Global Changes , Regional Impacts : Climate Change in the Middle East

Abstract. In the Middle East, an area where pressure on water resources is intensified by political conflict and natural scarcity, the possibility of future climate change looms as yet another compounding factor. An integrated approach, taking economic, social, political and climate factors into consideration, is embodied in the CLIMSOC model. Before using global model data for a future period as input into the regional scale CLIMSOC model, the global climate model data must first be tested for the present period. The work summarised here examines monthly preeipitation data from a Hadley Centre Global Climate Model, comparing it to an observed climatology, for the present period 1961–1990. The differences between the GCM and observed data are examined with an eye toward systematic discrepancies among the different months, spatial patterns and overall quantitative differences in preeipitation. Finally, a glimpse at future preeipitation, as estimated by the global climate model, is presented in the context of the comparison results.


Introduction
In terms of necessity for sustaining life, water can be considered our most essential natural resource.Espe¬ cially in areas where it is scarce, its value increases to the point where conflicts arise over its shared use.The Middle East region presents an example of such a water access conflict, exacerbated by long term political an- tagonism (McCaffrey 1993).Yet another critical fac¬ tor to be considered in this precarious Situation is climate change, where the possibility of modified temperatures and precipitation could shrink an already lim¬ ited resource.catchment basin, presents itself as a fitting example of an area struggling with conflict over already limited water resources.Whether or not climate change affects water availability in the region, population growth will continually increase pressure on water supply.Jordan and Israel already withdraw more water from the Sys¬ tem than can be renewed.Distribution of scarce water requires a very sophisticated water management Sys¬ tem, as for example in Israel (Gleick 1993b).As in the case of the Jordan River watershed, this Situation can lead to considerable political tensions.Following are examples ofthe conflict Situation, based on Stephen C.   McCaffrey's «Water, politics and international law», in «Water in Crisis -A Guide to the World's Fresh Wa¬ ter Resources», edited by Peter H. Gleick (1993a).
Extensive references are provided there for further read- ing (McCaffrey 1993).
As it flows from its sources down to the Dead Sea, the 93 kilometers of the Jordan River contact several bor¬ ders: Lebanon, Syria, Jordan, Israel and the Palestinian Territories.The headwaters are made up of three streams, each located in a different country.Since 1967, Israel controls these areas and therefore all the headwa¬ ters ofthe Jordan.The Jordan River has two tributaries, one entirely in Jordan, the other forming the border be¬ tween Jordan and its neighbors, Syria and Israel.This second tributary contributes 40% of the Jordan's total flow.Overall, 77% of the Jordan River's water origi¬ nales in Arab countries (Fig. 1).
Water resources can clearly be considered a significant element contributing to hostilities in this region.One such case is the Situation leading up to the 1967 war, in which Israel seized the West Bank, declaring the water of the West Bank and the Gaza Strip a Strategie resource under military control.In 1964, partly in response to Is- rael's construction of the National Water Carrier (diverting water from Lake Tiberias to Tel Aviv and the Negev Desert), Arab states developed a plan called the Headwater Diversion Project.This would divert Jordan Riv¬ er headwaters and störe them in the planned Mukheiba Dam, which has since not been construeted.According to estimates, this would have cut the amount of water available for the National Water Carrier by half.
Israel considered the diversion to be a violation of its sovereign rights, and Iaunched military strikes against the works after construction began.They culminated in April 1967 with air strikes deep within Syria.
P. Beaumont stated during the 1991 Conference on Transboundary Water Disputes in the Middle East that «the increase in water-related Arab-Israeli hostility was a major factor leading to the 1967 June War» (Mc¬ Caffrey 1993: 93).
Indeed, the limestone aquifer located below the West Bank provides one quarter of Israel's water.The politi¬ cal significance of water is apparent in the oecupied ar- The Jordan River and its riparian territories Der Jordan-Fluss und angrenzende Territorien La riviere Jordan et les terriloires bordieres eas, where water use is strictly controlled.The inequitable nature of that control is expressed through the ab- sence of running water in many Arab villages (about half) alongside some Israeli Settlements that have swimming pools (McCaffrey 1993).However, neither Pal- estinian Claims of increasing Israeli withdrawals nor any opposing Claims from Israelis can be verified, as Is¬ rael deems water statistics to be State secrets and therefore does not release them (Gehriger 1999, Gleick 1993b).
Obviously, traditional animosities are rendered more acute over water scarcity.In spite of aecords signed by the conflict parties, recent developments show the potential role that climate change could play in the Middle East political arena: after declaring 1999 to be an offi¬ cial drought year, Israel announced in April that it would deliver only one half of the previously agreed- upon quantity of water to Jordan (Gehriger 1999,  Middle East Water Commission 1995).Obviously, any hope for agreement over equitable sharing of water resources hinges on the current Middle East Peace Con¬ ference; at the same time, an accord over water access is clearly a prerequisite for progress in these peace talks.The possibility of climatic changes renders this all the more urgent.Fig. 2: The GCM data points and corresponding grid cells of the study area Das GCM-Raster des Stildiengebietes La grille GCM de la region etudiee 2 Methodology and Data Planning for future water resources in this region obvi¬ ously demands examination of multiple factors.An in¬ tegrated approach, as proposed by the CLIMSOC mod¬ el, can provide this type of framework.GCMs, though coneeived to simulate climate primarily at the global scale, represent a powerful means of projeeting future climate conditions.The Utility of modeis as planning tools is clear, particularly in view of possible changes in climatic conditions and their subsequent impacts on various sectors of society.A necessary preliminary step to employing GCM pro- jeetions involves testing present period scenarios against Observation data.This comparison between GCM-produced preeipitation and preeipitation data from an observed climatology constitutes the core ofthe work described here.The two sets of preeipitation data were compared for the period 1961 -1990, the ti me scale being limited to monthly averages over the 30 year pe¬ riod.
The GCM data are the results of HADCM2, an acronym for Haehey Centre Coupled Model v2, indicating the model's origin: the Hadley Centre for Climate Prediction and Research, Meteorological Office, United King¬ dom (Johns et al. 1997, Mitchell & Johns 1997).
A gridded climatology of monthly means is used to rep¬ resent observed preeipitation.Developed at the Univer¬ sity of East Anglia's Climatic Research Unit, the 0.5°x 0.5°g rids are produced using thin plate spline interpolations of Station data as a funetion of latitude, longitude and elevation (New et al. 1999).
A basic methodological consideration in comparing the global data with the observed climatology concerns scale.Two scales are involved: the model data is grid¬ ded at 3.75°l ongitude by 2.5°latitude, whereas the ob¬ served climatology has a 0.5" x 0.5°g rid.A fundamen¬ tal question to be answered, therefore, is: which scale should be used to compare the two sets of values?In order to preserve the information of each dataset, the comparisons were carried out at both scales.More ex¬ tensive descriptions of methodology are provided in McNamara (1999).
To compare at the observed data's 0.5°scale, the GCM results were interpolated bi-linearly, after necessary Steps of coordinate conversion to ensure a more aecurate representation of distance (Bugayevsky & Schnyder 1995, Golden Software Inc. 1990).In order to present quantitative differences for each GCM grid point value, thus comparing at the GCM scale, the observed data were either averaged over the entire corresponding grid cell (as in Osborn & Hulme 1998), oran average of only the observed points located dosest to the GCM point (up to 4 observed points) was calculated.
3 Comparison of General Circulation Model Data with Observed Climatology The main differences between GCM projeetions and the observed climatology have been found as overestimales of the GCM in the northern part of the study area in both winter and summer, and as underestimates in winter along the northern and eastern Mediterranean coast.
The highest underestimates occur over areas that are smaller than one GCM grid cell, and are therefore better detected through comparisons at the finer resolution of the 0.5°scale (Fig. 3).This is linked to the observed preeipitation pattern of this region, where maximal amounts fall along a narrow strip of the coast, and then decrease sharply inland.Such high spatial variability is impossible to capture at the GCM's spatial scale.
Several region-wide statistics were also employed to compare the GCM data and the observed climatology: region-wide totals and averages, pattern correlation and root mean Square error.Though the GCM usually produced higher overall total and average preeipitation, the monthly variations are fairly well reproduced.Pattern correlations are higher in summer than winter, and root mean square error is lower in summer months than in the winter period.These last two comparison results are logically consistent -since rainfall is more frequent and more substantial in winter than in summer, it is more difficult to produce aecurate winter projeetions.
It is clear that variations at the subgrid scale cannot be reproduced by the GCM: the scale ofthe model hinders its ability to project preeipitation as it oecurs at the re¬ gional scale.Two features significant to rainfall patterns in this region are not «seen» by the global model: the rain shadow in central Turkey, and the steep rain gradient inland from the coast.Observed maximum amounts fall over limited areas and vary considerably at the sub¬ grid scale.The GCM value, when taken as an average over each grid box (more appropriately), masks the var- iance within each grid.
4 Exploring Data Relationships In addition to the previous descriptive indicators, further analyses were carried out to examine the spatial distribution of January preeipitation in the two datasets, particularly in terms of its association with other fac¬ tors.The three factors considered are spatial continuity, topography and continentality.
Spatial continuity (autocorrelation), a characteristic of most earth science datasets, is expressed through data values that are grouped together rather than being randomly located in space.The distance and orientation of data grouping are revealed using variography analyses (Isaaks & Srivastava I989, Bonham-Carter I994, Pannatier 1996).It was found that January preeipita¬ tion displays autocorrelation up to a certain distance in both GCM and observed data, but that observed data show more variability over shorter distances.Variogra¬ phy further illuslrated that the orientation of data clustering is not always the same in observed and GCM data, depending on the sector considered.
Using correlation coefficient r, preeipitation distribu¬ tion was then examined for associations with topogra¬ phy and continentality.Correlations of rainfall with to¬ pography uncovered differences between GCM and observed preeipitation, with GCM rainfall showing in general a stronger association with elevation, which is consistent with assumptions regarding precipitation's determinant factors.The r values for observed data, however.were largely influenced by extreme points of higher observed preeipitation which occur at lower ele- vations; these extreme points cannot be reproduced at the GCM scale.Correlations between preeipitation and distance from the Mediterranean showed rainfall having a stronger association with continentality than with to¬ pography.Continentality displayed negative correla¬ tions up to certain distances.This distance was greater for the GCM data, which also had lower r values than observed preeipitation.

5
Projected Climate Change: Implications for the Future Having located probable areas of the GCM results requiring adjustment, it is interesting then to look at the HADCM2 projeetions for the period 2010-2039.These are shown in Fig. 4, presented as change with respect to the period 1961-1990.Acknowledging that it is prefer- able to consider projeetions as a ränge of possible changes stemming from several modeis, the ränge ofthe Hadley Centre model's results agree generally with those presented by Wigley (1992).
It is interesting to note that, based on the comparisons carried out here, model error is often quantitatively much larger than projected change.Over the entire study area and all months, the model projeetions ränge from -0.59 to +0.41 mm/day (corresponding to -21 to +23% of averaged-observed preeipitation at those points).The ränge of model error (GCM versus aver¬ aged-observed) is considerably greater: from -2.18 to +3.04 mm/day (corresponding to -65 to +203% of aver¬ aged-observed at those points).When presented as a percentage of observed rainfall, the projected changes alone are already considerable, and could require serious adaptations in water resources planning.Once mod¬ el errors are «corrected», the resulting magnitude of projected change may well represent an urgent need for action in water management.
Taking into account the inherent uncertainty of model results, plus the corrections necessary to «downscale» global scale projeetions, substantial changes could be possible, and could be severe.Yet, in this area where resources are already strained beyond the limits of the renewable System, any change in the physical availability of water could present significant problems.This could seriously affect access to available water which, as mentioned earlier, is already a very delicate issue.
6 Concluding Remarks In conclusion, it is important to remember that the point here is not simply to highlight the shortcomings of glo¬ bal modeis.If any criticism were to be expressed, it would need to be directed at improper applications of GCM results, a problem which is related to an appreciation of the specificity of scale.Global general circulation modeis produce results at the global scale, and as such these results represent a useful tool for assessing conditions over the entire globe.Preeipitation, though of course related to synoptic level circulation, is a phe¬ nomenon with many regional and local influences.Before employing GCM results, for example in a regionalscale model of water resource use such as CLIMSOC, it is necessary first to recognize local and regional influ¬ ences and then to integrate them, in an appropriate man¬ ner, into the global setting.
Summary: Global Changes, Regional Impacts: Climate Change in the Middle East In the Middle East, an area where pressure on water re¬ sources is intensified by political conflict and natural scarcity, the possibility of future climate change looms as yet another compounding factor.An integrated ap¬ proach, taking economic, social, political and climate factors into consideration, is embodied in the CLIMSOC model.Before using global model data for a future period as input into the regional scale CLIMSOC model, the global climate model data must first be tested for the present period.The work summarised here examines monthly preeipitation data from a Hadley Centre Global Climate Model, comparing it to an ob¬ served climatology, for the present period 1961-1990.The differences between the GCM and observed data are examined with an eye toward systematic discrepancies among the different months, spatial patterns and overall quantitative differences in preeipitation.Final¬ ly, a glimpse at future preeipitation, as estimated by the global climate model, is presented in the context of the comparison results.