Overlay mapping trend analysis technique and its application in Euphrates Basin, Turkey

Authors


ABSTRACT

New investigation techniques for detecting hydro-meteorological data trends are required due to significant changes in precipitation and streamflow affecting the planning and operation of water resources. In this study, the trends analysis is carried out for the observation of climatic and hydrologic parameters such as temperature, humidity, precipitation and streamflow in the Euphrates Basin located in the Southeastern Anatolia Project (referred to as GAP from the Turkish acronym for Güneydoǧu Anadolu Projesi) region, which is the most important integrated water resources project in Turkey. The relationships between different meteorological and hydrological parameters in the basin are evaluated using the overlay mapping technique, popular in various engineering disciplines. The significant trends are indicated on the thematic maps and the visible changes are observed in different parts of the basin. Decreasing trends are detected in minimum streamflow, while the increasing trends in the annual mean and maximum temperature and in the annual mean and maximum humidity parameters are observed for many stations. The study concludes that the overlay mapping technique can be used effectively to observe the climatic changes and trends in hydrometeorological variables. The results are expected to help water resources managers and policy makers in decision-making for better planning in water resources planning and management.

1. Introduction

It is well known that there are significant changes in precipitation and streamflow, which may affect water resource planning and operation. Recent climate change studies have shown significant impacts on hydrological cycle components such as streamflow, precipitation, evapotranspiration and temperature (Arnell, 1999; Lettenmaier et al., 1999; Payne et al., 2004). The increasing impact of climate change on flood occurrences, landslides, droughts, reduction in agricultural products, pollution, hydro-electric energy shortage and water scarcity reinforce the need to determine the relationship between climatic changes and hydrological system performances. The adjustment to climate change is possible through climatic parameters on which there are effects by human activities and, therefore, future climatic change predictions play a significant role in any water resources design. Climatic change impacts could be explained by means of hydrometeorological data, decreases in glacier extent, sea level rise, water temperature increments in lakes, and aerosol concentrations.

Recent research in various parts of the world has shown major changes in rainfall and river flows. However, there is insufficient information today about how climatic change would affect the rainfall in any region. The annual variability of the relationship between rainfall and runoff in different seasons shows a great difference compared to the previous years. As a result, water management has a great significance in meeting the water demand not only temporally and spatially but also qualitatively and quantitatively. Flood values are also required in operation, design and management in controlling structures such as spillways (Bayazıt et al., 2002).

Observational and historical streamflow data are the most important information sources in planning and designing water resources projects. These data show time dependent characteristics and are affected by many factors such as anthropogenic activities and their consequent climate change impacts. One of the main steps in water resources works is to identify possible trends in observed streamflow data with their temporal and spatial occurrences (Yenigun et al., 2008).

Climatic change and its effects on water resources are investigated extensively by various researchers. For example, Burn and Elnur (2002) described the development and application of a procedure that identifies the trends in hydrological variables. Their study is focused on the quantification of trends in hydrometeorological variables and the investigation of the relationship between trends in hydrological variables in Canada. Whitfield and Cannon (2000) compared the hydrometeorological data for Canada from two different decades and found the recent decade to be generally warmer with the occurrence of both increases and decreases in precipitation and streamflow. Birsan et al. (2005) analysed the mean daily streamflow records from 48 watersheds in Switzerland with an undisturbed runoff regime for the trends using the Mann–Kendall non-parametric test for three study periods. Their results pointed out that the mountainous basins are the most vulnerable environments from the climate change point of view, because their watershed properties promote fast runoff with their fundamental vulnerability to temperature changes, which affect rainfall, snowfall, and snow and ice melt. Christensen et al. (2004) assessed the potential effects of climate change on the hydrology and water resources of the Colorado River basin by comparing simulated hydrological and water resources scenarios. They used climate simulations of the U.S. Department of Energy/National Center for Atmospheric Research Parallel Climate Model (PCM) to obtain scenarios driven by observed historical (1950–1999) climate. Vicuna and Dracup (2007) reviewed the climate change impacts on hydrology and water resources in California by three major fields in climate change: (1) studies of historical trends of streamflow and snowpack in the geophysical record; (2) predicted potential future effects of climate change on streamflow, and, (3) predicted changes in natural runoff in order to determine their economic, ecological or institutional impacts. Miller et al. (2003) investigated the potential impacts of climate change in California by using two general circulation models. According to them, it appears that all snowmelt from the runoff basins and snow accumulation late in winter decreases by 50% toward the end of this century. Wit et al. (2007) analysed the observed precipitation, temperature and discharge records from the Meuse basin (North Europe) for the period 1911–2003. They also simulated the impact of climate change on low flows.

Some other examples related to climatic change impacts on water resources are available. Gan (1998) applied Kendall's test to temperature, precipitation, evapotranspiration and natural streamflow data in Canada. Langan et al. (2001) studied the variation in river water temperatures in an upland stream over a 30 year period for Girnock in Scotland. Hiscock et al. (2001) studied the flow records of the Rivers Bure, Nar and Wensum in eastern England with the aim of identifying long-term changes in flow behaviour relating to variations in rainfall amount, land use, land drainage intensity and water resources use. Morrison et al. (2002) attempted to determine the trends in the historic flows and water temperatures of the Fraser River system in Canada considering both the annual flow profiles and the summer temperatures provided that these trends are likely to continue under the conditions predicted by various global circulation models (GCMs). Statistical relationships were established between three indices of atmospheric circulation, daily catchment precipitation and potential evapotranspiration in order to downscale from the GCM to the Upper Wye experimental catchment in mid-Wales, UK by Pilling and Jones (2002). Legesse et al. (2003) studied hydrological modelling at a catchment scale to investigate the impact of climatic and land use changes on water resources in Southern Ethiopia where the data are scarce, using a distributed precipitation-runoff modelling system. Fu et al. (2004) used Kendall's test to analyse the hydro-climatic trends of the Yellow River over the last half century in China. The potential effects of climate change on the hydrology and water resources of the Columbia River Basin were evaluated using simulations from the U.S. Department of Energy and National Center for Atmospheric Research Parallel Climate Model by Payne et al. (2004). Ye and Glantz (2005) reviewed the changes in the short-term climate information for water management in China after the 1998 Great Flood in the Yangtze River basin. They found that the assessments of the 1998 Great Flood in the Yangtze River basin helped the central government and water resources agencies to recognize the weaknesses of the existing flood control system, the mismanagement in the ecological systems, and the need for developing a national water resource management plan to deal with the problems of too much water, too little water, and very polluted water situations. Nawaz and Adeloye (2006) carried out the Monte Carlo Simulations technique to characterize the sampling uncertainties in assessed water resources impacts in Yorkshire, England. Kim et al. (2007) investigated and evaluated the climate change impact on the runoff and water resources by considering double carbon dioxide increase of Yongdam basin, Korea.

Yılmaz (1999) studied the climatic trend detection and its effects in the East Black Sea Basin in Turkey and found that the average temperature of the basin tends to decrease as quantities of the rainfall decrease along the coastal region of the basin and it tends to increase in higher parts of the basin. Kosif (1999) analysed six different climate elements (mean temperature, total precipitation, average flow, total evaporation, average duration of sunshine and clouds). Significant decreases in the mean temperatures, total evaporation and mean sunshine duration and increases in total rainfall and in mean flows, are observed with regional changes in cloudiness data of the basin.

The ‘Overlay Mapping Technique’ (OMT) is used in wide application areas (Bailey, 1988; Sumathi et al., 2008). This approach includes various features including geology, topography and soil, climatic and other related conditions of the region and makes joint comparison through map transparencies possible (Yesilnacar and Cetin, 2008).

To the best of the authors' knowledge, no study has yet been reported on the determination of climatic change effects on water resources using OMT. This provides an impetus to investigate the potential of the OMT for better understanding the process. The main aim of this study is to develop a suitable OMT model for climate change impacts on water resources with the applications in the study area, the Euphrates basin in Turkey. The usefulness of the method lies in its ability to provide a superposition process between the trend maps of climatic and hydrological parameters.

2. Materials and methods

2.1. Study area

The study area is the Euphrates Basin, located in the Southeastern Anatolia region of Turkey. It is the largest basin in the country and has the highest mean annual streamflow in Turkey. It is part of the Southeastern Anatolia Project, which is a multi-sector and integrated regional development iniative in sustainable development. The Basin's mean annual average discharge downstream (at the Syrian border) is 950 m3 s−1 (Yenigun et al., 2008). This region of the Euphrates basin has a continental sub-tropical climate, with extremes of heat in summer and cold in winter, as well as high diurnal variations. The total area of the Euphrates Basin in Turkey is 120 917 km2 (the overall area of the Euphrates basin is 765 831 km2) and its average elevation is about 1000 m. The general characteristics of the Euphrates Basin (as an average of all stations that are located on the main river bed) are given in Table 1 (EİEİ, 2003). The analysis does not cover the whole Euphrates River Basin due to the lack of hydrometeorological data at downstream of Turkey. The study area lies from 36°46′58″ to 43°55′50″E and 36°37′40″ to 40°27′50″N. Details of the Euphrates Basin are shown in Figure 1 with the locations of the selected hydrometric and meteorological stations (EİEİ, 2003; DSİ, 2009).

Figure 1.

Study area (a) and some details of the basin: (b) important irrigation areas, (c) the cities, (d) the rivers, (e) the selected hydrometric and meteorological stations

Table 1. General characteristics of Euphrates Basin
CharacteristicValue
Mean annual streamflow (m3 s−1)995.08
Volume of annual streamflow (m3)31.38 × 109
Annual runoff (mm)259.52
Mean annual efficiency (l s−1 km2)8.23
Ratio of discharge in the Euphrates Basin to total discharge in Turkey (%)16.75

2.2. Data

The streamflow gauging stations are operated by the General Directorate of Electrical Power Resources Survey and Development Administration (Turkish initials ‘EİEİ’) and the General Directorate of State Hydraulic Works (Turkish initials ‘DSİ’). The meteorological stations are operated by the Turkish State Meteorological Service (Turkish initials ‘DMİ’). In this study, 22 streamflow stations and 39 meteorological stations are employed. The selection of these stations is based on the record length, reliability and continuity of the data. The long term recorded daily meteorological data between 1960 and 2000 are used in the analysis. The measurement devices and measurement techniques are reliable, so the probable errors in the measurements are not taken into account in the analysis. The analysis is based on the data given by DSI and DMI measurements. The rivers with streamflow gauging stations are not regulated by large storage reservoirs. Not affected by urbanization they have relatively natural flows. Therefore, in the analysis, it is considered that there are no potential anthropogenic effects on the catchment. The details of the selected streamflow and meteorological stations are given in Tables 2 and 3, respectively (DMİ, 2009; DSİ, 2009). Since the results of the trend tests depend strongly on the chosen period, the same period is used for every parameter in the analysis. Detailed raster and vector maps are obtained from General Command of Mapping (Turkish initials ‘HGK’). These maps are developed by the authors using Arcmap.

Table 2. The details of selected streamflow gauging stations
Station identification number (SID)Part of basinName of stationLongitude (° ' ”E)Latitude (° ' ”N)Altitude (m)
2102MiddleMurat Nehri-Palu39 55 5238 41 18852
2115LowerGöksu Nehri-Malpınar38 09 3237 29 22390
2119UpperFırat Nehri-Kemah Boǧazı39 23 3739 41 021123
2122MiddleMurat Nehri-Tutak42 46 4939 32 191552
2123LowerÇaǧçaǧ Suyu-Çınarköy41 18 1437 11 38560
2124LowerTohma Suyu-Yazıköy37 26 3538 40 23995
2131LowerBeyderesi-Kılayık38 12 3838 19 47892
2132LowerCulpsuyu-İncirli39 02 0237 09 37470
2135LowerBulam Çayı-Fatopaşa38 14 1337 59 381252
2141UpperPersisuyu-Korudibi40 06 2839 09 131035
2145LowerTohma Suyu-Hisarcık37 41 0838 28 34933
2149UpperMunzur Suyu-Miskidaǧ39 32 3339 06 37900
2151UpperFırat Nehri-Demirkapı40 10 0539 34 451355
2154UpperKarasu-Aşaǧı Kaǧdariç40 45 3539 56 201675
2156UpperFırat Nehri-Baǧıştaş38 27 0439 26 05865
2157MiddleKarasu-Karaköprü41 29 4338 47 021250
2158MiddleBingöl Çayı-A.Paşa Köp.41 29 1439 06 301310
2164MiddleGöynük Çayı-Çayaǧzı40 33 1738 48 31990
2165LowerZerkan Suyu-Hocaköy40 30 3037 08 32445
2166UpperPeri Suyu-Loǧmar39 48 5238 51 30845
2167UpperÇaltı Suyu-Dazlak38 15 3339 20 52890
2168UpperDumlu Suyu-Yeşildere41 24 3640 08 172000
Table 3. The details of the selected meteorological stations (DMİ, 2009)
Station numberStation nameLocationStation details
 Latitude (°')NLongitude (°'E)Altitude (m)
17094ErzincanErzincan39 4439 301218
17096ErzurumErzurum39 5741 101758
17099AǧriAǧri39 4343 031632
17165TunceliTunceli39 0639 33981
17199MalatyaMalatya38 2138 18948
17201ElaziǧElaziǧ38 4039 13990
17203BingölBingöl38 5340 301177
17204MuşMuş38 4441 291323
17261GaziantepGaziantep37 0437 23854
17262KilisKilis36 4337 05638
17265AdiyamanAdiyaman37 4638 16672
17270ŞanliurfaŞanliurfa37 0938 47547
17275MardinMardin37 1840 461050
17718TercanErzincan39 4640 241425
17734DivriǧiSivas39 2238 071120
17736MazgirtTunceli39 0139 361400
17740HinisErzurum39 2141 421715
17762KangalSivas39 1437 231541
17764ArapkirMalatya39 0238 291200
17766AǧinElaziǧ38 5638 43900
17768Çemiş gezekTunceli39 0338 56953
17774KarakoçanElaziǧ38 5740 031090
17776SolhanBingöl38 5841 031366
17780MalazgirtMuş39 0942 331565
17804KebanElaziǧ38 4838 43808
17806PaluElaziǧ38 4239 571000
17808GençBingöl38 4540 361250
17842BalabanMalatya38 2837 351123
17843BaskilElaziǧ38 3438 491300
17844SivriceElaziǧ38 2639 181240
17872DoǧanşehirMalatya38 0537 531280
17874ÇermikDiyarbakir38 0839 28700
17910KahtaAdiyaman37 4638 38675
17912SiverekŞanliurfa37 4539 19801
17948NusaybinMardin37 0541 13500
17966BirecikŞanliurfa37 0137 57345
17968CeylanpinarŞanliurfa36 5040 02360
17980AkçakaleŞanliurfa36 4338 57361

2.3. Trend detection tests

In statistical terms, the trend analysis is used to understand whether a series of observations has generally increasing or decreasing behaviour, or the probability distribution function (pdf) changes with time or space. Several tests are available for the detection and/or quantification of trends such as non-parametric, mixed and parametric (Helsel and Hirsch, 1992). Non-parametric tests are widely used in trend analysis of climatic and hydrological data, which are robust with respect to missing and tied values, seasonality, non-normality, non-linearity and serial dependency. In this study, non-parametric Mann–Kendall and Spearman's Rho tests are used for trend detection in the streamflow of the Euphrates Basin. The Mann–Kendall rank correlation test is used for determining trend beginning year. The linear slopes of trends are calculated by using Sen's estimator of slope technique.

The Mann–Kendall method (Mann, 1945; Kendall, 1975) is one of the widely used non-parametric tests for detecting trends in climatological and hydrological time series. It has been used by the World Meteorological Organization (WMO) to assess the trend in environmental data time series (WMO, 1988). The method is simple, robust and handles missing and below detection limit values (Hamed and Rao, 1998; Burn and Elnur, 2002; Xu et al., 2003; Kahya and Kalayci, 2004; Silva, 2004). The effect of serial correlation problems are taken into account in the Mann–Kendall test. Choosing the level of significance is an arbitrary task. Therefore, the 0.05 significance level, which is chosen traditionally for many applications, is also used in the analysis. The most common approach for removing the serial correlation from a data set is pre-whitening (Burn and Elnur, 2002; Onoz and Bayazıt, 2003; Yue et al., 2003; Bayazıt and Onoz, 2004; Burn et al., 2004a, 2004b). The pre-whitening approach involves calculating the serial correlation and removing the correlation if the calculated serial correlation is significant at the 5% level. In this study, the pre-whitening procedure is applied to the data set before executing the tests.

The rank-based non-parametric statistical Spearman's Rho test can also be used to detect a monotonic trend in a time series (Yue et al., 2002). Spearman's rank correlations test is a quick and simple test to determine whether any significant correlation may exist between two classifications of the same series. In this test, a significant trend occurs only if the correlation between time steps and streamflow observations are found to be significant.

The Mann–Kendall rank correlation test gives the start point in time of a developed trend. This test does not take differences in magnitude of the values into account, it only counts the number of consecutive values where the value increases or decreases compared with the value before.

If a linear trend is present, the true slope (change per unit time) can be estimated by using a simple non-parametric procedure (Sen's estimator of slope) developed by Sen (1968). Details of the tests are given by Yenigun et al. (2008).

The binomial probability distribution function is employed in this study, which has been previously used by some researchers (Livezey and Chen, 1983; Yue et al., 2002) for assessing the significance of the trend. The probability of showing or exceeding a downward trend at five sites by chance, at a significance level of 0.05, is 0.8% according to the binominal probability distribution function. Thus, it is assumed that the detected trends in the region may not be due to chance alone. So, the cross correlation does not have influence on the test results.

In this study, a computer program called TAFW (Trend Analysis For Windows) which contains the above statistical trend tests is used for analysing the trend in streamflow (Gumus, 2006). This is a visual program developed by Gumus and Yenigun using Borland Delphi (Yenigun et al., 2008).

2.4. Overlay Mapping Technique (OMT)

This technique is an approach that includes various features of the study region (geology, topography and soil, climatic conditions) and as mentioned earlier makes the joint comparison possible through map transparencies (Yesilnacar and Cetin, 2008). Another definition of overlay is an analysis procedure for determining the spatial coincidence of geographic features. The overlay function output is capable of creating composite maps by combining diverse data sets. These outputs can reflect simple operations such as laying a road map over a map of local wetlands, or more sophisticated operations such as multiplying and adding map attributes of different values to determine averages and co-occurrences. At its simplest, this could be a visual operation, but analytical operations require one or more data layers to be joined physically.

In this study the following steps are performed in sequence.

  1. Trend features are determined for annual mean, minimum and maximum flows and current trends in the flows, including directions of the changes and the starting year. To ensure practicality, the detected trends in maximum/minimum/mean flows have been the first step for this study (Yenigün et al., 2008).

  2. The climatic change parameters (temperature, precipitation and relative humidity) which are considered to be effective in climate are used in the Mann–Kendall test for investigation of flow trends. The trend tests, described by Yenigun et al. (2008), are used to estimate trends in the hydrometeorological data.

  3. The Overlay Mapping Technique is employed in order to examine the changes in hydrological and meteorological parameters and the trend results are posted to the basin map. GIS (Geographical Information Systems) are used for the purpose of thematic map preparations. Rivers, cities, borders and all trend areas are taken as general parameters. Results from all stations and their trend analysis outputs are transferred spatially to these maps in digital form by using ARC-GIS software. Here, the specific criteria for ARC-GIS are the significant trends detected for stations. The trend areas in the maps as the results of trend analysis are generated by using Arcmap interface.

  4. Finally, the changes of climate and water resources are observed by the trend relationships using the superposition of different trends on thematic maps. In this way, past situations are evaluated and, accordingly, the climatic change and its visual relationship to water resources are interpreted.

3. Results and discussion

The results of the Mann–Kendall and Spearman's Rho test applications to annual minimum, maximum and mean streamflow of the selected stations are summarized in Table 4. The significant trends are more common in minimum streamflow than annual mean streamflow (see Yenigun et al. (2008) for detailed information).

Table 4. The results of Mann–Kendall and Spearman's Rho test for streamflow gauging stations
SIDPart of basinP-value of Mann–Kendall testP-value of Spearman's Rho test
  Mean streamflowMinimum streamflowMaximum streamflowMean streamflowMinimum streamflowMaximum streamflow
  1. equation image: upward trend, equation image: downward trend; bold values.

2102Middle0.41650.0230equation image0.45480.40520.0250equation image0.4562
2115Lower0.37900.07910.08920.37450.12920.0708
2119Upper0.19710.34920.08920.24830.39360.3336
2122Middle0.21050.0074equation image0.40600.24200.0071equation image0.3745
2123Lower0.26350.0094equation image0.11190.16850.0089equation image0.1335
2124Lower0.26350.19760.08480.25780.16850.0708
2131Lower0.03270.0085equation image0.04910.04750.0069equation image0.0537
2132Lower0.0099equation image0.32000.48370.0104equation image0.33720.4286
2135Lower0.25680.13290.06160.04460.04460.0694
2141Upper0.41740.19630.32320.36320.20050.2709
2145Lower0.14470.03870.05980.10380.03440.0537
2149Upper0.24770.50000.48270.27090.46810.4880
2151Upper0.33640.21080.11800.29810.20900.1190
2154Upper0.44590.14600.35420.42470.10750.3632
2156Upper0.04790.0136equation image0.41920.04850.0122equation image0.3409
2157Middle0.44590.11050.29330.45220.04460.2546
2158Middle0.38520.25920.07670.46020.21190.0694
2164Middle0.45480.07670.18840.39360.05050.2061
2165Lower0.24890.43560.16940.22660.30850.1922
2166Upper0.29690.48580.34730.28430.43250.3121
2167Upper0.28060.41630.24610.26760.46810.2206
2168Upper0.27970.0092equation image0.14310.28430.0078equation image0.1401

Significant decreasing trends are also detected in the annual daily minimum streamflow of stations 2102, 2122, 2123, 2131 and 2168 (5 of the 22 stations), which may indicate the presence of dry periods within a year in these rivers. Only station 2156 shows an upward trend in the minimum streamflow. For annual daily maximum streamflow, none of the stations has a significant trend. The decrease in low flows is especially important for the location of the water treatment facility, the quantity of irrigation, and water supply. The changes in low-flow statistics also affect the minimum water quantity released by the dams downstream for sustainable protection of ecological cycles.

The start years of the observed trends are given in Table 5. The range of the beginning detected trend years is between 1971 and 1994. There are 35 dams used for irrigation and energy production in the Euphrates Basin, with the largest ones being the Keban, Karakaya and Ataturk dams (DSİ, 2007). Although the rivers in this study have a relatively natural flow, it is interesting that the starting years for operation of those dams (1975, 1987 and 1992) are in the period of the start years of the observed trends (1971–1994). This may be attributed to other factors such as climatologic variables including precipitation and temperature. So, the presence of trends in Turkish streamflow patterns are generally attributed to the decreases in rainfall (Partal and Kahya, 2006; Kahya and Kalayci, 2004; Cigizoglu et al., 2005).

Table 5. The beginning year of observed trends for stream flow gauging stations according to Mann–Kendall rank correlation test and the values of Sen's slope
SIDPart of basinThe beginning yearSen's slope (m3 s−1 per year)
  Mean streamflowMinimum streamflowMaximum streamflowMean streamflowMinimum streamflowMaximum streamflow
  1. The beginning years belonging to the bold values.

2102Middle1982− 1.05920.5623− 9.3611
2115Lower− 0.4501− 0.2707− 7.2287
2119Upper0.1928− 0.04611.9500
2122Middle1985− 0.08920.0922− 3.8750
2123Lower1983− 0.06070.06410.2366
2124Lower− 0.0135− 0.0271− 0.5400
2131Lower1972− 0.01420.0073− 0.3046
2132Lower1972− 0.01470.0000− 0.0295
2135Lower− 0.0093− 0.0126− 0.3400
2141Upper0.0000− 0.11715.2166
2145Lower− 0.1461− 0.1363− 1.2166
2149Upper0.0350− 0.06530.4000
2151Upper− 0.1961− 0.0943− 2.8249
2154Upper− 0.0489− 0.0155− 0.0801
2156Upper19940.76110.4294− 0.8333
2157Middle− 0.02080.00790.2798
2158Middle− 0.0133− 0.00802.0000
2164Middle0.0348− 0.0503− 4.0277
2165Lower− 0.0091− 0.0017− 0.7333
2166Upper0.33000.0100− 0.9285
2167Upper− 0.2530− 0.0245− 4.5634
2168Upper1990− 0.00070.0028− 0.0768

The Sen's slopes of all trends in streamflow are also presented in Table 5. The statistically significant slopes, which are founded by ArcGIS, are highlighted with bold font. The sign of the slopes are consistent with the Mann–Kendall and Spearman's Rho test results (Table 4). In Figure 2, the detected trends in annual minimum and mean streamflow are shown on the Euphrates Basin map, respectively. Downward trends are observed in both annual mean and minimum streamflow for the stations at the lower part of Euphrates Basin near the Syrian border. An upward trend exists in annual minimum streamflow for only one station at the upper part of the Euphrates Basin.

Figure 2.

The detected trends in the annual minimum streamflow (a) and in the annual mean streamflow (b) on Euphrates Basin

The results of the Mann–Kendall and Spearman's Rho tests' applications to meteorological parameters at selected stations are summarized in Tables 6 and 7, which indicate that the Mann–Kendall and Spearman's Rho tests consistently yield almost the same results. No meaningful trend has been observed in the computations for daily minimum precipitation. The statistically significant Sen's slopes are presented in Table 8 for all parameters. The same period has been used for all trend calculations.

Table 6. The results of Mann–Kendall test for meteorological observation stations
Station numberStation namePrecipitationTemperatureHumidity
  TotalMaxMeanMaxMinMeanMaxMin
  1. (+): upward trend; (−): downward trend; (○): no trend.

17094Erzincan(+)(+)(+)(+)(+)
17096Erzurum(−)(−)(−)(+)(−)(+)(+)
17099Aǧri(+)(+)(+)
17165Tunceli(+)(+)
17199Malatya(+)(+)(+)
17201Elaziǧ(+)
17203Bingöl(+)
17204Muş(+)(+)
17261Gaziantep(+)(+)(+)(+)
17262Kilis(+)(+)
17265Adiyaman(+)(+)(+)
17270Şanliurfa(+)(+)(+)(+)(+)(+)
17275Mardin(+)(+)(−)(−)
17718Tercan(+)(−)(+)
17734Divriǧi(+)(+)
17736Mazgirt(+)
17740Hinis(+)
17744Karakoçan(+)(+)
17762Kangal(+)(+)
17764Arapkir(+)(+)(+)(−)(−)
17766Aǧin(+)
17768Çemişgezek(+)(+)(+)(+)
17776Solhan(+)(+)
17780Malazgirt(+)(+)(−)(−)
17804Keban(+)
17806Palu(+)(−)(−)
17808Genç(+)(−)
17842Balaban(−)(+)(+)(−)
17843Baskil(−)(−)(−)
17844Sivrice(+)(+)(−)(−)
17872Doǧanşehir(+)(+)
17874Çermik(+)(+)(+)
17910Kahta(+)
17912Siverek(+)(+)(+)(+)(+)(+)
17948Nusaybin(+)(−)(−)
17968Ceylanpinar(+)
17980Akçakale(−)(+)(+)
Table 7. The results of Spearman's Rho test for meteorological observation stations
Station numberStation namePrecipitationTemperatureHumidity
  TotalMaxMeanMaxMinMeanMaxMin
  1. (+): upward trend; (−): downward trend; (○): no trend.

17094Erzincan(+)(+)(+)(+)(+)
17096Erzurum(−)(−)(−)(+)(−)(+)(+)
17099Aǧri(+)(+)(+)
17165Tunceli(+)(+)
17199Malatya(+)(+)(+)
17201Elaziǧ(+)
17203Bingöl(+)
17204Muş(+)(+)
17261Gaziantep(+)(+)(+)(+)
17262Kilis(+)(+)
17265Adiyaman(+)(+)(+)
17270Şanliurfa(+)(+)(+)(+)(+)(+)
17275Mardin(+)(+)(−)(−)
17718Tercan(+)(−)(+)
17734Divriǧi(+)(+)
17736Mazgirt(+)
17740Hinis(+)
17744Karakoçan(+)(+)
17762Kangal(+)(+)
17764Arapkir(+)(+)(+)(−)(−)
17766Aǧin(+)(+)(+)
17768Çemişgezek(+)(+)(+)(+)
17776Solhan(+)(+)
17780Malazgirt(+)(+)(−)(−)
17804Keban(+)
17806Palu(+)(−)(−)
17808Genç(+)(−)
17842Balaban(−)(+)(+)(−)
17843Baskil(−)(−)(−)
17844Sivrice(+)(+)(−)(−)
17872Doǧanşehir(+)(+)
17874Çermik(+)(+)(+)
17910Kahta(+)
17912Siverek(+)(+)(+)(+)(+)(+)
17948Nusaybin(+)(−)(−)
17968Ceylanpinar(+)
17980Akçakale(−)(+)
Table 8. The values of Sen's slope for meteorological observation stations
Station numberStation namePrecipitationTemperatureHumidity
  Total cm year−1Max mm year−1Mean °C year−1Max °C year−1Min °C year−1Mean % year−1Max % year−1Min % year−1
17094Erzincan0.01150.03480.08250.04760.1443
17096Erzurum− 0.03390.0153− 0.05160.07340.0519
17099Aǧri− 0.01390.02340.03150.1413
17165Tunceli0.02340.1956
17199Malatya0.0360.02120.0172
17201Elaziǧ0.0183
17203Bingöl0.1464
17204Muş0.03460.0589
17261Gaziantep0.01960.02380.05370.0833
17262Kilis0.01740.0584
17265Adiyaman0.01730.03910.1917
17270Şanliurfa0.00880.00950.04180.09400.14290.0783
17275Mardin0.00920.0292− 0.0814− 0.0590
17718Tercan0.0273− 0.10820.1375
17734Divriǧi0.02720.0441
17736Mazgirt0.0774
17740Hinis0.0279
17744Karakoça0.34870.1809
17762Kangal0.04310.1192
17764Arapkir0.03940.03130.0367− 0.1743− 0.1140
17766Aǧin0.03730.04770.1409
17768Çemişgezek0.02660.06110.14510.1660
17776Solhan0.12650.1771
17780Malazgirt0.02420.0507− 0.3693− 0.1326
17804Keban0.0211
17806Palu0.0379− 0.28230.3935
17808Genç0.0534− 0.2891
17842Balaban0.12810.1439− 0.5833
17843Baskil− 0.5410− 0.3125− 0.4028
17844Sivrice0.04770.0750− 0.35211− 0.7578
17872Doǧanşehi0.03500.0444
17874Çermik0.46090.36110.4808
17910Kahta0.7202
17912Siverek0.01920.03330.03030.28420.16050.2439
17948Nusaybin0.0272− 0.0403− 0.2579
17968Ceylanpin0.1319
17980Akçakale0.1382

The thematic maps are drawn by ArcGIS allowing a variety of queries by overlay mapping techniques. In Figure 3, the detected trends are shown for total and maximum precipitation; mean, maximum and minimum temperature; mean, maximum and minimum humidity of the Euphrates Basin.

Figure 3.

The detected trends for Euphrates Basin: (a) annual total precipitation, (b) annual maximum precipitation, (c) annual mean temperature, (d) annual maximum temperature, (e) annual minimum temperature, (f) annual mean humidity, (g) annual maximum humidity, (h) annual minimum humidity

By examining the watershed-based queries (with a significant decreasing trend in the western, southern and southwestern sections of the basin for annual minimum flow), one can obtain meaningful results and see the links between the temperature and humidity values. By taking the thematic maps derived from provincial areas into account, the following interpretation may be derived from Figure 4.

Figure 4.

The overlayered map between (a) the downward trend of annual min. streamflow and the downward trend of total precipitation, (b) the downward trend of annual min. streamflow and the downward trend of max. precipitation, (c) the downward trend of annual minimum streamflow and the upward trend of mean humidity, (d) the downward trend of annual minimum streamflow and the upward trend of maximum humidity, (e) the downward trend of annual min. streamflow and the upward trend of mean temperature, (f) the downward trend of annual minimum streamflow and the upward trend of maximum temperature, (g) among the downward trend of annual minimum streamflow and the upward trend of mean humidity and the upward trend of mean temperature, (h) among the downward trend of annual min. streamflow and the upward trend of maximum humidity and the upward trend of maximum temperature for Euphrates

An obvious upward trend appears around Şanlıurfa, partially upward trends are observed around Erzurum and Erzincan and a downward trend is present around Mardin. However, partially downward trends are found around Aǧrı and Muş for the mean humidity value; again upward trends appear around Şanlıurfa and Gaziantep. On the other hand, partially upward trends become obvious around Adıyaman, Erzincan and Erzurum for the maximum humidity values in addition to a significant upward trend around Şanlıurfa Erzincan and Aǧrı with partially upward trends around Adıyaman for the minimum humidity values. Furthermore, upward trends exist around Gaziantep, Kilis and Malatya with partially upward trends around Şanlıurfa, Adıyaman, Tunceli, Sivas and Erzincan for the mean temperature values. Finally, there are clear upward trends around all cities for the maximum temperature values.

Partially upward trends are observable around Gaziantep, Kilis, Şanlıurfa and Erzincan in comparison to partially downward trends around Erzurum for the minimum temperature values. In the east a partially downward trend emerged around Erzurum for the total precipitation values together with a partially downward trend for the maximum precipitation values.

The generated spatial maps are overlayed according to their data types and the significant ones are given in Figure 4. The following results can be drawn from this figure.

Whilst a clear downward trend is observed around Şanlıurfa for annual mean minimum streamflow, an upward trend exists for the annual mean humidity values. In the meantime, obvious downward trends are found around Şanlıurfa and Adıyaman for annual mean minimum streamflow, with an upward trend for the annual maximum humidity values. However, significant downward trends prevail around Gaziantep, Kilis, Şanlıurfa, Adıyaman and Malatya for the annual mean minimum streamflow with an upward trend for the annual mean temperature values. Gaziantep, Mardin, Şanlıurfa, Adıyaman, Malatya, Aǧrı and Erzurum have downward trends for the annual mean minimum streamflow in addition to an upward trend for the annual maximum temperature values. Around Şanlıurfa there are downward trends for the annual mean minimum streamflow with upward counterparts for the annual mean temperature and humidity values. Important downward trends are observed around Şanlıurfa and Gaziantep for the annual mean minimum streamflow together with an upward trend for the annual maximum temperature and humidity values.

4. Conclusions

This work is an original case study employing trend analysis and development of thematic maps covering the upstream portion of the Euphrates basin in Turkey. The results are shown on a GIS based map.

The trend values in temperature, humidity and precipitation are processed in the form of maps, which are then used for the overlay method. In this manner, significant areas are indicated on the maps leading to the observable changes on thematic maps. Moreover, using this type of climatic change map, it is also possible to process and monitor the effects of other parameters, as well as topographical or other physical conditions. The new man-made changes (such as dams, new irrigation) could be included on the developed database-map and accordingly it will be possible to evaluate the positive/negative outcomes.

The overlay mapping technique supports detection of the climatic parameters' effects on the available water resources. It also helps to show the effect of climate change on water resources. Moreover, these maps are also important to evaluate the effects of climate change on the various water resources in the field. As a result of the visual evaluation of these maps, it may be possible to make effective assessments on parameters influencing the water resources to achieve their high performance, measures and economic management of the watershed.

By the evaluation of thematic regions and considering both the new irrigation areas and dams built in the GAP project within the last 15 years, the climate change effects can be assessed properly. For example, while an increasing humidity is observed in the lower Euphrates basin (at irrigation areas), the lower humidity values are observed in Mardin (at the high altitude areas). This is valid for both the mean and the maximum humidity values.

Serious increasing trends of the mean and maximum temperatures in conjunction with the partial decreasing trends of the annual total and maximum rainfall are important to make logical decisions concerning the significant decrease in streamflow. Besides, information of the addition of the decreasing trends on annual mean streamflow around the trans-boundary waters are significant for water resources planners.

The study has important results for the GAP region which includes many dams and irrigation projects. Because the GAP project is built on the Euphrates and Tigris rivers, which are the largest rivers of Turkey, the GAP project is still under construction and operation. Also, these rivers are very important transboundary waters of the countries of Turkey, Syria and Iraq.

The results of this study concerning the Euphrates Basin are likely to be useful for the following purposes:

  • investigation of the climatic change effects on water resources;

  • determination of possible decreases and its causes for water resources, and,

  • evaluation of water scarcity in future and drought situations on the water resources situations.

Even so, there may be some other uncertainties in the results. However this study is a first step for examining the relationship between climatic and hydrologic parameters of previous time spans (or periods) by using the Overlay Mapping Technique (OMT). Other analyses and techniques as well as OMT should be used in order to better decipher the complex relationships between climate change and water resources in this region.

Acknowledgements

This study is supported by the Scientific Research Projects Committee of Harran University (Project no: HÜBAK-862).

Ancillary