.3
0
26°
0.3
28°
30°
0.6
0.9
16°E
18°
20°
22°
24°
Precipitation (mm d –1)
Fig. 8. Seasonal mean precipitation change between 2071−2100 and 1961−1990 for (a) winter, (b) spring, (c) summer and (d) autumn based on the high-resolution RCM10. The line shading indicates areas in which the differences between future runs and
the control run are not statistically significant at the 95% confidence level according to a 2-tailed paired t-test
3.4. Future projections in climate indices
RCM10 simulations were used to calculate future
changes in the FUT1 and FUT2 periods relative to the
control period in climate indices including (1) the
number of hot days with Tmax > 35°C; (2) the number of warm nights with Tmin > 20°C; (3) the number
of night frost days with Tmin < 0°C; (4) the number of
continuous dry spell days with daily precipitation < 1
mm; and (5) the length of the growing season, which
corresponds to the number of days between the last
spring frost and the first autumn frost. These indicative climate indices were selected following the
assessment of regional climate change for Greece by
the Climate Change Impacts Study Committee
(Zerefos et al. 2011).
For the FUT1 period, an increase in annual hot
days is projected, with Tmax > 35°C on the order of
14 to 21 d for most of Greece except in Peloponnese,
where the increase is slightly higher, ranging between 21 and 28 d (Fig. 9a). For the FUT2 period, we
note larger increases in the number of hot days, ranging between 49 and 56 d for most of Greece, with the
largest increases in the range of 56 to 63 d in southwestern Peloponnese (Fig. 9b). The number of warm
nights with Tmin > 20°C is projected to increase by
roughly 10 to 30 d in the FUT1 period (Fig. 9c) and by
about 50 to 80 d in the FUT2 period (Fig. 9d), with the
largest increases seen over southern Greece (Peloponnese and Crete).
Concerning the number of night frost days, RCM10
simulations in the FUT1 period project a decrease in
the range of 4 to 8 d for southern Greece and a
slightly higher decrease in northern Greece, between 8 and 12 d (Fig. 9e). The latitudinal distinction
in the decrease of night frost days becomes stronger
in the FUT2 period, when decreases are projected in
the range of 20 to 28 d for northern Greece, 12 to 20 d
Zanis et al.: High-resolution climate simulation for Greece
40 d in the FUT2 period, with the largest increases in
northern Greece (Fig. 10c,d). For the island of Crete,
no change in night frost days or growing season is
projected for both the FUT1 and FUT2 periods, because of the generally higher temperatures throughout the year. All of the above-mentioned changes in
hot days, warm nights, night frost days and length of
the growing season for both the FUT1 and FUT2
periods are consistent with the generally warmer climate projected for the A1B IPCC scenario, as discussed in Section 3.3.
in central Greece and 4 to 16 d in southern Greece
(Fig. 9f). The larger decrease in night frost days over
northern Greece is linked to the generally higher
temperatures and the lower possibility for the development of night frosts. The changes in the number of
hot days, warm nights and night frosts are statistically significant at the 95% confidence level according to a 2-tailed paired t-test for all grid points for
both the FUT1 and FUT2 periods.
The growing season is also projected to extend by
about 4 to 16 d in the FUT1 period and by about 20 to
2021–2050 – 1961–1990
44°N
135
2071–2100 – 1961–1990
44°N
42°
42°
40°
40°
42
38°
38°
28
36°
36°
56
14
34°
a)
16°E
34°
18°
20°
22°
24°
26°
28°
30°
44°N
0
b)
16°E
No. of hot days
70
18°
20°
22°
24°
26°
28°
30°
44°N
100
42°
80
40°
40°
38°
38°
36°
36°
c)
60
40
20
d)
34°
34°
16°E
No. warm days
42°
18°
20°
22°
24°
26°
28°
30°
16°E
44°N
44°N
42°
42°
18°
20°
22°
24°
26°
28°
30°
0
–8
40°
40°
–16
38°
38°
36°
36°
e)
–24
–32
f)
34°
34°
16°E
No. night frost days
0
18°
20°
22°
24°
26°
28°
30°
–40
16°E
18°
20°
22°
24°
26°
28°
30°
Fig. 9. Annual change between 2021−2050 and 1961−1990 as well as between 2071−2100 and 1961−1990 for (a,b) the number of hot days with Tmax > 35°C, (c,d) the number of warm with days with Tmin > 20°C and (e,f) the number of night frost
days with Tmin < 0°C based on the high-resolution simulation RCM10. All differences are statistically significant at the 95%
confidence level according to a 2-tailed paired t-test
Clim Res 64: 123–140, 2015
136
2021–2050 – 1961–1990
44°N
2071–2100 – 1961–1990
44°N
50
40°
40°
38°
38°
36°
36°
30
20
10
0
–10
–20
34°
a)
16°E
34°
18°
20°
22°
24°
26°
28°
30°
–30
16°E
44°N
44°N
42°
42°
–40
b)
–50
18°
20°
22°
24°
26°
28°
30°
40
32
40°
40°
24
38°
38°
36°
36°
c)
16
8
d)
34°
34°
16°E
18°
20°
22°
24°
26°
28°
30°
16°E
18°
20°
22°
24°
26°
28°
30°
No. growing season days
42°
No. dry spell days
40
42°
0
Fig. 10. Similar to Fig. 9 but for (a,b) dry spell days and (c,d) growing season days. The line shading indicates areas in which
the differences between future runs and the control run are not statistically significant at the 95% confidence level according
to a 2-tailed paired t-test
Finally, concerning the number of continuous dry
spell days with daily precipitation < 1 mm, there is a
tendency for an increase of about 5 to 20 d for the
FUT1 period in most parts of northwestern, central
and southern Greece, but there are areas with no
change in northern Greece (e.g. central and eastern
Macedonia) or with even a small decrease such as in
northeastern Greece (Thrace) and Crete (Fig. 10a).
This is consistent with the pattern of precipitation
changes in the FUT1 period (Fig. 6a). In the FUT2
period, a general increase in the number of continuous dry spell days in the range of 10 to 40 d is projected for the majority of Greece due to the anticipated overall reduction in precipitation (Fig. 10b).
Furthermore, increases in the number of continuous
dry spell days are higher over southern Greece than
over northern Greece in accordance with the pattern
of precipitation changes in the FUT2 period (Fig. 6b).
4. DISCUSSION AND CONCLUSIONS
A high-resolution simulation (RCM10) was carried
out with RegCM3 over the period 1960−2100 under
the A1B scenario, which is a transient regional pro-
jection for future climate change, for the first time to
our knowledge with 10 km spatial resolution for
Greece. Based on a previous study for optimisation of
the model configuration concerning the convection
scheme (Mystakidis et al. 2012), the RCM10 simulation was performed using the Emanuel scheme with
modified parameters in the rate at which the cloudbase upward mass flux is relaxed to steady state, and
in the warm cloud autoconversion threshold.
The RCM10 simulation was evaluated over the
period 1975−2000 using temperature and precipitation measurements from 79 stations of the HN MS
observational network. Selection of the period 1975−
2000 was based on optimum spatial coverage of stations over this period. RCM10 precipitation and temperature fields depict the finer regional characteristics over the complex Greek terrain compared to
RCM25. However, the station-based evaluation does
not reveal a considerable improvement in RCM10
compared to RCM25, with only a slight reduction in
the annual bias in near-surface temperature and precipitation over mountain regions, presumably related
to the different convective schemes implemented
through a feedback mechanism, as has also been discussed by Zanis et al. (2009b) and Mystakidis et al.
Zanis et al.: High-resolution climate simulation for Greece
(2012). The higher spatial resolution in RCM10 versus RCM25 does not lead to improvements of the
temporal and spatial correlations of temperature and
precipitation, presumably because of the limited
number of stations over the mountainous regions,
and the complex terrain. The projected changes in
mean annual near-surface temperature for the nearfuture period (2021−2050) are in the range of up to
1.4°C over central and northern Greece to up to 1.8°C
over parts of Peloponnese and Crete as well as
islands of the Aegean Sea. Concerning precipitation,
the near-future changes are statistically non-significant at the 95% confidence level. At the end of the
21st century, larger changes in both near-surface
temperature and precipitation are projected, with
mean annual temperature increasing by about 3.4 to
4.2°C over land and by 2.6 to 3.4°C over the sea and
precipitation decreasing by 10 to 40%, with a positive gradient from the north to the south. As has been
discussed in Tolika et al. (2012), the different sea−
land warming, with a maximum in summer, could be
attributed to the differences in heat capacity and
evaporation of the sea water. These RCM10 simulated changes in annual temperature and precipitation are in accordance with a regional climate
change ensemble from the EN SEMBLES project
under the A1B scenario indicating a temperature
increase over the inland area of Greece of about 3.5
to 4.0°C and precipitation decrease of about 15 to
25% (Jacob et al. 2014). Furthermore, the recent
high-resolution climate change ensemble from the
EURO-CORDEX initiative indicated a temperature
increase over the inland area of Greece of about 2 to
2.5°C and a small precipitation decrease up to15% in
southern Greece for the RCP4.5 scenario, while
RCP8.5 indicated a temperature increase of about 4
to 5°C and a precipitation decrease of 5 to 35%
(Jacob et al. 2014).
The future projected climate change signal in the
RCM10 simulation is also in agreement with previous
model simulations based on RCMs indicating a consistent warming and drying of the Mediterranean
region mainly over the last decades of the 21st century for various emission scenarios (e.g. Gibelin &
Déqué 2003, Gao et al. 2006, Diffenbaugh et al. 2007,
Goubanova & Li 2007, Giorgi & Lionello 2008, Önol &
Semazzi 2009, Krichak et al. 2011, Lelieveld et al.
2013). The simulated RCM25 projected changes in
near-surface temperature and precipitation are similar to RCM10 in magnitude and spatial distribution,
indicating the robustness of the RegCM future projections, despite the different resolution and the different convective schemes implemented.
137
Future changes in maximum annual temperature
are slightly higher, and changes in minimum annual
temperature are slightly lower, when compared to
the changes in mean annual temperature. Similar
results showing higher sensitivity of Tmax than Tmin
in future climate projections for the Mediterranean
have also been reported in previous studies based on
both GCMs and RCMs (Sánchez et al. 2004, Lobell et
al. 2007, Elguindi et al. 2013). This could be linked to
drying and its impacts on Tmax during the warm
period of the year through decreases in cloud cover
and the surface latent heat flux (Dai et al. 2004, van
der Schrier et al. 2006), pointing also to the role of soil
moisture depletion in the amplification of climate
warming in the EM (Zittis et al. 2014).
The largest seasonal changes in mean near-surface
temperature are seen during summer over the Greek
mainland, with values in the late 21st century from
3.8 to 4.6°C for southern Greece, 4.6 to 5°C for northeastern Greece and > 5°C for northwestern Greece.
These simulated changes in summer mean near-surface temperature are in agreement with the study of
Tolika et al. (2012) that reported changes in the
range of 4 to 5.5°C for Greece from an ensemble of 8
RCMs under the A1B scenario. The changes in maximum near-surface temperature in summer for FUT2
are slightly higher, with values > 5°C covering a large
part of northern Greece, possibly related to the
winter and spring drying over the same regions and
the soil moisture−temperature feedback discussed
by Zittis et al. (2014). These RCM10 projected summertime changes in Tmax are in the range of simulated Tmax changes with the PRECIS model under
A1B for the EM (Lelieveld et al. 2013). Lelieveld et al.
(2013) reported that across the Balkan Peninsula and
Turkey, climate change is particularly rapid, and
temperatures, especially in summer, are expected to
increase strongly. Concerning seasonal precipitation
changes in the late 21st century, winter shows the
largest decreases in precipitation rate in western and
southern Greece, while spring shows the largest decreases in northern Greece. The simulated decrease
in precipitation in winter over the Mediterranean (especially at the end of the 21st century) is linked to an
increase in the sea level pressure (SLP) gradient between the Azores High and the Icelandic Low that results in an intensified zonal circulation over Europe
and a northward shift of the location of the storm
tracks, which in turn leads to more precipitation in
central Europe and less precipitation in southern Europe. This is a common finding in many GCM and
RCM simulations for the area of interest (Giorgi & Lionello 2008, Tolika et al. 2012). The summer drying
Clim Res 64: 123–140, 2015
138
over the area of interest is probably associated with
the SLP increase over the British Isles and western
Europe, leading to weakening of the westerly flow,
while a less significant soil moisture−precipitation
feedback due to soil moisture depletion is expected
to further amplify the dryer and warmer conditions
over the domain of the study (Seneviratne et al. 2006,
Zanis et al. 2009b, Zittis et al. 2014). Also, the model’s
SLP decrease over the Mediterranean is probably a
result of the intense temperature increase and is consistent with a weakening and poleward expansion of
the Hadley circulation in the climate change simulations of the IPCC AR4 project (Lu et al. 2007).
The number of hot days, warm nights and night
frost days and the length of the growing season are
projected to increase in the near- and late-future
periods in accordance with the generally warmer
climate projected from the RCM10 simulation for
Greece. Furthermore, there is a tendency for a slight
increase in the number of continuous dry spell days
in the near-future period in most parts of northwestern, central and southern Greece (but not for north
central and northeastern Greece), while in the late
21st century, a general increase is projected for the
whole of Greece in accordance with the projected
overall reduction in precipitation. The reported
future climatic changes based on this high-resolution
RCM simulation could be a valuable dataset for other
researchers for the study of impacts in vital sectors
such as water resources, agriculture, tourism, forest
fire risk and energy demand for the region of Greece.
Acknowledgements. The research was co-financed by the
European Union (European Regional Development Fund)
and Greek national funds through the Operational Program
‘Competitiveness and Entrepreneurship’ of the N ational
Strategic Reference Framework (N SRF)−Research Funding
Program COOPERATION 2009 (no. 09SYN-31-1094, ‘Development of a Geographic Climate Information System’). The
RCM25 simulations over Europe, used as lateral boundary
conditions for our RCM10 simulations, were provided by the
Earth System Physics section of the International Centre for
Theoretical Physics (ICTP), Trieste, Italy, and we acknowledge support by Filippo Giorgi and Erika Coppola from
ICTP. The high-resolution regional climate simulations were
performed in the EGI/HellasGrid infrastructure. We acknowledge the E-OBS dataset from the EU-FP6 project ENSEMBLES (http://ensembles-eu.metoffice.com), data providers in
the ECA&D project (www.ecad.eu), and the forcing GCM
ECHAM5 r3 download from the CERA database (http://cerawww.dkrz.de/CERA/index.html). Finally, we thank the reviewers for their constructive comments.
LITERATURE CITED
➤ Christensen JH, Christensen OB (2007) A summary of the
PRUDEN CE model projections of changes in European
climate by the end of this century. Clim Change 81:7−30
➤ Crétat J, Pohl B (2012) How physical parameterizations can
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
modulate internal variability in a regional climate model.
J Atmos Sci 69:714−724
Dai A, Trenberth K, Qian T (2004) A global dataset of Palmer
drought severity index for 1870−2002: relationship with
soil moisture and effects of surface warming. J Hydrometeorol 5:1117−1130
Déqué M, Rowell DP, Lüthi D, Giorgi F and others (2007) An
intercomparison of regional climate simulations for
Europe: assessing uncertainties in model projections.
Clim Change 81:53−70
Dickinson R, Errico R, Giorgi F, Bates G (1989) A regional
climate model for the western United States. Clim
Change 15:383−422
Dickinson R, Henderson-Sellers A, Kennedy PJ (1993) Biosphere−Atmosphere Transfer Scheme, BATS: version 1E
as coupled to the NCAR community climate model. Tech
N ote N CAR/TN -387+ STR, N ational Center for Atmospheric Research, Boulder, CO
Diffenbaugh NS, Pal JS, Giorgi F, Gao X (2007) Heat stress
intensification in the Mediterranean climate change
hotspot. Geophys Res Lett 34:L11706, doi:10.1029/2007
GL030000
Elguindi N , Rauscher SA, Giorgi F (2013) Historical and
future changes in maximum and minimum temperature
records over Europe. Clim Change 117:415−431
Emanuel KA, Živkovi -Rothman M (1999) Development and
evaluation of a convection scheme for use in climate
models. J Atmos Sci 56:1766−1782
Fritsch JM, Chappell CF (1980) Numerical prediction of convectively driven mesoscale pressure systems. I. Convective parameterization. J Atmos Sci 37:1722−1733
Gao X, Pal JS, Giorgi F (2006) Projected changes in mean
and extreme precipitation over the Mediterranean region
from a high resolution double nested RCM simulation.
Geophys Res Lett 33:L03706, doi:10.1029/2005GL024954
Gibelin AL, Déqué M (2003) Anthropogenic climate change
over the Mediterranean region simulated by a global
variable resolution model. Clim Dyn 20:327−339
Giorgi F (2006) Climate change hot-spots. Geophys Res Lett
33:L08707, doi:10.1029/2006GL025734
Giorgi F, Pal JS, Bi XQ, Sloan LC, Elguindi N , Solmon F
(2006) Introduction to the TAC special issue: the RegCNET network. Theor Appl Climatol 86:1−4
Giorgi F, Bi X (2000) A study of internal variability of a
regional climate model. J Geophys Res 105:29503−29521
Giorgi F, Lionello P (2008) Climate change projections for
the Mediterranean region. Global Planet Change 63:
90−104
Giorgi F, Mearns LO (1999) Introduction to special section:
regional climate modeling revisited. J Geophys Res 104:
6335−6352
Giorgi F, Marinucci MR, Visconti G (1990) Use of a limitedarea model nested in a general circulation model for
regional climate simulation over Europe. J Geophys Res
Atmos 95:18413−18431
Giorgi F, Bi X, Pal J (2004) Mean, interannual variability and
trends in a regional climate change experiment over
Europe. II. Climate change scenarios (2071−2100). Clim
Dyn 23:839−858
Goubanova K, Li L (2007) Extremes in temperature and precipitation around the Mediterranean basin in an ensemble of future climate scenario simulations. Global Planet
Change 57:27−42
Zanis et al.: High-resolution climate simulation for Greece
➤ Grell GA (1993) Prognostic evaluation of assumptions used
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
by cumulus parameterizations. Mon Weather Rev 121:
764−787
Grell GA, Dudhia J, Stauer DR (1994) A description of the
fifth-generation Penn State/N CAR mesoscale model
(mm5). Tech Note NCAR/TN-398+ STR, National Center
for Atmospheric Research, Boulder, CO
Hadjinicolaou P, Giannakopoulos C, Zerefos C, Lange MA,
Pashiardis S, Lelieveld J (2011) Mid-21st century climate
and weather extremes in Cyprus as projected by six
regional climate models. Reg Environ Change 11:
441−457
Haylock MR, Hofstra N , Klein Tank AMG, Klok EJ, Jones
PD, N ew M (2008) A European daily high-resolution
gridded data set of surface temperature and precipitation
for 1950−2006. J Geophys Res 113:D20119, doi:10.1029/
2008JD010201
Held I, Zhao M (2011) The response of tropical cyclone statistics to an increase in CO2 with fixed sea surface temperatures. J Clim 24:5353−5364
Holtslag AAM, de Bruijn EIF, Pan HL (1990) A high resolution air mass transformation model for short-range
weather forecasting. Mon Weather Rev 118:1561−1575
Im ES, Coppola E, Giorgi F, Bi X (2010) Validation of a high
resolution regional climate model for the Alpine region
and effects of a subgrid-scale topography and land use
representation. J Clim 23:1854−1873
IPCC (2007) Climate change 2007: the physical science
basis. In: Solomon S, Qin D, Manning M, Chen Z and
others (eds) Contribution of Working Group I to the
Fourth Assessment Report of the Intergovernmental
Panel on Climate Change, Cambridge University Press,
Cambridge
Jacob D, Bärring L, Christensen OB, Christensen JH and
others (2007) An inter-comparison of regional climate
models for Europe: model performance in present-day
climate. Clim Change 81:31−52
Jacob D, Petersen J, Eggert B, Alias A and others (2014)
EURO-CORDEX: new high-resolution climate change
projections for European impact research. Reg Environ
Change 14:563−578
Kiehl JT, Hack JJ, Bonan GB, Boville BA, Breigleb BP,
Williamson D, Rasch P (1996) Description of the N CAR
community climate model (ccm3). Tech Note NCAR/TN420+ STR, N ational Center for Atmospheric Research,
Boulder, CO
Kostopoulou E, Giannakopoulos C, Anagnostopoulou C,
Tolika K, Maheras P, Vafiadis M, Founda D (2007) Simulating maximum and minimum temperature over
Greece: a comparison of three downscaling techniques.
Theor Appl Climatol 90:65−82
Krichak SO, Breitgand JS, Samuels R, Alpert P (2011) A double-resolution transient RCM climate change simulation
experiment for near-coastal eastern zone of the eastern
Mediterranean region. Theor Appl Climatol 103:167−195
Laprise R, de Elía R, Caya D, Biner S and others (2008)
Challenging some tenets of regional climate modelling.
Meteorol Atmos Phys 100:3−22
Lelieveld J, Hadjinicolaou P, Kostopoulou E, Giannakopoulos C, Pozzer A, Tanarhte M, Tyrlis E (2013) Model projected heat extremes and air pollution in the eastern
Mediterranean and Middle East in the twenty-first century. Reg Environ Change 14:1937–1949
Lobell DB, Bonfils C, Duffy PB (2007) Climate change uncertainty for daily minimum and maximum temperatures: a
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
139
model inter-comparison. Geophys Res Lett 34:L05715,
doi:10.1029/2006GL028726
Lu J, Vecchi GA, Reichler T (2007) Expansion of the Hadley
cell under global warming. Geophys Res Lett 34:L06805,
doi:10.1029/2006GL028443
Maheras P, Flocas HA, Patrikas I, Anagnostopoulou C (2001)
A 40 year objective climatology of surface cyclones in the
Mediterranean region: spatial and temporal distribution.
Int J Climatol 21:109−130
Maheras P, Anagnostopoulou C (2003) Circulation types and
their influence on the interannual variability and precipitation changes in Greece. In: Bolle HJ (ed) Mediterranean
climate: variability and trends. Springer, Berlin, p 215−239
Marougianni G, Melas D, Kioutsioukis I, Feidas H, Zanis P,
Anadranistakis E (2012) Trend analysis for climatic time
series for Greece. In: Helmis CG, N astos PT (eds)
Advances in meteorology, climatology and atmospheric
physics. Springer, Berlin, p 583−589
Meinshausen M, Smith SJ, Calvin K, Daniel JS and others
(2011) The RCP greenhouse gas concentrations and their
extensions from 1765 to 2300. Clim Change 109:213−241
Mystakidis S, Zanis P, Dogras C, Katragkou E and others
(2012) Optimization of a regional climate model for high
resolution simulations over Greece. In: Helmis CG, Nastos PT (eds) Advances in meteorology, climatology and
atmospheric physics. Springer, Berlin, p 623−629
Nastos PT, Politi N, Kapsomenakis J (2013) Spatial and temporal variability of the aridity index in Greece. Atmos Res
119:140−152
O’Brien TA, Sloan LC, Snyder MA (2011) Can ensembles of
regional climate model simulations improve results from
sensitivity studies? Clim Dyn 37:1111−1118
Önol B (2012) Effects of coastal topography on climate: highresolution simulation with a regional climate model.
Clim Res 52:159−174
Önol B, Semazzi FHM (2009) Regionalization of climate
change simulations over the eastern Mediterranean.
J Clim 22:1944−1961
Pal JS, Small EE, Eltahir EAB (2000) Simulation of regionalscale water and energy budgets: representation of subgrid cloud and precipitation processes within RegCM.
J Geophys Res 105:29579−29594
Pal JS, Giorgi F, Bi X, Elguindi N and others (2007) Regional
climate modeling for the developing world: the ICTP
RegCM3 and RegCN ET. Bull Am Meteorol Soc 88:
1395−1409
Rauscher S, Coppola E, Piani C, Giorgi F (2010) Resolution
effects on regional climate model simulations of seasonal
precipitation over Europe. Clim Dyn 35:685−711
Rodwell MJ, Hoskins BJ (1996) Monsoons and the dynamics
of deserts. QJR Meteorol Soc 122:1385−1404
Roeckner E (2006) ENSEMBLES ECHAM5-MPI-OM 20C3M
run1, monthly mean values. CERA database. World Data
Center for Climate, Hamburg. Available at: http://cerawww.dkrz.de/WDCC/ui/Compact.jsp?acronym=ENSEM
BLES_MPEH5_20C3M_1_MM
Roeckner E (2008) EN SEMBLES ECHAM5-MPI-OM
SRESA1B run3, monthly mean values. CERA database.
World Data Center for Climate, Hamburg. Available at:
http://cera-www.dkrz.de/WDCC/ui/Compact.jsp?
acronym=ENSEMBLES_MPEH5_SRA1B_3_MM
Rogelj J, Meinshausen M, Knutti R (2012) Global warming
under old and new scenarios using IPCC climate sensitivity range estimates. Nat Clim Change 2:248−253
Sánchez E, Gallardo C, Gaertner MA, Arribas A, Castro M
140
➤
➤
➤
➤
➤
➤
Clim Res 64: 123–140, 2015
(2004) Future climate extreme events in the Mediterranean simulated by a regional climate model: a first
approach. Global Planet Change 44:163−180
Sanchez-Gomez E, Somot S, Déqué M (2009) Ability of
an ensemble of regional climate models to reproduce
weather regimes over Europe-Atlantic during the period
1961−2000. Clim Dyn 33:723−736
Segele ZT, Leslie LM, Lamb PJ (2009) Evaluation and adaptation of a regional climate model for the Horn of Africa:
rainfall climatology and interannual variability. Int J Climatol 29:47−65
Seneviratne SI, Lüthi D, Litschi M, Schär C (2006) Landatmosphere coupling and climate change in Europe.
Nature 443:205−209
Tolika K, Anagnostopoulou C, Maheras P, Vafiadis M (2008)
Simulation of future changes in extreme rainfall and
temperature conditions over the Greek area: a comparison of two statistical downscaling approaches. Global
Planet Change 63:132−151
Tolika K, Zanis P, Anagnostopoulou C (2012) Regional climate change scenarios for Greece: future temperature
and precipitation projections from ensembles of RCMs.
Global NEST J 14:407−421
Torma C, Coppola E, Giorgi F, Bartholy J, Pongrácz R (2011)
Validation of a high-resolution version of the regional climate model RegCM3 over the Carpathian Basin. J
Hydrometeorol 12:84−100
Tselioudis G, Douvis C, Zerefos C (2012) Does dynamical
downscaling introduce novel information in climate
model simulations of precipitation change over a complex topography region? Int J Climatol 32:1572−1578
Tyrlis E, Lelieveld J, Steil B (2013) The summer circulation
over the eastern Mediterranean and the Middle East:
Editorial responsibility: Filippo Giorgi,
Trieste, Italy
➤
➤
➤
➤
➤
➤
influence of the South Asian monsoon. Clim Dyn 40:
1103–1123
van der Schrier G, Briffa KR, Jones PD, Osborn TJ (2006)
Summer moisture variability across Europe. J Clim 19:
2818−2834
Xoplaki E, González-Rouco JF, Luterbacher J, Wanner H
(2004) Wet season Mediterranean precipitation variability: influence of large-scale dynamics and trends. Clim
Dyn 23:63−78
Xue Y, Janjic Z, Dudhia J, Vasic R, De Sales F (2014) A
review on regional dynamical downscaling in intraseasonal to seasonal simulation/prediction and major factors that affect downscaling ability. Atmos Res 147−148:
68−85
Zanis P, Douvis C, Kapsomenakis I, Kioutsioukis I, Melas D,
Pal JS (2009a) A sensitivity study of the regional climate
model (RegCM3) to the convective scheme with emphasis in central eastern and southeastern Europe. Theor
Appl Climatol 97:327−337
Zanis P, Kapsomenakis I, Philandras C, Douvis K and others
(2009b) Analysis of an ensemble of present day and
future regional climate simulations for Greece. Int J Climatol 29:1614−1633
Zerefos C, Repapis C, Giannakopoulos C, Kapsomenakis J
and others (2011) The climate of the eastern Mediterranean and Greece: past, present and future. In: The environmental, economic and social impacts of climate
change in Greece. Climate Change Impacts Study Committee, Bank of Greece, Athens, p 1−126
Zittis G, Hadjinicolaou P, Lelieveld J (2014) Role of soil moisture in the amplification of climate warming in the eastern Mediterranean and the Middle East. Clim Res 59:
27−37
Submitted: September 10, 2014; Accepted: April 20, 2015
Proofs received from author(s): July 2, 2015