INFORMATION SHEET ON FUTURE CLIMATE AND IMPACTS IN THE COASTAL CASE STUDIES: GULF OF GABÈS, TUNISIA

Summary► Mean air temperature over the Gulf of Gabès is projected to increase during the period 1950-2050.

Daily minimum and maximum temperatures show a similar increase. Precipitation is decreasing slightly. The simulated trends are very weak before the year 2000 and the rate of increase accelerates thereafter. There are no significant trends for wind speed and relative humidity.

► The Gulf of Gabès waters are warming, also at higher rates after the year 2000. Salinity is increasing at a very low rate. There is evidence for salinity decrease at long time-scales. Sea level is increasing mainly due to the steric effect.

► As a consequence of water warming alien marine species are projected to increase.

► Similar to observations, the summer season is projected to lengthen leading to an extended “tourist season”.

► Similar to observations, the number of days “favourable” for tourism activities deduced from climate projections, decreases whereas the number of those just “acceptable” increases. Such changes are accentuated in summer; in contrast the number of “favourable” days increases in winter.

► Socio economic impacts of changes in climate conditions on tourism activity are estimated based on model simulations. They show that impacts increase after the year 2000.

1. Introduction This information sheet is the third one in the series on the present and future impacts of climate change over the Gulf of Gabès. The first information sheet focused on observed climate-marine conditions (Harzallah, 2010). It showed that mean air temperature is increasing over the Gulf of Gabès. Daily maximum air temperature is also increasing particularly in summer and autumn. Rainfall showed no clear trends. The summer season has lengthened over the observation period with a tendency for an earlier start date and a later end date. The second information sheet (Harzallah et al., 2010) showed important shoreline modifications along the Gulf of Gabès coasts, although these were mainly related to human activity. An increase in the annual maximum swell height deduced from wind statistics together with the observed sea-level rise may reinforce the vulnerability of the coasts. The number of alien marine species discovered in the Gulf is found to increase, which may negatively impact biodiversity in the

Gulf and possibly the fishing activity in the region. The tourism sector may be impacted by beach erosion and by the increase of ‘favourable’ days in winter and their decrease in summer.

This information sheet shows climate projections for the present and future periods in the Gulf of Gabès (1950-2050), based on the most recent set of model simulations available, including for the first time a coupling of the Mediterranean Sea with the atmosphere. The set of atmospheric and oceanic variables simulated by the models are presented and compared to the observed ones where available. Biogeophysical impacts are investigated along with some economic and social future trends.

Indicators are presented for: ► Climate conditions► Marine conditions► Alien marine species► Seasonal shift Index► Daily Climate Tourism Potential Index.► Socio-economic tourism impacts.

1

Climate models

The CIRCE project climate model runs (see Information sheet on future climate: Climate projections for the CIRCE case studies, http://www.cru.uea.ac.uk/projects/circe/Future_infosheet_overview_final.doc) are used to investigate present and future climate changes in the Gulf of Gabès for the period, 1950-2050. After the year 2000, the A1B scenario for the GHG aerosols concentration is used (Nakićenović et al, 2000). The CIRCE climate models are of differing spatial resolution and the areas covered for the Gulf of Gabès are shown in Figure 1. The following coupled atmosphere-ocean CIRCE models are used: ENEA; ENEA-ERA40-2 (using observation reanalyses for the period 1958-2000); INGV; LMDglo (also referred to as IPSL1 elsewhere); MF (also referred to as CNRM elsewhere); and MPI.

Figure 1: Gulf of Gabès areas covered by the atmospheric and oceanic components of the CIRCE climate models. Left (atmosphere); right (ocean)

In addition, two ocean model runs (INSTMED06 and INSTMCOTR) are used for the three 10-year periods (1960-1969, 1991-2000 and 2050-2059).

2. Climate and marine projections

Climate conditions

What is it? Figure 2 shows key weather variables in the Gulf of Gabès simulated by the ensemble of CIRCE models for the period, 1950-2050. Mean values and trends during the ‘present climate’, 1961-1990 and the ‘mid-century’, 2021-2050 are shown in Table 1. Also shown are the ‘long-term’, mean changes and mean change rates between these two periods. The Mann-Kendall test (Mann, 1945; Kendall, 1975) and the student-t test are used to asses the statistical significance of trends and of the mean changes (and mean change rates). These tests are also used to asses the statistical significance of the subsequent series shown in the present information sheet.

Figure 2: Simulated atmospheric variables in the Gulf of Gabès (averaged over the areas shown in Figure 1) from a set of numerical models for the period, 1950-2050. The black line is the average series based on the different simulations (except ENA-ERA40-2). Details on models and simulations are shown above. The variables are mean, minimum and maximum air temperature (2m), precipitation, wind speed and relative humidity

2. Climate and marine projections

Table 1: Mean values and trends of key variables simulated by the five coupled models in the Gulf of Gabès area (Figure 1) based on ensemble averages, for the present 1961-1990 and mid-century 2021-2050 periods. Mean long-term changes between the periods 1961-1990 and 2021-2050 are also shown. Inter-model ranges are shown as the standard deviations between estimates from the five models. Bold values are significant.

Variable present climate (1961-1990)

mid-century (2021-2050)

Long-term (1961/1990- 2021/2050)

T mean (°C) (°C/decade)

17.1±0.1+ 0.013±0.0120

18.4±1.0 + 0.54±0.021

+1.4±0.50

+ 0.23±0.080

T max (°C) (°C/decade)

20.6±1.7+ 0.003 ±0.140

21.9±1.5+ 0.55±0.191

+ 1.4±0.40

+ 0.22±0.070

T min (°C) (°C/decade)

14.2±1.3 + 0.02 ±0.120

15.6±1.3+ 0.55±0.151

+1.4±0.50

+ 0.23±0.080

Precipitation (mm) (mm/ decade)

163±51- 1.8 ±160

147±47-8.1±120

-16±100

- 2.7±1.70

Wind speed (m/s) (m/s/ decade)

4.3±0.5+ 0.01 ±0.040

4.25±0.5- 0.005±0.10

-0.05±0.060

-0.008±0.010

Relative humidity (%) (%/decade)

61.0±4.0+ 0.1 ±0.30

60.7±3.5- 0.04±0.70

-0.3±0.70

-0.05±0.120

1: Inter-model Mann-Kendall trend test: the null hypothesis of an absence of trend is rejected at the 0.05 significance level; 0: a failure to reject the null hypothesis.1: Inter-model Student’s t-test: the null hypothesis means are equal is rejected at the 0.05 significance level; 0: a failure to reject the null hypothesis.

What does this show? Mean air temperature is projected to increase at an accelerating rate. The different models show similar behaviour. The observed annual temperature for the period 1961-1990 from data archived at the Goddard Institute for Space Studies (GISS, www.giss.nasa.gov) range from 19 to 20°C at Gabès and 18-19°C at Sfax. The coupled models slightly underestimate the temperature in the Gulf of Gabès (a general characteristic of the CIRCE simulations showing a cool bias of around -2°C over lands). However, model data cover the areas shown in Figure 1, whereas observations are for stations. The simulated trends are also slightly less than the observed ones and show that the mean, minimum and maximum temperatures increase by around 1.4 °C between 1961-1990 and 2021-2050 and 2°C from 1950 to 2050. Temperature trends are very weak during the ‘present’ period and are stronger and significant during the ‘mid-century’ period. The simulated precipitation series show a generally declining trend after the year 2000; however, trends are not significant. In

addition, wind speed and relative humidity show no clear trends.

2. Climate and marine projections

Why is it relevant?The simulated trends in mean temperature for ’present’ climate conditions are lower than observed ones (+0.013°C/decade simulated in 1961-1990 versus +0.26°C/decade for the 1948-2008 and +0.46°C/decade for the period 1973-2008, from the GISS archive at Gabès station) but it is important to remember that model projections are for climate changes due to observed and projected greenhouse gas concentration under the A1B scenario. Part of the stronger trends observed may result from

decadal variability which is not necessarily reproduced by the models.

More importantly, the CIRCE simulations show for the first time projections for the future situation based on multi-model outputs with the atmospheric component coupled to the Mediterranean Sea at a high resolution. There is a clear acceleration of the changes during the 21st

century most prominent for the air temperature. The CIRCE models project more severe conditions with a warmer climate and slightly reduced precipitation.

Marine conditions

What is it? Annual mean sea surface temperature (SST) averaged over the Gulf of Gabès area from the five CIRCE coupled models (and the ENEA-ERA40-2 model) are shown in Figure 3 for the period 1950-2050. In addition Figure 3 shows the SST from the two forced ocean models (the Mediterranean Sea model: INSTMED06, and the higher resolution regional model: INSTMCOTR) for the three simulated periods 1960-1969, 1991-2000 and 2050-2059. Figure 4 shows sea surface salinity (SSS) simulated by the coupled models and the average series for the period, 1950-2050.

Figure 3: Simulated sea surface temperatures in the Gulf of Gabès. The black line is the average series based on the different simulations (1950-2050).

Details of models and simulations are shown above.

Figure 4: Simulated sea surface salinity in the Gulf of Gabès, 1950-2050. The black line is the average series based on the different simulations (except

ENA-ERA40-2).

Table 2a shows the mean values and trends for SST and SSS for the periods 1961-1990 (present) and 2021-2050 (mid-century). It also shows the rate of change between these two periods (‘long-term’ changes). The inter-model ranges are shown as the standard deviations between estimates from the five models. Table

2b shows the rates of change for SST, SSS and SSH (sea surface height) between the three simulated periods by the INSTMED06 and INSTMCOTR forced models. For the INSTMCOTR model SSH represents the contribution from water circulation with reference to the central Mediterranean. For the

2. Climate and marine projections

INSTMED06 model SSH represents the effect of volume expansion and contraction due to the

variation in temperature and salinity (the steric effect).

Table 2a: Mean values and trends in SST and SSS from five CIRCE coupled models in the Gulf of Gabès, based on series averages for two periods, 1961-1990 (present) and 2021-2050 (mid-century). Mean changes and mean rates of change between the present and mid-century periods (long-term) are shown. Inter-model ranges are shown as standard deviations between estimates from the five models. Bold values are significant.

Coupled modelsVariable

present-climate(1961-1990)

mid-century(2021-2050)

long term(1961-1990 to 2021-2050)

SST(°C) (°C/decade)

17.9±1.1-0.01±0.070

19.1±0.9+0.42±0.11

+1.2±0.40

+0.2±0.070

SSS (PSU) (PSU/ decade)

37.45±0.47+ 0.03±0.050

37.29±0.51+ 0.05±0.091

-0.16±0.20

-0.03±0.030

1: Inter-model Mann-Kendall trend test: the null hypothesis of an absence of trend is rejected at the 0.05 significance level; 0: a failure to reject the null hypothesis. 1: Inter-model Student’s t-test: the null hypothesis means are equal is rejected at the 0.05 significance level; 0: a failure to reject the null hypothesis.

Table 2b: Mean rates of change for SST, SSS and SSH between the indicated periods from the forced models INSTMED06 (left values) and INSTMCOTR (right values). Bold values are significant.

INSTMED06/INSTMCOTRVariable

present-climate(1960-1969

to 1991-2000)

mid-century(1991-2000

to 2050-2059)

long-term(1960-1969

to 2050-2059)SST (°C/decade) +0.0130/+0.030 +0.251/+0.301 +0.171/+0.211

SSS (PSU/ decade) -0.081/-0.091 +0.0211/+0.0140 -0.0150/ -0.0231

SSH (cm/ decade) +0.50*/+1.01** +2.51*/-0.31** +1.81*/+0.10**

*SSH for the INSTMED06 model is relative to the steric effect in the Mediterranean basin; **SSH for the INSTMCOTR model is relative to the part related to the ocean circulation with a reference in the central Mediterranean. 1Student’s t-test: The null hypothesis means are equal is rejected at the 0.05 significance level; 0a failure to reject the null hypothesis.

What does this show? Projected water temperatures are lower than expected for the Gulf of Gabès where observations show an average value of 20.6°C. Nevertheless all models show an increase in SST with acceleration during the 21st century. SSS shows weak trends, although there is a tendency for a slight lowering of salinity at the long-term time-scale. For SSH the change is mainly due to the steric effect, +0.5 cm/decade between the periods 1960/1969 and 1991/2000, and +2.5 cm/decade between the periods 1991/2000 and 2050/2059. The observed sea-level trends are +2.6 cm/decade for the period 1999-2007 from data provided by the ‘Centre d’Hydrographie et d’Océanographie de la Marine Nationale, Ministère de la Défense Nationale’, Tunisia at Sfax; +2.1 cm/decade is observed by altimetry data in the Mediterranean basin for the period 1992-2005, Criado-Aldeanueva et al., 2008). Estimates of the steric component are also obtained from the coupled

models but at the Mediterranean basin scale (not shown). Relative to the period 1961-1990, the changes projected are +10cm for the period, 2021-2050 and +14cm for 2050. The steric component accounts for approximately 70% of total sea-level change (IPCC 2007); the projected sea-level rise is +14cm for the period 2021-2050 and +20cm for 2050. A similar estimate for 2050-2059 relative to 1960-1969 (INSTMED06 model, Table 2b) leads to +23cm.

Why is it relevant?Similar to the atmospheric variables, the oceanic ones show stronger trends after the year 2000. For the period 2021-2050 models project +0.54°C/decade and +0.42°C/decade for air and sea temperatures trends respectively. These close values again reflect how fast the shallow gulf responds to air warming. The projected water warming reinforces the expected changes in the gulf ecosystem including changes in biodiversity, fish resources with ‘knock-on’

2. Climate and marine projections

impacts on the fishing activity in the region. The very weak change in projected salinity suggests that salinity will probably not constitute an additional constraint on the gulf ecosystem.

However a sea-level rise of nearly +20 cm is projected in 2050 which could accentuate the vulnerability of the gulf to erosion and inundation.

3. Future impacts and vulnerabilities

Alien marine species

What is it? Harzallah et al., (2010) showed that the number of alien species is increasing in the Gulf of Gabès at a rate of 15% decade when corrected for bias due to the increasing number of studies (on alien species). This increase is attributed to changes in marine water characteristics (mainly increase in water temperature). The cumulated number of new observations corrected for the increasing number of studies is shown in Figure 5 together with an exponential fit. The reference date for the corrected series and exponential fit is the year 2007 with 40 new observations. With the hypothesis that alien species are related to the SST increase, a linear transformation is performed on the average SST projected by the different models shown in Figure 3 (thick line), so that the transformed SST equals the fitted exponential at two dates, 1966 and 2007 (first and last observation dates). The fitted SST gives the linear transformation needed to obtain an estimate of the new species from SST: Fitted SST=42.8*SST-741.6. The cumulated number of new species increases with SST at a rate of 42.8%/°C.

Figure 5: Estimated number of alien species in the Gulf of Gabès (red) obtained as a fitted exponential function to SST (Fitted SST=42.8*SST-741.6). An exponential fit (blue line) to observations with a trend of

15%/decade is shown. The observations (blue dots) are cumulative number of new species, (see Harzallah et al. 2010) and are adjusted to correct for the increasing number of studies with a reference the year 2007.

What does this show? Estimates of future behaviour in alien species can be estimated using numerical simulations based on the hypothesis that alien species observed in the Gulf of Gabès are related to sea warming. The projected increasing trend in alien species corresponds to 42.8% /°C.

Why is it relevant?In 2007 the total number of new species is estimated as 40. During the period 2007-2050 an increase in the sea-surface temperature of 1.6°C is projected (see Figure 3). Using the

above linear transformation the total number of new species is estimated as nearly 106 in 2050. Such a large number of new species would constitute an additional stress on the gulf ecosystem in particular for its biodiversity. It is therefore important to consider how the ecosystem would respond to such a large number of projected new species. However, it is unlikely that temperature warming is the only driver of this increase in invasive species. Their occurrence is more likely resulting from a combination of different mechanisms,

3. Future impacts and vulnerabilities

ecological, other environmental process and human activity.

Seasonal Shift Index

What is it? Using an index for the summer season extension, Harzallah et al. (2010) showed that there has been a lengthening of this season with an earlier start and a later end during the period 1960–2008. The devised Seasonal Shift Index, measures the start and the end of the summer season, occurring when the observed daily mean air temperature reaches 23°C for the first time. The index is similarly constructed using model data for the period 1950-2050, and is shown in Figure 6. The index is constructed for each model. An average index is then obtained. Table 3 shows the dates and trends calculated for the start, the end and the duration of the summer season for the ‘present’ 1961-1990 and ‘mid-century’ 2021-2050 periods. It also shows the corresponding ‘long-term’ mean changes and mean rates of change between these two periods. Prior to calculations, the average temperature series for each model for the period 1973-2003 is adjusted to the mean observed temperature at Djerba during the same period. This technique helps to correct for bias between model data and observations in the calculation of the start and end dates.

Figure 6: Seasonal Shift Index (T>23°C): start and end of the summer season estimated from six different CIRCE, models for the period 1950-2050. An average index based on these climate models (except ENEA-ERA40-2) is also shown. Days are counted from March 1st. Prior to calculations, the temperature series of each model (mean daily) is adjusted to the mean observed temperature at Djerba during the period 1973–

2003.

What does this show? The different models show a tendency for an earlier start and a later end to the summer season. Similar to observations, the summer season becomes longer. This lengthening is more substantial during the 21st century. In the ’present’ climate the summer season lengthens slightly (~ +0.4 days/decade) mainly due to a delay of its end (~ +1 day/decade) but the trends are not significant. In the ‘mid-century’ the summer season shows further lengthening (~ +4.9 days/decade) again mainly due to a delay in the end date (~ +3.9 days/decade). The projected

trends in the ‘mid-century’ are slightly weaker than for observed trends (~ +8 days/decade for the summer duration during the period 1960–2008, Harzallah, 2010).

Why is this important? Both observations and model simulations show a lengthening of the summer season mostly due to a delay in the end of the season. A few days added to the summer season constitutes a positive impact for the tourism activity. Moreover most added days are projected to occur in late summer which allows an extension of the August high season into September.

Table 3: Mean values and trends for the start, the end and the duration of the summer season based on the mean air temperature simulated by the CIRCE models for the Gulf of Gabès area shown in Figure 1. The average index is based on the different coupled models (except ENEA-ERA40-2). Trends are shown for two periods (present climate: 1961-1990; mid-century: 2021-2050). Mean changes and mean rates of change between the two periods 1961-1990 and 2021-2050 (long-term changes) are also shown. Inter-model ranges are shown as the standard deviations between estimates from the five models. Bold values are significant. Days are counted from March 1st.

Summer season Present climate (1961-1990)

start (days) (days/decade)

66±3+ 0.6± 1.60

end (days) (days/decade)

199+4+ 1±2.50

212±5+ 3.9±2.9

duration (days) (days/decade)

133±5+ 0.4±3.20

149±6+ 4.9±3.5

1: Inter-model Mann-Kendall trend test: the null hypothesis of an absence of trend is rejected at the 0.05 significance level; 0: a failure to reject the null hypothesis.1: Inter-model Student’s t-test: The null hypothesis means are equal is rejected at the 0.05 significance level; 0: a failure to reject the null hypothesis.

Daily Climate-Tourism Potential Index

What is it? The Daily Climate-Tourism Potential Index (DCTPI; Henia and Alouane, 2007) evaluates the impact of weather conditions on tourism activity (Harzallah et al. 2010). Five classes are defined as 0: ‘highly unfavourable’ days; 1: ‘unfavourable’ days; 2: ‘acceptable’ days; 3: ‘favourable’ days; 4: ‘highly favourable days’. The DCTPI is calculated for the period 1950-2050 using output from the five CIRCE coupled models, and annual and seasonal values for the index are calculated for the period 1961-1990 (present) and 2021-2050 (mid-century). For each variable used to calculate the DCTPI, the average of the different models is adjusted to that of observations for the period 1973-2003 at Djerba. This allows a model-based indicator to be constructed which is coherent with observations.

Figure 7: Histogram of the mean annual Daily Climate-Tourism Potential Index (DCTPI) for the Gulf of Gabès for the periods 1961-1990 (left) and 2021-2050 (right). DCTPI is the percentage of days corresponding to the

classes, 1: ‘Unfavourable’; 2: ‘Acceptable’; 3: ‘Favourable’; 4; ‘Highly favourable’. The DCTPI is the average estimate based on the five CIRCE coupled models. Individual model estimates are shown by coloured circles.

What does this show? For the ‘present climate’ (Figure 7; left), a majority of days (~62% annually) are ‘favourable’ for tourism and nearly 22% of days are ‘highly favourable’. Days ‘acceptable’ and ‘unfavourable’ are fewer (~15% and <1% respectively). For the ‘mid-century’ (Figure 7; right), models project more ‘acceptable’ days (~21%) and less ‘favourable’ days (~53%). Hence some ‘favourable’ days are replaced by ‘acceptable’ days (+6% between the two periods; nearly 22 days).

Trends are projected to be higher in the 21st

century. Table 4 shows that the trend of days degraded to ‘acceptable’ is +0.56% /decade for the ‘present’ climate whereas it is (+2.5% /decade) for the ‘mid-century’. This change is most prominent in summer when the percentage of ‘acceptable’ days increases to about +1.3%

/decade and +6.2% /decade respectively for the ‘present’ and ‘mid-century’ periods. In winter, there is an increase in the percentage of ‘acceptable’ and ‘favourable’ days in the ‘present’ climate but trends remain weak. For the ‘mid-century’, winter changes correspond to an increase of ‘highly favourable’ days (+4.1%). The 1950-2050 time series for the percent of ‘favourable’ and ‘acceptable’ days (Figure 8) clearly show the continuous increase in the number of ‘acceptable’ days and the continuous decrease in the number of ‘favorable’ days. Rates for the modelled ‘present’ climate are however smaller than those obtained for observations at Djerba (Harzallah et al., 2010).

Why is it relevant?The above results show that the CIRCE models reproduce the increase in the climate-related potential of the region in winter and its decrease in summer. The changes are only important during the 21st century when climate changes become significant. Agreement in the nature of changes between CIRCE model projections and

observations is found for the entire year but also for the winter and summer seasons separately. This indicates that the proposed changes are robust and that such model simulations constitute an important tool providing an indication of future climate-related potential for tourism activity.

Figure 8: DCTPI time series (1950-2050) for ‘favourable days’ (a) and ‘acceptable days’ (b) for the Gulf of Gabès, calculated using the output from five CIRCE coupled models, and the ensemble mean (black)

Table 4: Trends (upper: 1961-1990 present climate; middle: 2021-2050 ‘mid-century’) in the annual and seasonal (relative to number of days in the season) percentage of days corresponding to a given DCTPI class (in % days/decade) for the Gulf of Gabès. Inter-model ranges are shown as the standard deviations between estimates from the five models. Lower: mean rates of change between the two periods 1961-1990 and 2021-2050 (long-term). Bold values are significant.

Season DCTPI Class ‘present’ climate‘Acceptable’ ‘Favorable’ ‘Highly favorable’

Annual + 0.6±0.90 - 0.5±0.90 - 0.06±0.80

Winter + 0.2±0.80 + 0.8±1.60 - 1±1.90

Summer + 1.3±30 - 1.3±30 - 0.05±0.40

Season DCTPI Class ‘mid-century’‘Acceptable’ ‘Favorable’ ‘Highly favorable’

Annual + 2.5±0.91 - 2.7±0.40 + 0.1±1.10

Winter - 0.3±0.30 - 3.8±1.91 + 4.1±1.81Summer +6.2±2.31 - 6.0±2.21 - 0.5±0.51

1: Inter-model Mann-Kendall trend test: the null hypothesis of a trend absence is rejected at the 0.05 significance level; 0: a failure to reject the null hypothesis.

Season DCTPI Class ‘long-term’‘Acceptable’ ‘Favorable’ ‘Highly favorable’

Annual +1.2±0.61 -1.4±0.51 +0.1±0.10

Winter -0.2±0.11 -1.6±0.30 +1.8±0.30

Summer +3.2±1.51 -3.0±1.41 -0.2±0.11

1: Inter-model Student’s t-test: The null hypothesis means are equal is rejected at the 0.05 significance level; 0: a failure to reject the null hypothesis.

Socio-economic impacts on tourism activityWhat is it? This is a theoretical simulation of the socio-economic impacts resulting from days in a year becoming ‘acceptable’ (rather than ‘favourable’) in the Djerba-Zarzis tourist area (Harzallah et al. 2010). The hypothesis is that hotels are devalued when weather conditions become ‘acceptable’ such that prices equate to inland hotels (-20 Tunisian dinar /night devaluation; MEDD/PNUD, 2008). Changes are based on the annual percentage of days becoming ‘acceptable’ from the five CIRCE coupled models (Figure 8b). Ensemble mean annual changes are smoothed to remove interannual variability (Figure 9). Changes are assumed to start in 1960. Capital and gains losses are based on the calculations given in MEDD/PNUD (2008) and are relative to GDP (2007 rates). Employment is considered dependent on hotel capacity. Tourism statistics in the Djerba-Zarzis area are based on ONTT (2007).

Figure 9: A theoretical simulation of socio-economic impacts of days becoming “acceptable” in the Djerba-Zarzis tourist area for the period 1960-2050, based on the ensemble average of five CIRCE coupled models

(adjusted to that of observations for the period 1973-2006). Results are shown as annual percentages of 2007 GDP. Projections of the direct and indirect employee loss should be multiplied by 100,000.

What does this show? For climate model data, 11 days are degraded (relative to 1960) to ‘acceptable’ in 2007, a lower number than that based on observed weather conditions (51). Changes accelerate during the 21st century. In 2050, this number is projected to be 43. In 2007, losses of capital and of annual gains are both projected as 0.014% (of the 2007 GDP); in 2050 they are projected to increase to 0.023% and 0.11%, respectively. Such losses may induce a hypothetical reduction in the direct and indirect number of employed persons of around 5,000 and 14,000 respectively.

Why is it relevant?The projected acceleration of changes in economic and social impacts shown in Figure 9 reflects the similar changes simulated for weather conditions and the resulting changes for the DCTPI. The theoretical socio-economic impacts are projected to be more severe in the future. However, it is important to remember that in addition to the weather conditions simulated by the different models, the above results are largely dependent on the underlying hypotheses. Moreover other system drivers and pressures are not included.

4. Socio-economic trends

Climate change is only one of multiple factors acting on the Gulf of Gabès. In addition, climate and non-climate drivers interact in a complex way accentuating or diminishing future climate change impacts. Figure 10 shows some socio-economic indicators for Tunisia. Population growth in Tunisia (Figure 10a) is one of the lowest in the southern Mediterranean; lower than the word average. In addition, population growth is expected to decrease constantly and population to stop increasing around 2050. This suggests that there may be less severe social and economic pressures and hence less severe pressure on the marine environment (such as more controlled increase of fishing activity). Although population will still continue to grow in the short term, a steady increase in GDP

(Figure 10c) should provide increased financial resources for climate adaptation. However, due to the process of urbanisation, urban populations are expected to grow rapidly at the expense of rural populations (Figure 10b). A large influx of people in urban areas, together with an increase in life expectancy (Figure 10d) could create additional pressures for urban areas along the coast and may impact key sectors such as tourism and fishing. Such non-climate pressures will likely interact with climate change-induced pressures. It is expected that most social pressures will accentuate climate impacts. Investing in the less populated coastal areas may help to reduce these pressures. An integrated approach to climate vulnerability and impact assessment is therefore essential.

0

0.5

1

1.5

2

2.5

3

1950

-195

5

1960

-196

5

1970

-197

5

1980

-198

5

1990

-199

5

2000

-200

5

2010

-201

5

2020

-202

5

2030

-203

5

2040

-204

5

Five-year periods

Ann

ual %

cha

nge

in m

id-y

ear p

opul

atio

n in

th

e in

dica

ted

perio

d Tunisia World

a) Annual population growth rate. Source: UN World Population Prospect 2006

0

2000

4000

6000

8000

10000

1200019

50

1960

1970

1980

1990

2000

2010

2020

2030

2040

2050

Rural population(thousands)Urban population(thousands)

b) Urbanisation (urban and rural population). Source: UN World Urbanization Prospects: The 2007 Revision,

http://esa.un.org/unup.

0.00

2,000.00

4,000.00

6,000.00

8,000.00

10,000.00

12,000.00

14,000.00

2008 2009 2010 2011 2012 2013 2014 2015

PPP per capita GDP

c) Gross domestic product based on purchasing-power-parity (PPP) per capita GDP. Source: IMF World Economic Outlook Database, April 2010

0

10

20

30

40

50

60

70

80

90

1950

-195

5

1960

-196

5

1970

-197

5

1980

-198

5

1990

-199

5

2000

-200

5

2010

-201

5

2020

-202

5

2030

-203

5

2040

-204

5

year

s

Females Males

d) Life expectancy at birth. Source: UN World Population Prospects: The 2008 Revision,

http://esa.un.org/unpp

Figure 10: Selected socio-economic indicators for Tunisia

5. Uncertainties

The inter-model approach followed in CIRCE is based on a limited number of coupled models. The structural uncertainty inherent to this approach is therefore a major concern in the CIRCE simulations. The results shown for the Gulf of Gabès are mostly based only on five models. Nevertheless with this limited set the statistical significance of the means and the trends are tested. Ocean-only models (three ocean forced models) are also used to blend the results and allow a comparison of coupled and forced approaches.

The ability of the models to reproduce the observed climate is also a major concern in the CIRCE simulations. For example most CIRCE models simulate nearly 2°C cooler temperatures in the Gulf of Gabès. In addition the observed trends of several parameters are not well reproduced by the models (e.g., the simulated air temperatures trends in the Gulf of Gabès are lower than observed ones, see “climate conditions” above). However, the observed trends may be biased by large decadal and longer time-scale variability and hence do not reflect a true trend related to climate change. The bias found between observations and model data are reduced through model result adjustments. For example, simulated air temperatures are adjusted to observed ones during the period 1973-2003

when constructing the seasonal shift index (see Figure 6). Similarly, for each variable used to calculate the Daily Climate-Tourism Potential Index (DCTPI) at Djerba (Figure 7), the average of the different models is adjusted to that of observations for the period 1973-2003.

Additional uncertainties result from geographical differences between the location of observations (generally stations) and the model results (area averages). Such differences tend to be reduced by using monthly or annual averages so that local characteristics to the stations are reduced.

The approach used in the analysis of the model output may also introduce some uncertainties into the results. For example, socio-economic projections for tourism are based on a simple analytical model with crude simplifications (e.g., the 2007 annual GDP is considered, Figure 9). In addition only the direct effects (climate warming) on tourist attractiveness are considered. Other impacts of climate change (beach erosion and loss and water availability) need to be taken into account.

Finally all CIRCE models use the same choice of emissions scenario for the future (the SRES A1B scenario). Other scenarios (projecting more or less greenhouse emissions related to population and economical growth, policies, etc.) are not addressed in the CIRCE simulations.

6. Integrated assessment

Tables 5 and 6 provide a summary of climate trends and changes in the Gulf of Gabès based on the CIRCE model projections for the ‘present’ climate 1961-1990 and the ‘mid-century’ 2021-2050 periods. There is clear air warming for both mean values and daily extremes. Associated to this air warming, waters in the gulf are also warming. Their salinities are increasing but only at a very low rate. In addition, there is some evidence of salinity decrease at the long time-scale. Sea level is increasing mainly due to the steric effect in the Mediterranean basin. Precipitation is decreasing although at a low rate (~ -4% per decade in 2021-2050). Wind speed and relative humidity show no significant changes.

Those changes are expected to lead to biogeophysical and socioeconomic impacts in the Gulf of Gabès (see the linkages diagram, Figure.11, Harzallah et al, 2010). Alien species already recorded in the gulf are estimated to increase further in number, which may lower biodiversity. Similar to observations, a summer season lengthening is projected, which may

constitute a positive impact on the tourism activity in the southern gulf. Also similar to observations, changes in weather conditions show mitigating impacts, with a decrease of ‘favourable’ conditions in summer and an increase in winter with potential economic and social consequences. For most of the simulated variables and the indicators derived, the year 1990 seems to be the start of the period when changes become appreciable.

In general the changes in climate and the corresponding impacts deduced from the set of CIRCE climate models qualitatively agree with observations. The decadal variability modulating several climate variables (such as sea-surface temperature and sea-surface salinity) and the design of the CIRCE climate models initiated with a relatively short spin-up may explain part of the quantitative differences. Most of the simulated changes during the 21st century are significant. Hence the CIRCE climate models provide important information on projected changes in the first half of the 21st century relative to the second half of the 20th century, the base-line period. Observations demonstrate that changes are already occurring in the Gulf of Gabès; the CIRCE climate models project these changes increase in the future.

Table 5: Summary of observed (last few decades) and projected (2021-2050) impacts and vulnerabilities for Tunisia. ↑: observed/projected increase, ↓: observed/projected decrease, ↔: no trend/change identified. Brackets () indicate very small/uncertain trends/changes. -: not analysed.

Observed Projected (2021-2050)

SST (°C) (↑) ↑SSS (PSU) ↔ ↔SSH (steric, cm) (↑) ↑Estimated number of alien species ↑ ↑Season start ↓ ↓Season end (days) ↑ ↑Season duration (days) ↑ ↑‘Acceptable’ annual days (↑) ↑‘Acceptable’ summer days ↑ ↑‘Highly favourable’ winter days ↑ ↑Tourism: ‘Annual gain’ losses (↑) ↑Tourism: ‘Direct employee’ losses (↑) ↑

Table 6: Summary of trends and changes (long-term) in the Gulf of Gabès for the ‘present’ 1961-1990 and the ‘mid-century’ 2021-2050 periods based on output from five CIRCE climate models.

Climate Indicator (hazard)

present (1961-1990)

trends impact

mid-century long-term (2021-2050) (1961-1990 to 2021-2050)

trends changes impact T mean (°C/decade, °C) + 0.013 no significant

change + 0.54 +1.4 air warming

T max (°C/ decade,°C) + 0.003 no significant

change + 0.55 +1.4 day air warming

T min (°C/decade, °C) + 0.024 no significant

change + 0.55 +1.4 night air warming

Precipitation (mm/decade, mm) - 1.9 Small decrease

not significant -8.1 -16 Small decrease but not significant

Wind speed (m/s/ decade, m/s) + 0.01 no significant

change - 0.005 -0.05 not significant change

Relative humidity (% /decade, %) + 0.09 no significant

change - 0.1 -0.3 very weak decrease but not significant

SST (°C/decade, °C)

-0.01 +0.013* +0.03*

no significant change

+0.42 +0.25* +0.30*

+1.2+1.5*

+1.9*Significant water warming

SSS (PSU/decade, PSU)

+0.025 -0.08*

-0.09*

no significant change

+0.05 +0.02*

+0.014*

-0.16 -0.14*

-0.21*

weak increase but there is also evidence of decrease at

the long-term time-scale

SSH (steric, cm/decade, cm)Total (cm)

+0.5*

0

small sea level rise

base period

+2.5*

-

+16*

~ +20

significant sea-level rise

(very rough projection)Estimated number of alien species 40 in 2007 first records of

40 new species 106 in 2050 more than doubling projected

Season start (days/decade, days) +0.6 no significant

change -1.0 -3 earlier start

Season end(days/decade, days) +1.0 moderately later

end +4 +13. later end

Season duration (days/decade, days) +0.4 small summer

lengthening +5 +16. summer lengthening

‘Acceptable’ Annual(% days/decade, % days)

+0.53 slightly less (~2) favourable days/decade, especially in

summer: ~ -1 day/decade

+2.5 +7 less favourable days:(~ -4 to -9 days/decade)

especially in summer(~ +3 to +6 days/decade)

‘Acceptable’ Summer(% days/decade, % days)

+1.3 +6.2 +19

‘Highly favourable’ winter(% days/decade, % days)

-1.0

slightly less favourable days: (~ -1

day/decade)

+4.1 +11 more favourable days:(~ +2 to +4 days/decade)

Tourism: ‘Annual gain’ losses estimate (% 2007 GDP)

+0.014% in 2007(only

theoretical)

small losses +0.11% in 2050(only theoretical)

higher impacts

Tourism: ‘Direct employee’ losses estimate

700 in 2007(only

theoretical)small losses 5000 in 2050

(only theoretical) higher impacts

*

19

*Periods considered are 1960/1969 -1991/2000; 1991/2000-2050/2059 and 1960/1969-2050/2059

20

AcknowledgementsCIRCE (Climate Change and Impact Research: the Mediterranean Environment) is funded by the Commission of the European Union (Contract No 036961 GOCE) http://www.circeproject.eu/. This information sheet forms part of the CIRCE deliverable D11.5.7. The following data sources were used: relative sea level for the Sfax harbour provided by ‘Centre d’Hydrographie et d’Océanographie de la Marine Nationale, Ministère de la Défense Nationale’, Tunisia. The CIRCE climate model data were provided by INGV-CMCC (Gualdi Silvio, Enrico Scoccimarro); MF (Florence Sevault, Clotilde Dubois); IPSL-CNRS (Laurnet Li); ENEA (Alessandro Dell'Aquila, Adriana Carillo); MPIM-HH (Alberto Elizalde Arellano); INSTMED06 and INSTMCOTR (Ali Harzallah).

References► Criado-Aldeanueva F., Del Rio Vera J. and J. Garcia-Lafuente, 2008. Steric and mass-induced Mediterranean sea

level trends from 14 years of altimetry data. Global and Planetary Change, 60, 563-575.► Harzallah A. 2010. Information sheet on observed climate indicators for the coastal case studies : Gulf of Gabès,

Tunisia, M. Agnew and and C. Goodess editors, Climatic Research Unit, School of Environmental Sciences, University of East Anglia, Norwich, UK.

► Harzallah A., Bradai M.N., Ben Salem, S. and A. Hattour, 2010. Biogeophysical and social indicators: Coastal case studies information sheet: Gulf of Gabès, Tunisia, M. Agnew and and C. Goodess editors, Climatic Research Unit, School of Environmental Sciences, University of East Anglia, Norwich, UK

► Henia, L. and T. Alouane, 2007: Le potentiel climato-touristique de la Tunisie, Actes du X Xème Colloque de l’Association de Climatologie, Carthage, 3-8 Septembre 2007.

► Kendall, M. G., 1975. Rank Correlation Methods, Charles Griffin, London, UK.► Mann, H. B., 1945. Nonparametric tests against trend, Econometrica, 13, 245– 259.► MEDD/PNUD, 2008.: Etude de la vulnérabilité environnementale et Socio-économique du littoral Tunisien face à

une élévation accélérée du niveau de la mer due aux changements climatiques et identification d’une stratégie d’adaptation, Phase II, Ministère de l’Environnement et du Développement Durable. 127 pp.

► Nakićenović, N., and R. Swart (eds.) 2000. Special Report on Emissions Scenarios. A Special Report of Working Group III of the Intergovernmental Panel on Climate Change. Cambridge University Press, Cambridge, United Kingdom and New York, NY, USA, 599 pp.

Author: Ali Harzallah, Mohamed Nejmeddine Bradai and Scander Ben Salem INSTMContact: Ali Harzallah: ali.harzallah@instm.rnrt.tn.

Editors: Maureen Agnew (m.agnew@uea.ac.uk) and Clare Goodess (c.goodess@uea.ac.uk), Climatic Research Unit, School of Environmental Sciences, University of East Anglia, Norwich, UK

Date: June 2011

21