close

Вход

Забыли?

вход по аккаунту

?

j.jhydrol.2017.10.054

код для вставкиСкачать
Accepted Manuscript
Research papers
Historical trends and high-resolution future climate projections in Northern
Tuscany (Italy)
Marco D'Oria, Massimo Ferraresi, Maria Giovanna Tanda
PII:
DOI:
Reference:
S0022-1694(17)30730-8
https://doi.org/10.1016/j.jhydrol.2017.10.054
HYDROL 22334
To appear in:
Journal of Hydrology
Received Date:
Revised Date:
Accepted Date:
6 June 2017
4 September 2017
23 October 2017
Please cite this article as: D'Oria, M., Ferraresi, M., Tanda, M.G., Historical trends and high-resolution future climate
projections in Northern Tuscany (Italy), Journal of Hydrology (2017), doi: https://doi.org/10.1016/j.jhydrol.
2017.10.054
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers
we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and
review of the resulting proof before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Historical trends and high-resolution future climate projections in Northern Tuscany (Italy)
Marco D’Oriaa,1 , Massimo Ferraresia,2, Maria Giovanna Tandaa,3
a
Department of Engineering and Architecture, University of Parma, Parco Area delle Scienze
181/A, 43124, Parma, Italy
Abstract
This paper analyzes the historical precipitation and temperature trends and the future climate
projections with reference to the northern part of Tuscany (Italy). The trends are identified and
quantified at monthly and annual scale at gauging stations with data collected for long periods (6090 years). An ensemble of 13 Regional Climate Models (RCMs), based on two Representative
Concentration Pathways (RCP4.5 and RCP8.5), was then used to assess local scale future
precipitation and temperature projections and to represent the uncertainty in the results. The
historical data highlight a general decrease of the annual rainfall at a mean rate of 22 mm per
decade but, in many cases, the tendencies are not statistically significant. Conversely, the annual
mean temperature exhibits an upward trend, statistically significant in the majority of cases, with a
warming rate of about 0.1 °C per decade.
With reference to the model projections and the annual precipitation, the results are not concordant;
the deviations between models in the same period are higher than the future changes at medium(2031-2040) and long-term (2051-2060) and highlight that the model uncertainty and variability is
high.
According to the climate model projections, the warming of the study area is unequivocal; a mean
positive increment of 0.8 °C at medium-term and 1.1 °C at long-term is expected with respect to the
reference period (2003-2012) and the scenario RCP4.5; the increments grow to 0.9 °C and 1.9 °C
for the RCP8.5. Finally, in order to check the observed climate change signals, the climate model
1
Corresponding author. Tel.: +39 (0)521 905931; fax: +39 (0)521 905912.
e-mail address: marco.doria@unipr.it
2
e-mail address: massimo.ferraresi@unipr.it
3
e-mail address: mariagiovanna.tanda@unipr.it
1
projections were compared with the trends based on the historical data. A satisfactory agreement is
obtained with reference to the precipitation; a systematic underestimation of the trend values with
respect to the models, at medium- and long-term, is observed for the temperature data.
Keywords: climate change; historical trends; precipitation; temperature; Regional Climate Model
1
Introduction
In recent years, climate change and variability have received a great deal of attention by the
scientific community and their effects are already experienced in many parts of the world. The Fifth
Assessment Report (AR5) of the Intergovernmental Panel on Climate Change (IPCC, 2014)
denounces an unequivocal warming of the climate system, and since the 1950s, many of the
observed changes are unprecedented over decades to millennia. In the Northern Hemisphere, the
30-year period 1983-2012 was likely the warmest of the last 1400 years and the rate of warming of
the global mean surface temperature, in the period 1951-2012, is equal to 0.12 °C per decade.
According to the AR5, the main cause of the observed warming since the mid-20th century is
extremely likely to be the anthropogenic emissions of greenhouse gases. Independently from the
future emission scenarios, surface temperature is expected to increase over the 21st century. A mean
surface temperature rise of 0.3-0.7 °C is estimated for the period 2016-2035 relative to 1986-2005;
these values are expected to be in the range 0.3-4.8 °C for the period 2081-2100, depending on the
future emission scenarios. About precipitation, the AR5 reports non-uniform changes for the future
and in many cases they are comparable with the natural internal variability.
For the above reasons, the interest in mitigation and adaptation strategies to climate change is
increased (Martinez et al., 2012) and because precipitation and temperature are two of the main
components of the hydrological cycle, their investigation is essential. Both the historical trends on
the meteorological variables and the projections of climate models can be used to identify and
quantify changes in climate.
2
Trend analysis on observed precipitation and temperature data have been performed worldwide
from regional to local scale (e.g. Qin et al., 2010; Martinez et al., 2012; Irizarry-Ortiz et al., 2013;
Lavado Casimiro et al., 2013, Tian et al., 2016). With regard to Italy, Brunetti et al. (2006), based
on secular records updated to 2003, observed a general negative precipitation trend (the slopes of
the trends were calculated by least-square linear fitting). Considering the average all over Italy, the
Authors identified a decreasing trend of about 0.5% per decade in the annual mean; the variations,
spatially averaged over six regions, reach values up to -2% per decade in the different seasons but
the tendencies were rarely significant. About the temperature, the Authors identified an increasing
trend of 0.1 °C per decade in the annual values averaged all over Italy; the differences between
seasons, spatially averaged over three regions, are not significant (0.08-0.12 °C per decade).
According to Fatichi and Caporali (2009), in Tuscany (Italy) in the period 1916-2003, there is no
evidence of trend in the average precipitation regime; only few gauging stations show a very slight
decreasing trend in the total rainfall. The Authors, in a complex climatic system like central Italy,
recommend similar studies together with up-to-date rainfall measurement to continue checking the
possible consequences of global warming on the Tuscany precipitation regime. Bartolini et al.
(2008) studied the temperature trends in Tuscany looking only at the summer values; the Authors
found in the period 1955-2004 an increase in the minimum and maximum temperature of about
0.4 °C per decade, averaged over the entire region, with mean values ranging from 0.2 to 0.5 °C per
decade in the four clusters into which Tuscany was divided. The mean spatial variations are even
higher considering only the statistically significant results; the deviation between the trend at each
station and the corresponding cluster mean are up to ±0.3 °C per decade.
Ongoing trends in the observed climate variables can be used to develop future scenarios and
projections by means of a simple extrapolation (Mearns et al., 2001). This approach assumes that
the identified recent trends will continue in the future (Hennessy et al., 2011) and, even if there are
several reasons to carefully evaluate the results, they can be useful to assess the reliability of other
scenarios (such as climate model scenarios) especially over a short projection period (Mearns et al.,
3
2001). However, the extrapolation of climate variables is clearly not physically-based and does not
account for present or future greenhouse forcing. In order to overcome these limitations, Global
Circulation Models (GCMs) are adopted to simulate the response of the climate system to different
greenhouse gas concentrations (scenarios) and to obtain future projections. In AR5, the IPCC
adopted a new set of scenarios, named Representative Concentration Pathways (RCPs, see IPCC
(2013) for details), for the new climate model simulations carried out under the framework of the
Coupled Model Intercomparison Project Phase 5 (CMIP5, Taylor et al., 2012) of the World Climate
Research Programme. Usually, the GCM spatial resolution is approximately in the range 100-500
km and the outputs cannot be directly used in local studies where often the spatial domain is smaller
than a single GCM grid element. Many efforts have been made to dynamically downscale GCMs to
obtain higher resolution Regional Climate Models (RCMs) with a more realistic representation of
important surface heterogeneities (Kotlarski et al., 2014). The RCM resolution is typically 10-50
km and the output variables are generally available on a daily scale. On European scale,
international research projects such as PRUDENCE (Christensen et al., 2007), ENSEMBLE (van
der Linden and Mitchell, 2009) or the more recent EURO-CORDEX (Jacob et al., 2014) represent
some initiatives that provide RCM data. However, it is well established that temperature and
precipitation simulated by the RCMs are affected by systematic errors (Villani et al., 2015). RCM
simulations, especially in regions with complex topography, may not still be representative for the
local scale climate and may have inadequate spatial resolution for climate impact assessment
(Lafflamme et al., 2016). As a consequence, the RCM data need to be post-processed (using
downscaling techniques/bias correction methods) in order to produce high spatial resolution and
reliable estimates of local scale climate (Teutschbein and Seibert, 2012, Gudmundsson et al., 2012).
Moreover, it is recommended to use an ensemble of simulations (different combination of GCMs
and RCMs and forcing scenarios) to represent the uncertainty in the results (Sulis et al., 2012,
Teutschbein and Seibert, 2012).
4
The Italian National Institute for Environmental Protection and Research (ISPRA) applied four
RCMs provided in the framework of Med-CORDEX (Ruti et al., 2016) to obtain future climate
projections for Italy (Desiato et al., 2015). In this study, the Authors did not apply any downscaling
techniques or bias correction methods; the future climate variations were simply evaluated as
anomalies with respect to the modeled values in a reference period. The RCM grid resolution was
0.44° (~50 km) and the RCP4.5 and RCP8.5 scenarios were analyzed. The Authors found a mean
temperature rise of 2.5 °C (ensemble mean) for the RCP4.5 and 4.4 °C for the RCP8.5 in the period
2061-2090relative to the reference period 1971-2000, with higher increments in the summer season.
All the analyzed models predict a temperature rise in the future but significant differences are
identified between the climate projections. The spatial distribution of the temperature anomalies
differs between models; the ensemble mean markedly smooths the variations between regions over
Italy. Future precipitation anomalies are more uncertain than the temperature ones and the climate
models results are not concordant. Considering the ensemble mean, both the scenarios indicate, in
the period 2061-2090, a decrease of precipitation. The reduction is around 1.5% with respect to the
reference period 1971-2000 and this decline is concentrated in spring, summer and autumn; a slight
increase is forecasted in winter. With reference to Tuscany and the ensemble mean, decreasing
precipitation up to 50 mm are predicted for the RCP4.5 scenario in the period 2061-2090; the
reduction is slightly pronounced for the RCP8.5. Even if the previous results give an overview of
the situation in Italy, the RCM grid resolution was too low to highlight local scale climate
deviations. For this reason, the Authors recommend further investigations using RCMs with higher
grid resolution and downscaling techniques to obtain detailed local scale climate projections.
Local scale temperature and precipitation forecasts are becoming essential for understanding,
mitigating and adapting to the impact of climate change in different contexts such as the
hydrological cycle and the water resources. The Italian report on the National Strategy for Adapting
to Climate Change (Castellari et al., 2014) recalls that, although the causes of climate change are
predominantly global, the extent and the way in which these changes and their impacts occur are
5
typical of the local scale. The Authors encourage and suggest the use of RCMs in combination with
downscaling methods to obtain very high-resolution climate projections and confirm the need of an
ensemble of RCM simulations to highlight the uncertainty in the results.
The present work focuses on the northern part of Tuscany (Italy) and investigates the area served by
a local water company. The company is interested in understanding the potential impacts of climate
change on the water sources within the area in order to plan and start implementing effective
adaptation strategies. This study is preliminary to the quantification of the impacts and its main
contribution is to gain a better and conclusive understanding of the past, present and future climate
in the investigated area. This work attempts to detect signals of ongoing climate changes at local
scale and aims at assessing future scenarios based on both historical analyses and climate model
projections. In particular the study aims at: 1) identifying local trends in the observed precipitation
and temperature time series at gauging stations with data collected for long periods (60-90 years).
The trends are evaluated at monthly and annual scale; 2) assessing the effect of climate change on
the meteorological variables over the study area; 3) comparing the climate model projections with
the historical trends. We made the attempt to evaluate how well the results of climate models based
on different RCPs reproduce the values extrapolated from the trends, based on the historical data, at
different times.
The remainder of this paper is divided as follows. Section 2 describes the study area and presents
the available observed and climate model datasets. The methods used to analyze both the historical
and the RCM time series are then described in Section 3. Section 4 presents the results in terms of
historical trends and future projections of climate models and a comparison of the outcomes.
Section 5 provides a discussion of the results and draws some conclusions.
2
Study area and available data
The study area is located in the northern part of Tuscany (Italy). It covers about 3000 km2 and
extends from the Apennines to the Tyrrhenian Sea including also the Apuan Alps. As a result, the
6
topography is very complex ranging from mountains about 2000 m high to an extended coastline
(Figure1). Two main river basins (Magra and Serchio basins) are within the area together with five
smaller coastal river basins (Figure 1). 128 rain gauges and 94 temperature stations are located
within and surrounding the area (Figure 1) with daily data available up to 2012. However, only few
of them have a long record of historical observations (last 60-90 years) and are suitable for a trend
analysis. In the decade 2003-2012 the annual mean precipitation, averaged over the study area, was
about 1600 mm and the mean temperature was about 12 °C. As a result of the complex landscape,
precipitation and temperature show a great spatial variation; as an example Figure 2 shows the map
(Kriging interpolation together with the isohyets) of the annual mean rainfall in the period 20032012. A high heterogeneity is noticeable in the spatial distribution of precipitation; the annual mean
values range from about 900 to almost 2700 mm. Similarly, the temperature spatial variations are
large with annual mean values ranging from about 4 to 17 °C (decade 2003-2012).
Here Figure 1.
Here Figure 2.
According to the length of the time series, the quality of the data, the number of missing values and
the spatial distribution of the stations, we selected 18 rain and 14 temperature gauges to evaluate the
climate trends (Figure 3). In these sites, daily precipitation and daily temperature data were
available from 1916/1951 to 2012. Table 1 reports the characteristics of these gauges in terms of
altitude, record length and percentage of available daily data. With reference to the rain gauges, all
the selected stations have a percentage of missing value less than 25% with a high number of sites
where the percentage of available data approaches or exceeds 90%. Many temperature stations have
a percentage of available data greater than 70%; for few of them we accepted a percentage of gaps
up to 40% since temperature values show very high correlation between close stations (Section 3.1).
Here Table 1.
In addition to the instrumental data, we used climate model results to investigate future precipitation
and temperature. At this aim, the EURO-CORDEX project (Jacob et al., 2014), the European
7
branch of CORDEX (Giorgi et al., 2009), provides a multi-model ensemble of Regional Climate
Models (RCMs), driven by different General Circulation Model (GCM) projections, for Europe, at a
grid resolution of 0.11° (EUR-11, ~12.5 km) or 0.44° (EUR-44, ~50 km). Since the present study
focuses on local scale climate projections, we used 13 of those models available with the finer grid
resolution (EUR-11), resulting from the combination of the GCMs and the RCMs reported in
Table 2. A portion of the RCM grid superimposed to the study area is shown in Figure 3. The
climate model data are available for the period 1950/1970-2100 and consist of a historical
simulation until 2005 and a scenario period from 2006 to 2100. For this work, we considered daily
precipitation and mean daily temperature based on two of the Representative Concentration
Pathways (RCPs), adopted by the IPCC for the Fifth Assessment Report (AR5), the RCP4.5 and the
RCP8.5. We used the Gregorian calendar as a standard and those models that consider 360-day or
365-day calendars were converted.
Here Table 2.
Here Figure 3.
3
3.1
Methods
Historical data
We used the daily data collected for long periods at the gauging stations (Table 1) to identify trends
in the meteorological variables. Preliminarily, the missing values in the recorded precipitation and
temperature time series for one station were filled by means of records available in the best
correlated (according to the Pearson correlation coefficient) neighboring site making use of linear
relationships (Allen et al., 1998). In this stage, with the aim at filling the missing data in the stations
listed in Table 1, all the available gauges were considered. Then, we evaluated the trends in annual
and monthly precipitation and mean temperature time series by means of the nonparametric MannKendall (MK) test (Mann, 1945; Kendall, 1975). MK test is widely used to identify monotonic
trends in meteorological time series (e.g. Tian et al., 2013; Irizarry-Ortiz et al., 2013; Wang et al.,
8
2014) and, being rank-based, is not affected by the data distribution (Hamed, 2008). According to
the method, the null hypothesis (H) assumes that the data are independent and randomly ordered (no
trend) and this assumption is checked against the alternative hypothesis of presence of trend.
Because the existence of autocorrelation in the data can affect the test performance, we made use of
the modified MK test as proposed by Hamed and Rao (1998) and implemented by Fatichi (2009). A
rejection of the null hypothesis was evaluated at the 0.05 significance level.
The non-parametric Theil-Sen slope estimator (Sen, 1968) was then used to quantify linear trends in
time series. According to the method, the slope is defined as the median of the slopes of all possible
pairs of the considered data.
3.2
Regional Climate Model data
In this work, we made use of Regional Climate Model (RCM) data (at daily scale, Section 2) for
assessing future precipitation and temperature projections over the study area. In order to keep local
climate heterogeneities, also with the aim at using such data as inputs for the distributed or semidistributed hydrological models useful for the water resource assessment, the RCM outputs have
been downscaled at each weather station locations of Figure 1. In particular, an inverse distance
method (power of 2) was adopted to evaluate the site-specific time series; the cell that covers the
considered station and its eight neighboring cells were used in the interpolation process. However,
many literature studies (e.g. Teutschbein and Seibert, 2012; Smith et al., 2014) report that the
results of climate models are often affected by a systematic error; a bias correction of the RCM
simulated meteorological variables is needed to properly reproduce the observed time series in a
chosen control period (period in which both modeled and observed data are available).
Several bias correction methods are available in literature and a review of the most common
approaches can be found in e.g. Teutschbein and Seibert (2012). In this study we made use of the
Distribution Mapping method (also known as quantile-quantile mapping or statistical downscaling)
that is found to outperform other approaches in many literature works (e.g. Teutschbein and Seibert,
2012; Teng et al., 2015). We applied the correction to the local daily time series obtained
9
interpolating the RCM outputs at each site. According to the method, with reference to a control
period, the cumulative distribution function of simulated precipitation or temperature values is
corrected to agree with the one of the observed data. The same transfer function is then used to
modify the projected values since the assumption usually made is that the correction identified for
the control period holds for the scenario runs (Teutschbein and Seibert, 2013). We used the Gamma
distribution to model the wet-day rainfall (Piani et al., 2010) and the Gaussian distribution for the
temperature time series (Teutschbein and Seibert, 2012). The correction of the daily climate
variables was made on a monthly scale, i.e. for each month separately.
Following Teutschbein and Seibert (2012), with reference to the day d of the month m of the control
*
period, the corrected rainfall Pcontr is obtained from the RCM modeled rainfall Pcontr using the
following equation:
*
(d ) = Fγ−1 (Fγ (Pcontr(d) | αcontr,m ) | αobs,m )
Pcontr
−1
where Fγ is the Gamma cumulative distribution function and Fγ is its inverse, α contr,m and α obs ,m are
the Gamma distribution parameter vectors of the daily rainfall simulated by the RCM and observed
for the month m in the control period, respectively. The corrected daily precipitation for the
*
scenario period Pscen is then evaluated from the RCM modeled rainfall Pscen as:
*
(d ) = Fγ−1(Fγ (Pscen(d ) | αcontr,m ) | αobs,m )
Pscen
Prior to the Distribution Mapping correction, we adjusted the wet-day frequencies identifying an
RCM-specific threshold (Teutschbein and Seibert, 2012) such that the number of rainy days in the
model is equal to the one of the observed data. Then, all days with precipitation below the threshold
were considered dry.
The Distribution Mapping method, without any frequency adjustment, was adopted to correct the
bias in the temperature time series.
10
4
4.1
Results
Historical trends
We implemented the Mann-Kendall (MK) test and the Theil-Sen (TS) slope estimator for all the
selected 18 precipitation and 14 temperature time series, reported in Table 1, at monthly and annual
scale. For all sites, the analysis period ranges from the beginning of the observations (see Table 1)
to 2012.
Table 3 and Table 4 report the results of the MK test in terms of acceptance (no trend, H=A) or
rejection (trend exists, H=R) of the null hypothesis (0.05 significance level) and the trend gradient
per decade β as evaluated by the TS slope estimator for the precipitation and temperature time
series, respectively.
With reference to the annual values of Table 3, precipitation exhibits a negative trend for all sites
but Viareggio; however, in only 30% of cases the trend is statistically significant. The annual
average rate of decreasing precipitation is about 22 mm per decade. These results are highlighted in
Figure 4 that reports the values of the annual precipitation trends and the indication of their
significance at the station locations. At monthly scale, there are no clear indications of any
systematic tendency at all sites except for February, March and May. These months show an overall
decreasing trend but only in May the tendency is statistically significant in 70% of sites.
Here Table 3.
Here Table 4.
Here Figure 4.
The analyses on mean temperature, reported in Table 4, indicate a diffuse warming over the study
area at monthly and annual scale. The mean annual warming rate is about 0.1 °C per decade and the
trend is significant in 80% of sites. Only two time series show a decreasing trend in mean
temperature; Figure 5 summarizes these results. At monthly scale, January and May present the
mean highest positive trend gradients (0.16 °C per decade and 0.17 °C per decade, respectively) and
11
only September exhibits a decreasing trend (0.05 °C/decade) which, however, is not significant in
85% of cases.
4.2
Regional Climate Model data
We analyzed a set of 13 RCM simulations (Section 2), carried out in the framework of the EUROCORDEX project, to investigate the future climate.
As reported in Section 3.2, to evaluate the local climate over the study domain, the RCM outputs
were downscaled and bias corrected at 128 rain and 94 temperature gauges (Figure 1) that have an
adequate amount of data collected in the control period 1976-2005 (30 years). The correction
factors, identified for the control period, were then used to evaluate the climate in three decades: the
more recent-past period 2003-2012 with available observed data (selected as reference period) and a
medium- (2031-2040) and a long-term (2051-2060) decade. Both the RCP4.5 and the RCP8.5
scenarios were considered. The high spatial density of the weather stations allows obtaining
climatic projections with a fine resolution over the study area, therefore preserving the local
heterogeneities.
Figure 6 shows the maps (Kriging interpolation of the data at the 128 rain gauges) of the changes in
the annual precipitation, based on the RCM ensemble mean, for the decades 2031-2040 and 20512060 relative to 2003-2012 (note that the period 2006-2012 falls in the scenario runs, Section 2)
under the RCP4.5 scenario. At medium term, the variations range from +1% to -2.5%; however, it
is evident a general decrease in the annual mean precipitation over the study area with just small
sub-areas that experience an increase. A decrease in rainfall is projected at long-term over the entire
study domain with values up to -5%. For the sake of brevity, the maps obtained under the RCP8.5
scenario are not presented; in the following, only the results averaged over the study area are
discussed.
Here Figure 6.
Figure 7 shows the comparison between observed and RCM simulated precipitation, averaged over
the study area, for the reference period 2003-2012. The RCM outputs are reported in terms of
12
ensemble mean for both the scenarios (RCP4.5 and RCP 8.5) and as box-whisker plots (the
whiskers extend to the minimum and maximum of all data) for the RCP4.5. With reference to the
annual values, both the scenarios perform well in reproducing the observed data; the absolute
percentage error is about 2% according to the ensemble mean (RCP4.5). Although the rainfall
regime is properly simulated, more or less marked differences are observed at monthly scale. In
November and December all 13 model outputs underestimate the observed values but, as for the
other months, the differences are comparable with the interannual natural variability of precipitation
over the area (up to 50% in the annual mean and up to 150% in the monthly mean in the reference
period).
Figure 8 shows the medium- and long-term precipitation projections, averaged over the study area,
of the RCMs for the two RCP scenarios. The climate model outputs for the two future periods are
reported in terms of both box-whisker plots and ensemble mean. These values are compared with
the observations and the RCM ensemble mean of the reference period (2003-2012). Looking at the
annual values, the medium- and long-term precipitation does not show a significant deviation with
respect to the reference period values and the differences are comparable with the ensemble
variability. With reference to the RCM ensemble mean (compare green and red lines), the annual
precipitation for the period 2051-2060 and the RCP4.5 scenario decreases of about 3% compared to
the reference decade; no appreciable changes are detected for the RCP8.5 scenario. At monthly
scale, according to the RCP4.5 scenario, precipitation shows a gradual decrease in the future (with
respect to the model predictions in the reference period; compare green and red lines) during the
first 6-9 months of the year; the annual mean is then almost restored in the remaining months. For
the RCP8.5 scenario, a systematic rainfall decrease is detectable in May, June, August and
December. Nevertheless, the annual mean is restored by irregular fluctuations in the other months.
Table 5 reports a summary of the results with reference to the annual precipitation values in the
three analyzed decades.
Here Figure 7.
13
Here Figure 8.
Here Table 5.
As for precipitation, Figure 9 shows the maps (Kriging interpolation of the data at the 94
temperature gauges) of the changes in the annual mean temperature, based on the RCM ensemble
mean, for the decades 2031-2040 and 2051-2060 relative to 2003-2012 under the RCP4.5 scenario.
A diffuse warming is forecasted for the study area with increments ranging from 0.65 to 0.9 °C at
medium term and from 1.0 to 1.3 °C at long-term. The higher increments are located in the middle
of the studied area. Also in this case, the maps under the RCP8.5 scenario are omitted and the
results averaged over the study domain are discussed.
Here Figure 9.
Figure 10 shows the monthly and annual mean temperatures observed in the decade 2003-2012
compared with the same period forecasts of the climate models: ensemble mean for both the RCP
scenarios and box-whisker plots for the RCP4.5. The climate models are fully capable of
reproducing the thermometric regime and the observed monthly and annual values; the ensemble
variability and the differences between the two scenarios are rather contained.
Here Figure 10.
Figure 11 shows the medium- and long-term mean temperature projections, averaged over the study
area, of the RCMs for the two RCP scenarios. The climate model outputs for the two future periods
are reported in terms of both box-whisker plots and ensemble mean. These values are compared
with the observations and the RCM ensemble mean of the reference period (2003-2012).
The analysis of the RCM ensemble mean (compare green and red lines) for the RCP4.5 scenario
reveals an increase in the annual mean temperature of 0.8 °C between 2003-2012 and the mediumterm decade (2031-2040); the warming rises up to 1.1 °C in the long-term period (2051-2060).
According to the RCP8.5 scenario, it is expected an increase in the annual mean temperature of
0.9 °C and 1.9 °C in the medium- and long-term period, respectively. All climate models suggest a
growth in temperature; however, the ensemble variability, of the annual mean temperature,
14
increases over time. The mean temperature exhibits a gradual warming in all months from the
medium-term period to the long-term decade. The higher changes are predicted in summer (June,
July and August) and they increase from the RCP4.5 to the RCP8.5 scenario.
Table 6 summarizes the above findings with reference to the annual mean temperature in the three
analyzed decades.
Here Figure 11.
Here Table 6.
4.3
Comparison between observed trends and climate model projections
In order to check the observed climate change signals, we compared the climate model projections
with the observed trends based on the historical data. We assumed that the identified linear trends
would continue in the future with the same slope. The comparison was conducted for all the
precipitation and temperature stations of Table 1 at monthly and annual scale and for the reference
period (2003-2012), the medium- (2031-2040) and the long-term (2051-2060) decades. Both the
RCP4.5 and RCP8.5 scenarios were considered.
As an example, Figure 12 shows the comparison between the observed trends and the climate model
projections for the Massa temperature gauge for all months and the annual values. Each subplot of
Figure 12 reports the observed data, the linear trend, as estimated by the Theil-Sen method,
extrapolated at 2060 and the climate model projections, in terms of box-whiskers plots (the
whiskers extend to the minimum and maximum of all data), for the three considered decades and
the two RCP scenarios.
Here Figure 12.
From the analysis of Figure 12, it is evident that for a few months and for the three investigated
periods (2003-2012, 2031-204 and 2051-2060) the trend line intersects the box-whisker plots
obtained from the distribution of the climate model projections. In these cases, there is a good
agreement between the observed climate change indicators and the RCM forecasts. In some cases,
the extrapolated values from the trend line fall between the maximum and minimum values of the
15
climate model distribution, in others between the first and third quartiles (significant agreement). In
some months, such as September, the trend line deviates from the RCM projections; in particular,
the extrapolated values from the trend line underestimate the predictions of all the 13 RCMs.
Figure 13 and Figure 14 summarize, using a colored matrix, the results of the comparison for all
stations, months and analyzed decades with reference to the 18 precipitation and 14 temperature
sites, respectively. The figures, by means of a color scale, show where the extrapolated data from
the trend lines fall between the maximum and minimum values or between the first and third
quartile of the RCM distribution or underestimate or overestimate the climate model forecasts. A
number in each matrix cell indicates how many models (of the 13 RCMs) exceed the data
extrapolated from the trend line.
With reference to the precipitation (Figure 13), in most cases the extrapolated values from the trend
lines fall between the maximum and minimum of the 13 RCMs, with a high number of cases in
which the agreement is particularly significant (trends fall between the first and third quartile of the
climate model distributions). A systematic underestimation of the trend values with respect to the
RCMs, at medium- and long-term, is observed in March and May, for both the RCP scenarios, and
sometimes in April. The values extrapolated from the trend lines only few times overestimate the
climate model forecasts. Ultimately, all the analyzed stations show a more than acceptable
agreement between the trends identified from the historical data and the climate model projections
for the three analyzed decades.
Whereas, the temperature data (Figure 14) indicate that there is a good agreement between the
extrapolated values from the trend lines and the RCM projections only in the reference period
(2003-2012) with the exception of September where a systematic underestimation is observed
(Figure 14, top row). At medium- and even more at long-term, the values extrapolated from the
historical trends are frequently lower than the RCM projections; only sporadically, the trends
overestimate the climate model values. The disagreement between trend lines and RCMs increases
from the RCP4.5 to the RCP8.5 scenario.
16
Here Figure 13.
Here Figure 14.
5
Discussion and conclusions
This study investigated the historical precipitation and temperature trends and the future climate
projections in Northern Tuscany (Italy). The results of the analysis on the historical data showed a
general decrease of the annual rainfall at a mean rate of 22 mm per decade (about -1.4% per decade
relative to the 2003-2012 mean precipitation). However, according to the Mann-Kendall test, for
many sites and months, the tendencies are not statistically significant (0.05 significance level).
Conversely, the annual mean temperature, over the study area, exhibits an upward trend with a
warming rate of about 0.1 °C per decade. The tendencies are in the majority of cases statistically
significant.
With reference to the model projections and the annual precipitation, the results are not concordant;
the deviations between models in the same period (up to 14%, according to the first and third
quartile) are higher than the predicted changes, for given quartile, at medium- and long-term (up to
7%, with respect to the reference period). The differences between models are even higher
considering the extremes of the distribution of the model ensemble. Although the variations are
comparable with the natural fluctuations, these results highlight that the model uncertainty and
variability is high and comparable with or even greater than the predicted future changes. Therefore,
an ensemble of models, together with multiple emission scenarios, is needed to investigate the range
of possible variations and to obtain reliable results. The projections of a single model can lead to
misleading conclusions since it is not possible to statistically evaluate the outcomes. According to
the climate model projections, the warming of the study area is unequivocal; a mean positive
increment of 0.8 °C at medium-term and 1.1 °C at long-term is expected with respect to the
reference period (2003-2012) and the scenario RCP4.5; the increments grow to 0.9 °C and 1.9 °C
17
for the RCP8.5. The deviations between models are limited to about 4% at medium-term and about
5% at long-term, according to the first and third quartile of the distribution.
In this paper, we then extended the observed trends to the future comparing the results with the
ensemble of the climate model projections. A satisfactory agreement is obtained with reference to
the precipitation; in many cases, there is accordance between the values extrapolated from the
historical trends and the climate model projections. A systematic underestimation of the trend
values with respect to the models, at medium- and long-term, is observed for the temperature data.
However, both the historical tendencies and the model projections point out a warming of the
studied area. We are aware that extrapolating the observed trends to obtain future climate conditions
is not always reasonable or reliable, especially over a long projection period. Indeed, the tendencies
cannot be linear, they can be sensitive to the record length, they can be due to natural variability and
they do not account for future emission scenarios. However, if clear signals arise from the analysis
of the historical data, they should not be ignored but rather they can confirm the choice and the
reliability of the climate models.
This work provided a detailed local scale understanding of the past and future climate in the study
area; the downscaled/bias corrected projections allowed obtaining very high-resolution results. The
gauging station density is quite uniform and high (Figure 1) over the study domain; this is a novelty
in comparison with the previous literature at national (Desiato et al., 2015) and regional scale
(IPCC, 2013). The existing studies, because of their coarse spatial detail, cannot detect the climate
projection variability inside the study area. As an example, the temperature increments provided by
the available national projections (Desiato et al., 2015) appear constant for this sub-region.
Conversely, analyzing the annual mean temperature increase at medium- and long-term (depicted in
the contour lines of Figure 9), a maximum range (i.e. difference between minimum and maximum
increment) up to about 0.37 °C occurs. The spatial variation of temperature is not uniform during
the year and the monthly temperature range can vary from 0.2 to 0.8 °C over the study area
(according to the RCP4.5 scenario, whole data not shown for brevity), with the maximum in June at
18
both medium- and long-term. The maximum positive temperature increments are expected in June
reaching 1.7 °C at medium-term and 2.2 °C at long-term (RCP4.5 scenario). During the winter
(Dec, Jan, Feb) positive temperature increments are expected all over the study domain with values
varying from 0.4 to 1.3 °C at both medium- and long-term and the temperature range reaches 0.5 °C
(RCP4.5 scenario). In this small region, where high mountains exist and where the snowpack can
last for some months, supplying water in spring and summer, such differences have a great
importance since they can distinguish between snow and rain.
We think that the reliability of a study aimed at local scale impact of climate change and in areas
with complex topography, precipitation and temperature variations cannot avoid a high-resolution
analysis, especially for projects dealing with the assessment of water resources availability.
Acknowledgements
Financial assistant for this study was provided by GAIA S.p.A. (Italy) that supported the project
“Formulazione del bilancio idrologico del territorio di competenza di Gaia volto alla valutazione
delle variazioni indotte dal cambiamento climatico”. The authors are also grateful to GAIA S.p.A.
for the help during the data collection phase. The authors would like to thank the anonymous
Reviewers for the valuable and constructive comments on the initial version of this manuscript.
References
Allen R.G., Pereira L.S., Raes D., Smith M. (1998). Crop evapotranspiration - Guidelines for
computing crop water requirements - FAO Irrigation and drainage paper 56 - Annex 4. FAO, Rome.
ISBN: 92-5-104219-5
Bartolini, G., Morabito, M., Crisci, A., Grifoni, D., Torrigiani, T., Petralli, M., Maracchi, G.,
Orlandini, S. (2008). Recent trends in Tuscany (Italy) summer temperature and indices of extremes.
International Journal Climatology, 28, 1751-1760. DOI: 10.1002/joc.1673
19
Brunetti, M., Maugeri, M., Monti, F., Nanni, T. (2006). Temperature and precipitation variability in
Italy in the last two centuries from homogenised instrumental time series. International Journal
Climatology, 26, 345-381. DOI: 10.1002/joc.1251
Castellari S., Venturini S., Giordano F., Ballarin Denti A., Bigano A., Bindi M., Bosello F., Carrera
L., Chiriacò M.V., Danovaro R., Desiato F., Filpa A., Fusani S., Gatto M., Gaudioso D., Giovanardi
O., Giupponi C., Gualdi S., Guzzetti F., Lapi M., Luise A., Marino G., Mysiak J., Montanari A.,
Pasella D., Pierantonelli L., Ricchiuti A., Rudari R., Sabbioni C., Sciortino M., Sinisi L., Valentini
R., Viaroli P., Vurro M., Zavatarelli M. (2014). Elementi per una Strategia Nazionale di
adattamento ai Cambiamenti Climatici (Elements for a National strategy to adaptation to climate
changes). Ministero dell’Ambiente e della Tutela del Territorio e del Mare, Roma. In Italian. ISBN:
9788887728071
Christensen, J.H., Carter, T.R., Rummukainen, M., Amanatidis, G. (2007). Evaluating the
performance and utility of regional climate models: the PRUDENCE project. Climatic Change, 81,
1-6. DOI: 10.1007/s10584-006-9211-6
Desiato F., Fioravanti, G., Fraschetti, P., Perconti, W., Piervitali, E. (2015). Il clima futuro in Italia:
analisi delle proiezioni dei modelli regionali (The future climate in Italy: analysis of the Regional
model projections). The Italian National Institute for Environmental Protection and Research
(ISPRA), Stato dell’Ambiente, 58. In Italian. ISBN: 978-88-448-0723-8
Fatichi, S., Caporali, E. (2009). A comprehensive analysis of changes in precipitation regime in
Tuscany. International Journal Climatology, 29, 1883-1893. DOI: 10.1002/joc.1921
Fatichi, S. (2009). Mann-Kendall Modified test (version 1.0). File Exchange, MATLAB Central,
MathWorks. https://it.mathworks.com/matlabcentral/fileexchange/25531-mann-kendall-test
Giorgi, F., Jones, C., Asrar, G.R. (2009). Addressing climate information needs at the regional
level: the CORDEX framework. WMO Bulletin, 58, 175-183.
20
Gudmundsson, L., Bremnes, J.B., Haugen, J.E., Engen-Skaugen, T. (2012). Technical Note:
Downscaling RCM precipitation to the station scale using statistical transformations – a comparison
of methods. Hydrol. Earth Syst. Sci., 16, 3383-3390. DOI: 10.5194/hess-16-3383-2012, 2012.
Hamed, K.H. (2008). Trend detection in hydrologic data: The Mann–Kendall trend test under the
scaling hypothesis. Journal of Hydrology, 349, 350-363. DOI: 10.1016/j.jhydrol.2007.11.009
Hamed, K.H., Rao, A.R. (1998). A modified Mann-Kendall trend test for autocorrelated data.
Journal of Hydrology, 204, 182-196. DOI: 10.1016/S0022-1694(97)00125-X
Hennessy, K.J., Clarke, J., Ricketts, J.R. (2011). Methods for producing extreme temperature
projections for Australia. CAWCR Technical Report No. 038. The Centre for Australian Weather
and Climate Research. Melbourne. ISBN: 9781921826450
IPCC (2013). Climate Change 2013: The Physical Science Basis. Contribution of Working Group I
to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change [Stocker, T.F.,
D. Qin, G.-K. Plattner, M. Tignor, S.K. Allen, J. Boschung, A. Nauels, Y. Xia, V. Bex and P.M.
Midgley (eds.)]. Cambridge University Press, Cambridge, United Kingdom and New York, NY,
USA. ISBN 978-1-107-05799-1
IPCC (2014). Climate Change 2014: Synthesis Report. Contribution of Working Groups I, II and III
to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change [Core Writing
Team, R.K. Pachauri and L.A. Meyer (eds.)]. IPCC, Geneva, Switzerland. ISBN 978-92-9169-1432
Irizarry-Ortiz, M.M., Obeysekera, J., Park, J., Trimble, P., Barnes, J., Park-Said, W., Gadzinski, E.
(2013). Historical trends in Florida temperature and precipitation. Hydrological Processes, 27,
2225-2246. DOI: 10.1002/hyp.8259
Jacob, D., Petersen, J., Eggert, B., Alias, A., Christensen, O.B., Bouwer, L.M., Braun, A., Colette,
A., Déqué, M., Georgievski, G., Georgopoulou, E., Gobiet, A., Menut, L., Nikulin, G., Haensler,
21
A., Hempelmann, N., Jones, C., Keuler, K., Kovats, S., Kröner, N., Kotlarski, S., Kriegsmann, A.,
Martin, E., van Meijgaard, E., Moseley, C., Pfeifer, S., Preuschmann, S., Radermacher, C., Radtke,
K., Rechid, D., Rounsevell, M., Samuelsson, P., Somot, S., Soussana, J.-F., Teichmann, C.,
Valentini, R., Vautard, R., Weber, B., Yiou, P. (2014). EURO-CORDEX: new high-resolution
climate change projections for European impact research. Regional Environmental Change, 14,
563-578. DOI: 10.1007/s10113-013-0499-2
Kendall, M.G. (1975). Rank Correlation Methods (4th ed.). Griffin, London.
Kotlarski, S., Keuler, K., Christensen, O.B., Colette, A., Déqué, M., Gobiet, A., Goergen, K., Jacob,
D., Lüthi, D., van Meijgaard, E., Nikulin, G., Schär, C., Teichmann, C., Vautard, R., Warrach-Sagi,
K., Wulfmeyer, V. (2014). Regional climate modeling on European scales: a joint standard
evaluation of the EURO-CORDEX RCM ensemble. Geoscientific Model Development, 7, 12971333. DOI: 10.5194/gmd-7-1297-2014
Laflamme, E.M., Linderb, E., Panb, Y. (2016). Statistical downscaling of regional climate model
output to achieve projections of precipitation extremes. Weather and Climate Extremes, 12, 15-23.
DOI: 10.1016/j.wace.2015.12.001
Lavado Casimiro, W.S., Labat, D., Ronchail, J., Espinoza, J. C., Guyot, J. L. (2013). Trends in
rainfall and temperature in the Peruvian Amazon–Andes basin over the last 40 years (1965–2007).
Hydrological Processes, 27, 2944-2957. DOI: 10.1002/hyp.9418
Mann, H.B. (1945). Nonparametric tests against trend. Econometrica, 13, 245 259. DOI:
10.2307/1907187
Martinez, C.J., Maleski, J.J., Miller, M.F. (2012). Trends in precipitation and temperature in
Florida, USA. Journal of Hydrology, 452-453, 259-281. DOI: 10.1016/j.jhydrol.2012.05.066
Mearns, L.O., Hulme, M., Carter, T.R., Leemans, R., Lal, M., Whetton P. (2001). Climate Scenario
Development. Climate Change 2001: The Scientific Basis. Contribution of Working Group I to the
22
Third Assessment Report of the Intergovernmental Panel on Climate Change [Houghton, J.T.,Y.
Ding, D.J. Griggs, M. Noguer, P.J. van der Linden, X. Dai, K. Maskell, and C.A. Johnson (eds.)].
Cambridge University Press, Cambridge, United Kingdom and New York, NY, USA. ISBN:
0521014956
Piani, C., Haerter, J.O., Coppola, E. (2010). Statistical bias correction for daily precipitation in
regional climate models over Europe. Theoretical and Applied Climatology, 99, 187–192. DOI:
10.1007/s00704-0090134-9, 2010
Qin, N.X., Chen, X., Fu, G.B., Zhai, J.Q., Xue, X.W. (2010). Precipitation and temperature trends
for the Southwest China: 1960–2007. Hydrological Processes, 24, 3733-3744. DOI:
10.1002/hyp.7792
Ruti, P.M., Somot S., Giorgi F., Dubois C., Flaounas E., Obermann A., Dell'Aquilla A., Pisacane
G., Harzallah A., Lombardi E., Ahrens B., Akhtar N., Alias A., Arsouze T., Aznar R., Bastin S.,
Bartholy J., Béranger K., Beuvier J., Bouffies-Cloché S., Brauch J., Cabos W., Calmanti S., Calvet
J.-C., Carillo A., Conte D., Coppola E., Djurdjevic V., Drobinski P., Elizalde-Arellano A., Gaertner
M., Galàn P., Gallardo C., Gualdi S., Goncalves M., Jorba O., Jordà G., L'Heveder B., LebeaupinBrossier C., Li L., Liguori G., Lionello P., Maciàs D., Nabat P., Önol B., Raikovic B., Ramage K.,
Sevault F., Sannino G., Struglia M.V., Sanna A., Torma C., and Vervatis V. (2016). MEDCORDEX initiative for Mediterranean Climate studies. Bulletin of the American Meteorological
Society, 97, 1187-1208. DOI: 10.1175/BAMS-D-14-00176.1
Sen, P.K. (1968). Estimates of the regression coefficient based on Kendall’s tau. Journal of the
American Statistical Association, 63, 1379-1389. DOI: 10.2307/2285891
Smith, A., Bates, P., Freer, J., Wetterhall, F. (2014). Investigating the application of climate models
in flood projection across the UK. Hydrological Processes, 28, 2810-2823. DOI: 10.1002/hyp.9815
23
Sulis M., Paniconi, C., Marrocu, M., Huard, D., Chaumont, D. (2012). Hydrologic response to
multimodel climate output using a physically based model of groundwater/surface water
interactions. Water Resource Research, 48, W12510. DOI: 10.1029/2012WR012304
Taylor, K.E., Stouffer, R.J., Meehl, G.A. (2012). An overview of CMIP5 and the experiment
design. Bulletin of the American Meteorological Society, 93, 485-498. DOI: 10.1175/BAMS-D-1100094.1
Teng, J., Potter, N.J., Chiew, F.H.S., Zhang, L., Vaze, J., Evans, J.P. (2015). How does bias
correction of RCM precipitation affect model runoff? Hydrology and Earth System Sciences, 19,
711-728. DOI: 10.5194/hess-19-711-2015
Teutschbein, C., Seibert J. (2012). Bias correction of regional climate model simulations for
hydrological climate-change impact studies: review and evaluation of different methods. Journal of
Hydrology, 456-457, 12-29. DOI: 10.1016/j.jhydrol.2012.05.052
Teutschbein, C., Seibert J. (2013). Is bias correction of regional climate model (RCM) simulations
possible for non-stationary conditions? Hydrology and Earth System Sciences, 17, 5061-5077. DOI:
10.5194/hess-17-5061-2013
Tian, Q., Prange, M., Merkel, U. (2016). Precipitation and temperature changes in the major
Chinese river basins during 1957–2013 and links to sea surface temperature. Journal of Hydrology,
536, 208-221. DOI: 10.1016/j.jhydrol.2016.02.048
van der Linden P., Mitchell J.F.B. (2009). ENSEMBLES: climate change and its impacts: summary
of research and results from the ENSEMBLES project. Met Office Hadley Centre, FitzRoy Road,
Exeter EX1 3PB, UK.
Villani, V., Rianna, G., Mercogliano, P., Zollo, A.L., Schiano, P. (2015). Statistical approaches
versus weather generator to downscale RCM outputs to point scale: a comparison of performances.
24
Journal of Urban and Environmental Engineering, 8, 142-154. DOI:
10.4090/juee.2014.v8n2.142154
Wang X., Yang X., Liu T., Li F., Gao R., Duan L., Luo Y. (2014). Trend and extreme occurrence of
precipitation in a mid-latitude Eurasian steppe watershed at various time scales. Hydrological
Processes, 28, 5547-5560. DOI: 10.1002/hyp.10054
25
Figure captions.
Figure 1. Study domain and river basins within the area. Rain and temperature gauges within and
surrounding the area.
Figure 2. Map and isohyets of the annual mean rainfall in the period 2003-2012 over the study area.
Figure 3. Study area and portion of EUR-11 RCM grid. Rain and temperature gauges used for trend
analysis.
Figure 4. Rate of the annual precipitation trend (mm/decade) and indication of trend significance at
0.05 level at all sites. The symbol size is proportional to the trend gradient.
Figure 5. Rate of the annual mean temperature trend (°C/decade) and indication of trend
significance at 0.05 level at all sites. The symbol size is proportional to the trend gradient.
Figure 6. Maps of the variations in the annual precipitation, based on the RCM ensemble mean, for
the decades 2031-2040 and 2051-2060 relative to 2003-2012 under the RCP4.5 scenario.
Figure 7. Monthly and annual precipitation, averaged over the study area, for the reference period
(2003-2012): observed values and RCM projections. The RCM outputs are reported in terms of
ensemble mean for both the scenarios (RCP4.5 and RCP 8.5) and as box-whisker plots (the
whiskers extend to the minimum and maximum of all data) for the RCP4.5.
Figure 8. Monthly and annual precipitation averaged over the study area: observed and RCM
ensemble mean in the period 2003-2012, RCM projections for the medium- (2031-2040) and longterm (2051-2060) period. The future RCM outputs are reported in terms of ensemble mean and as
box-whisker plots (the whiskers extend to the minimum and maximum of all data) for both the
RCP4.5 and RCP 8.5 scenarios.
Figure 9. Maps of the variations in the annual mean temperature, based on the RCM ensemble
mean, for the decades 2031-2040 and 2051-2060 relative to 2003-2012 under the RCP4.5 scenario.
26
Figure 10. Monthly and annual mean temperature, averaged over the study area, for the reference
period (2003-2012): observed values and RCM projections. The RCM outputs are reported in terms
of ensemble mean for both the scenarios (RCP4.5 and RCP 8.5) and as box-whisker plots (the
whiskers extend to the minimum and maximum of all data) for the RCP4.5.
Figure 11. Monthly and annual mean temperature averaged over the study area: observed and RCM
ensemble mean in the period 2003-2012, RCM projections for the medium- (2031-2040) and longterm (2051-2060) period. The future RCM outputs are reported in terms of ensemble mean and as
box-whisker plots (the whiskers extend to the minimum and maximum of all data) for both the
RCP4.5 and RCP 8.5 scenarios.
Figure 12. Comparison between the extrapolated historical trends up to 2060 and the projections of
the RCM models (in terms of box-whisker plots) for the three analyzed decades: 2003-2012, 20312040 and 2051-2060, with reference to the Massa temperature station. The range of the y-axis in all
the images is the same to easily compare the trend gradients.
Figure 13. Precipitation gauging stations: comparison between the extrapolated values from the
trend lines and the RCM data distribution, at monthly and annual scale and for the three analyzed
decades (2003-2012, 2031-2040 and 2051-2060). The meaning of the colors is reported in the
legend at the top of the figure. The number in each matrix cell indicates how many climate model
values (of the 13 RCMs) exceed the data extrapolated from the trend line.
Figure 14. Temperature gauging stations: comparison between the extrapolated values from the
trend lines and the RCM data distribution, at monthly and annual scale and for the three analyzed
decades (2003-2012, 2031-2040 and 2051-2060). The meaning of the colors is reported in the
legend at the top of the figure. The number in each matrix cell indicates how many climate model
values (of the 13 RCMs) exceed the data extrapolated from the trend line.
27
Tables.
Table 1. Precipitation and temperature gauge characteristics: altitude, record length and percentage
of available daily data.
Altitude
(m a.s.l.)
385
250
544
100
1340
402
112
343
23
285
439
5
20
65
763
119
215
251
622
26
603
3
500
133
Station name
Arlia
Bagnone
Bedonia
Borgo A Mozzano
Boscolungo
Calice Al Cornoviglio
Carrara
Casania
Castelmartini
Castelnuovo Garfagnana
Cembrano
La Spezia
Lucca
Massa
Palagnana
Pescia
Pontremoli
Pontremoli Verdeno
S. Marcello Pistoiese
Sarzana
Tavarone
Viareggio
Villacollemandina
Villafranca Lunigiana
Precipit.
record length
1934-2012
1919-2012
1921-2012
1921-2012
1922-2012
1921-2012
1925-2012
1916-2012
1916-2012
1923-2012
1921-2012
1916-2012
1921-2012
1921-2012
1920-2012
1921-2012
1921-2012
1920-2012
Available precip.
data (%)
87.1
76.1
99.2
90.2
95.5
87.7
83.6
88.4
91.2
92.3
87.9
86.7
97.1
99.5
93.1
89.7
99.9
80.1
Temper. record
length
1933-2012
1930-2012
1929-2012
1934-2012
1951-2012
1933-2012
1932-2012
1934-2012
1933-2012
1932-2012
1933-2012
1926-2012
1934-2012
1930-2012
-
Available temp.
data (%)
72.3
61.6
60.9
86.5
89.4
62.7
80.6
92.4
78.7
87.3
62.4
86.8
62.7
72.4
-
Table 2. Combination of GCMs and RCMs from the EURO-CORDEX project used in this study.
More information is available at www.euro-cordex.net.
RCM
CCLM4-8-17
HIRHAM5
WRF331F
RACMO22E
RCA4
CNRM-CM5
X
EC-EARTH
X
X
GCM
HadGEM2-ES
X
MPI-ESM-LR
X
IPSL-CM5A-MR
X
X
X
X
X
X
X
X
28
Table 3. Precipitation time series: monthly and annual results of the MK test and TS slope
estimator. H indicates the acceptance (A, no trend) or rejection (R, trend exists) of the MK test null
hypothesis (0.05 significance level). β (mm/decade) is the trend gradient.
Site
Arlia
Bagnone
Bedonia
Borgo a Mozzano
Calice al Cornoviglio
Carrara
Casania
Cembrano
Lucca
Massa
Palagnana
Pescia
Pontremoli Verdeno
S. Marcello Pistoiese
Sarzana
Viareggio
Villacollemandina
Villafranca Lunigiana
H
A
A
A
A
A
A
A
A
R
A
A
A
A
A
A
A
A
A
Jan
β
-3.98
-1.09
0.39
-0.67
2.06
-0.64
-2.36
-0.58
-5.99
0.00
-3.30
-2.59
1.77
-0.43
-0.82
-0.39
0.27
-0.01
H
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
Feb
β
-4.01
-1.74
-3.45
-5.10
-2.02
-3.35
-4.39
-3.28
-2.61
-2.08
-8.25
-0.93
-3.42
-5.17
-0.73
-2.70
-3.20
-3.60
Mar
H β
A -3.42
R -7.62
R -6.49
A -2.86
A -6.37
R -6.54
A -7.07
R -9.10
R -6.14
A -2.33
A -5.39
A -4.80
R -6.43
A -5.77
A -2.88
A -2.32
A -3.31
R -9.10
H
A
A
A
A
A
A
A
A
A
A
A
A
A
R
A
A
A
A
Apr
β
0.20
-2.77
1.42
-1.46
0.22
-3.72
0.43
-1.26
-1.42
-0.54
-2.51
-1.77
0.96
-3.68
2.44
0.35
-1.56
-6.01
H
A
R
R
A
R
R
R
A
R
R
R
A
R
R
R
R
A
R
May
β
-5.00
-5.63
-4.78
-4.25
-9.36
-6.39
-10.83
-5.33
-4.23
-5.15
-9.32
-2.51
-8.06
-4.98
-4.86
-4.47
-4.60
-8.27
H
A
A
A
A
A
A
A
A
A
A
A
A
A
A
R
A
A
A
Jun
β
2.43
-0.67
-1.72
-0.23
-1.29
0.94
-1.94
-1.12
0.39
1.76
-3.05
0.25
-1.22
-0.95
2.00
1.01
-0.23
-3.00
H
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
Jul
β
-0.41
-1.17
0.70
-0.29
0.82
-0.20
-3.14
0.21
-0.05
-0.07
-0.29
-0.23
-2.31
-0.50
0.30
0.10
-2.11
-0.85
Aug
H β
A -0.91
A -0.31
A 0.42
A 1.00
A 0.58
A 0.80
A -0.30
R 2.30
A -0.17
A 1.03
A 2.54
A 0.00
A 0.72
A 1.28
A 0.35
A 1.87
A 0.77
A -0.13
H
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
Sep
β
1.69
0.09
0.43
-0.52
0.20
4.04
-1.18
-1.67
2.21
1.63
3.40
0.84
-2.27
-1.00
2.44
3.31
-1.45
-2.27
H
A
A
R
A
A
A
A
A
A
A
A
A
A
A
A
R
A
A
Oct
β
-0.33
1.16
5.70
-2.65
0.82
-1.78
-8.93
1.69
-3.94
0.16
-0.39
-1.65
2.92
-3.32
-0.37
5.07
-0.54
-2.59
H
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
Nov
β
2.28
0.74
3.08
1.00
-0.71
-2.09
-10.13
-0.82
-1.01
-0.61
-8.04
0.64
2.36
-0.98
2.34
1.17
1.92
-4.49
H
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
Dec
β
-0.44
-0.85
-2.14
0.63
-1.69
-0.31
-0.22
0.75
-1.33
3.13
3.03
1.54
-0.16
-1.42
0.72
1.45
0.42
0.23
H
A
R
A
A
A
A
R
R
A
A
R
A
A
R
A
A
A
R
Year
β
-13.71
-33.39
-13.60
-17.59
-19.00
-21.85
-61.94
-27.01
-21.49
-9.00
-40.49
-11.16
-20.09
-27.96
-3.55
3.84
-14.82
-52.15
Table 4. Temperature time series: monthly and annual results of the MK test and TS slope
estimator. H indicates the acceptance (A, no trend) or rejection (R, trend exists) of the MK test null
hypothesis (0.05 significance level). β (°C/decade) is the trend gradient.
Site
Arlia
Bagnone
Boscolungo
Calice al Cornoviglio
Castelmartini
Castelnuovo Garf.
La Spezia
Lucca
Massa
Pontremoli
S. Marcello Pistoiese
Sarzana
Tavarone
Viareggio
H
A
R
R
R
A
R
R
R
R
A
R
A
R
A
Jan
β
0.03
0.33
0.19
0.17
-0.08
0.32
0.34
0.11
0.35
0.00
0.14
0.12
0.19
0.07
H
A
R
A
A
R
R
R
A
R
R
A
A
A
R
Feb
β
-0.12
0.25
0.10
0.09
-0.16
0.16
0.27
0.00
0.21
-0.17
0.01
0.07
0.06
0.07
Mar
H β
A -0.06
R 0.22
A 0.12
A 0.16
A 0.08
R 0.17
R 0.28
A 0.06
R 0.24
R -0.17
A 0.08
A 0.09
A 0.11
R 0.11
H
R
A
A
A
A
A
R
A
R
R
A
A
A
A
Apr
β
-0.17
0.05
0.02
0.04
0.04
0.06
0.18
0.07
0.14
-0.30
-0.09
0.00
-0.03
0.09
May
H β
A 0.05
R 0.17
R 0.16
R 0.20
R 0.21
R 0.25
R 0.32
R 0.21
R 0.28
R -0.15
A 0.13
A 0.13
R 0.20
R 0.22
H
A
A
A
R
A
R
R
R
R
R
A
A
A
R
Jun
β
-0.07
0.01
0.04
0.11
0.16
0.15
0.17
0.14
0.18
-0.20
0.05
0.06
0.08
0.15
H
R
A
A
R
A
R
A
R
R
R
A
A
A
R
Jul
β
-0.10
0.02
0.10
0.15
0.14
0.11
0.10
0.14
0.12
-0.22
0.01
0.12
0.08
0.15
Aug
H β
A -0.05
R 0.17
A 0.11
R 0.22
R 0.28
R 0.24
R 0.14
R 0.24
R 0.24
A -0.17
A 0.05
R 0.16
R 0.21
R 0.19
H
R
A
A
A
A
A
A
A
A
R
A
A
A
A
Sep
β
-0.25
0.05
-0.08
-0.01
-0.01
0.02
-0.03
0.04
0.08
-0.33
-0.10
-0.06
0.01
-0.01
H
A
R
A
R
A
R
R
R
R
R
A
R
R
R
Oct
β
-0.10
0.22
0.09
0.08
0.02
0.17
0.13
0.09
0.18
-0.19
0.03
0.10
0.15
0.08
H
A
R
A
A
A
R
R
A
R
R
A
A
A
A
Nov
β
-0.09
0.26
0.04
0.03
0.00
0.12
0.18
-0.01
0.23
-0.15
0.03
0.02
0.08
0.02
H
A
R
A
A
A
R
R
A
R
A
A
A
R
A
Dec
β
-0.07
0.31
0.09
0.08
-0.18
0.16
0.23
-0.02
0.27
-0.07
0.07
0.08
0.13
0.08
H
R
R
R
R
A
R
R
R
R
R
A
R
R
R
Year
β
-0.08
0.18
0.10
0.13
0.06
0.17
0.20
0.10
0.22
-0.18
0.04
0.10
0.11
0.12
29
Table 5. Annual precipitation, averaged over the study area, for the three analyzed decades: 20032012, 2031-2040 and 2051-2060. The RCM results are reported in terms of the first three quartiles
and the ensemble mean with reference to the RCP4.5 and RCP8.5 scenarios.
Observ.
2003-2012
2031-2040
2051-2060
1572
First
quart.
1503
1529
1454
Annual precipitation (mm)
RCMs - RCP4.5
Third
Ensemble
First
Median
quart.
mean
quart.
1586
1707
1606
1542
1653
1683
1591
1532
1516
1650
1550
1489
RCMs - RCP8.5
Third Ensemble
Median
quart.
mean
1593
1613
1577
1584
1725
1599
1516
1601
1570
Table 6. Annual mean temperature, averaged over the study area, for the three analyzed decades:
2003-2012, 2031-2040 and 2051-2060. The RCM results are reported in terms of the first three
quartiles and the ensemble mean with reference to the RCP4.5 and RCP8.5 scenarios.
Observ.
2003-2012
2031-2040
2051-2060
12.1
First
quart.
12.1
12.8
13.2
Annual mean temperature (°C)
RCMs - RCP4.5
Third
Ensemble
First
Median
quart.
mean
quart.
12.2
12.4
12.3
12.1
13.2
13.3
13.1
13.0
13.5
13.6
13.4
13.8
RCMs - RCP8.5
Third Ensemble
Median
quart.
mean
12.2
12.3
12.2
13.1
13.3
13.1
14.1
14.5
14.1
30
Highlights.
We analyze historical precipitation and temperature trends in northern Tuscany.
Regional Climate Models are used to assess local scale future climate projections.
An ensemble of 13 RCMs is used to represent the uncertainty in the results.
Downscaled/bias corrected projections allow obtaining very high-resolution results.
31
Figure 1
Click here to download high resolution image
Figure 2
Click here to download high resolution image
Figure 3
Click here to download high resolution image
Figure 4
Click here to download high resolution image
Figure 5
Click here to download high resolution image
Figure 6
Click here to download high resolution image
Figure 7
Click here to download high resolution image
Figure 8
Click here to download high resolution image
Figure 9
Click here to download high resolution image
Figure 10
Click here to download high resolution image
Figure 11
Click here to download high resolution image
Figure 12
Click here to download high resolution image
Figure 13
Click here to download high resolution image
Figure 14
Click here to download high resolution image
Документ
Категория
Без категории
Просмотров
4
Размер файла
1 310 Кб
Теги
2017, 054, jhydrol
1/--страниц
Пожаловаться на содержимое документа