• Bloomfield, P., 1992: Trends in global temperatures. Climatic Change, 21 , 116.

  • Burn, H. B., , and Elnur M. A. H. , 2002: Detection of hydrologic trends and variability. J. Hydrol., 255 , 107122.

  • Cayan, D. R., , and Peterson D. H. , 1989: The influence of North Pacific atmospheric circulation on streamflow in the West. Geophys. Monogr., No. 55, Amer. Geophys. Union, 375–397.

    • Crossref
    • Export Citation
  • Cigizoglu, H. K., , Bayazıt M. , , Önöz B. , , Yıldız M. , , and Malkoç Y. , 2002: Trends in the maxımum, mean and low flows of Turkish rivers (in Turkish). Istanbul Technical University (ITU) Research Foundation Project Report, 265 pp.

  • Douglas, E. M., , Vogel R. M. , , and Kroll C. N. , 2000: Trends in floods and low flows in the United States: Impact of spatial correlation. J. Hydrol., 240 , 90105.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Groisman, P. Y., , and Easterling D. R. , 1994: Variability and trends in total precipitation and snowfall over the United States and Canada. J. Climate, 7 , 184205.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Guttman, N. B., , Wallis J. R. , , and Hosking J. R. M. , 1992: Regional temporal trends of precipitation quantiles in the U.S. Research Rep. RC 18453 (80774), IBM Research Division, Yorktown Heights, NY.

  • Hirsch, R. M., , and Slack J. R. , 1984: A nonparametric trend test for seasonal data with serial dependence. Water Resour. Res., 20 , 727732.

  • Hubbard, E. F., , Landwehr J. M. , , and Barker A. R. , 1997: Temporal variability in the hydrologic regimes of the United States. FRIEND’97—Regional Hydrology: Concepts and Models for Sustainable Water Resource Management, A. Gustard, Ed., IAHS Publication 246.

    • Search Google Scholar
    • Export Citation
  • Houghton, J. T., , Ding Y. , , Griggs D. J. , , Noguer M. , , van der Linden P. J. , , and Xiaosu D. , 2001: Climate Change 2001: The Scientific Basis. Cambridge University Press, 944 pp.

    • Search Google Scholar
    • Export Citation
  • Kadioğlu, M., 1997: Trends in surface air temperature data over Turkey. Int. J. Climatol., 17 , 511520.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Kahya, E., , and Dracup J. A. , 1993: US streamflows patterns in relation to the El Nino/Southern Oscillation. Water Resour. Res., 29 , 24912503.

  • Kahya, E., , and Kalaycı S. , 2004: Trend analysis of streamflow in Turkey. J. Hydrol., 289 , 128144.

  • Karl, T. R., , and Knight R. W. , 1998: Secular trends of precipitation amount, frequency, and intensity in the Unites States. Bull Amer. Meteor. Soc., 79 , 231241.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Lettenmaier, D. P., , Wood E. F. , , and Wallis J. R. , 1994: Hydro-climatological trends in the continental United States, 1948–88. J. Climate, 7 , 586607.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Lins, H. F., 1985: Streamflow variability in the United States: 1931–1978. J. Climate Appl. Climatol., 24 , 463471.

  • Lins, H. F., , and Slack J. R. , 1999: Streamflow trends in the United States. Geophys. Res. Lett., 26 , 227230.

  • Olsen, J. R., , Stedinger J. R. , , Matalas N. C. , , and Stakhiv E. Z. , 1999: Climate variability and flood frequency estimation for the upper Mississippi and lower Missouri rivers. J. Amer. Water Resour. Assoc., 35 , 15091520.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Önöz, B., , and Bayazit M. , 1995: Best-fit distributions of largest available flood samples. J. Hydrol., 167 , 195208.

  • Potter, K. W., 1991: Hydrological impacts of changing land management practices in a moderate-sized agricultural catchment. Water Resour. Res., 27 , 845855.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Tayanç, M., , Karaca M. , , and Yenigün O. , 1997: Annual and seasonal air temperature trend patterns of climate change and urbanization effects in relation to air pollutants in Turkey. J. Geophys. Res., 102 , D2,. 19091919.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Toros, H. D., , Deniz A. , , and Karan H. , 1994: Statistical analysis of western Anatolia precipitation (in Turkish). Proc. National Hydrometeorology Symp., Istanbul, Turkey, 185–209.

  • Türkeş, M., 1990: Arid regions and significant dry years in Turkey (in Turkish). Ph.D. thesis, Istanbul University Marine Sciences and Geography Institute, 195 pp.

  • Türkeş, M., 1996: Spatial and temporal analysis of annual rainfall variations in Turkey. Int. J. Climatol., 16 , 10571076.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Türkeş, M., , Utku M. S. , , and Kılıç G. , 1995: Variations and trends in annual mean air temperatures in Turkey with respect to climatic variability. Int. J. Climatol., 15 , 557569.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Vinnikov, K. Y., , Groisman P. Y. , , and Lugina K. M. , 1990: Empirical data on contemporary global climate changes (temperature and precipitation). J. Climate, 3 , 662677.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Yu, Y. S., , Zou S. , , and Whittemore D. , 1993: Non-parametric trend analysis of water quality data of rivers in Kansas. J. Hydrol., 150 , 6180.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Zhang, X., , Harvey K. D. , , Hogg W. D. , , and Yuzyk T. R. , 2001: Trends in Canadian streamflow. Water Resour. Res., 37 , 987998.

  • View in gallery

    (a) Map showing the hydrologic regions of Turkey. (b) Map showing the flow gauging stations in Turkey.

  • View in gallery

    Flow stations where both parametric and nonparametric tests found trend on annual maximum flows (up arrow: positive trend; down arrow: negative trend).

  • View in gallery

    Flow stations where the t test found trend on annual mean flows (up arrow: positive trend; down arrow: negative trend).

  • View in gallery

    Flow stations where the Mann–Kendall test found trend on annual mean flows (up arrow: positive trend; down arrow: negative trend).

  • View in gallery

    Flow stations where the parametric test found trend on 1-day annual low flows (up arrow: positive trend; down arrow: negative trend).

  • View in gallery

    Flow stations where the parametric test found trend on 7-day annual low flows (up arrow: positive trend; down arrow: negative trend).

  • View in gallery

    Flow stations where the Mann–Kendall test found trend on 1-day annual low flows (up arrow: positive trend; down arrow: negative trend).

  • View in gallery

    Flow stations where Mann–Kendall test found trend on 7-day annual low flows (up arrow: positive trend, down arrow: negative trend).

  • View in gallery

    Flow stations where parametric test found trend on annual mean, 1-day and 7-day annual low flows (up arrow: positive trend, down arrow: negative trend).

  • View in gallery

    Flow stations where Mann–Kendall test found trend on annual mean, 1-day and 7-day annual low flows (up arrow: positive trend, down arrow: negative trend).

  • View in gallery

    Flow stations where parametric test found trend on annual maximum, annual mean, 1-day and 7-day annual low flows (up arrow: positive trend, down arrow: negative trend).

  • View in gallery

    Flow stations where Mann–Kendall test found trend on annual maximum, annual mean, 1-day and 7-day annual low flows (up arrow: positive trend, down arrow: negative trend).

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 175 175 14
PDF Downloads 107 107 14

Trends in the Maximum, Mean, and Low Flows of Turkish Rivers

View More View Less
  • 1 Civil Engineering Faculty, Division of Hydraulics, Istanbul Technical University, Maslak, Turkey
© Get Permissions
Full access

Abstract

In this study the existence of trend in maximum, mean, and low flows of Turkish rivers has been investigated. The data consisted of the daily mean flows of nearly 100 flow stations in 24 hydrological regions of Turkey. Trend analysis has been carried out using the parametric t test and nonparametric τ (Mann–Kendall) test. Both tests have been applied to annual maximum, mean, 1-day, and 7-day low flows. Trend existence was detected in the majority of rivers in western and southern Turkey and in some parts of central and eastern Turkey. Trends in mean and low flows were more common compared with maximum flows. Except at a few stations, flows showed a decreasing trend. In the time period of the last 30–60 yr, statistically significant decrease was found especially in the mean and low flows (and in some of the maximum flows) in western, central, and southern parts of Turkey. Such trends were not observed in other regions. These results are in agreement with those of the precipitation trend studies in Turkey.

Corresponding author address: Dr. H. K. Cigizoglu, Civil Engineering Faculty, Division of Hydraulics, Istanbul Technical University, Maslak 34469, Turkey. Email: cigiz@itu.edu.tr

Abstract

In this study the existence of trend in maximum, mean, and low flows of Turkish rivers has been investigated. The data consisted of the daily mean flows of nearly 100 flow stations in 24 hydrological regions of Turkey. Trend analysis has been carried out using the parametric t test and nonparametric τ (Mann–Kendall) test. Both tests have been applied to annual maximum, mean, 1-day, and 7-day low flows. Trend existence was detected in the majority of rivers in western and southern Turkey and in some parts of central and eastern Turkey. Trends in mean and low flows were more common compared with maximum flows. Except at a few stations, flows showed a decreasing trend. In the time period of the last 30–60 yr, statistically significant decrease was found especially in the mean and low flows (and in some of the maximum flows) in western, central, and southern parts of Turkey. Such trends were not observed in other regions. These results are in agreement with those of the precipitation trend studies in Turkey.

Corresponding author address: Dr. H. K. Cigizoglu, Civil Engineering Faculty, Division of Hydraulics, Istanbul Technical University, Maslak 34469, Turkey. Email: cigiz@itu.edu.tr

1. Introduction

In general, observational and historical hydroclimatologic data are used in the planning and designing of water resources projects. It has been documented that the global average surface temperature has increased by 0.6° ± 0.2°C over the twentieth century (Houghton et al. 2001) due to atmospheric concentrations of trace gases such as carbon dioxide. This process could lead to abnormalities in other climate parameters such as precipitation and evapotranspiration in a specific region. Streamflow is sensitive to changes in precipitation and other climate parameters. Hence it is informative to investigate whether streamflow records exhibit evidence of trends that may be linked to climate change.

Trend existence in the annual maximum (flood), mean, and low flows in rivers carries significance for different types of water resources problems. Floods and mean flows are considered in the design of flood mitigation structures and water storage reservoirs. Low flows are especially significant for the water quantity released to the downstream of a dam in order to protect the ecological sustainability. Among regional streamflow-trend studies in the world, Zhang et al. (2001) stated that monthly mean streamflow in Canada for most months decreased, with the strongest decrease in summer and autumn months, and there was almost no basin exhibiting upward trend. In contrast, Lettenmaier et al. (1994) presented the upward streamflow trend pattern, at its peak in midwinter, covering most of the United States with the exception of a small number of downtrends concentrated in the Northwest, Florida, and coastal Georgia. Lins and Slack (1999) came to a similar conclusion studying streamflow trends calculated for selected quantiles of discharge. Lettenmaier et al. (1994) also stressed that the trends in the streamflow are not fully parallel to the changes in precipitation and temperature because of a combination of climate and water management effects. However, Burn and Elnur (2002) indicated the similarities in trends and patterns in the hydrological variables and in meteorological variables at chosen locations in Canada, implying the relations between the two groups of variables.

Most of the previous studies regarding trends in surface climatic variables in Turkey were concentrated on temperature and precipitation patterns. For example, Türkeş et al. (1995) used various nonparametric tests to identify abrupt changes and trends in the long-term mean temperature of both individual stations and geographical regions in Turkey during the 1930–92 period. They found that climate tended to be warmer in the eastern Anatolia and to be cooler particularly in the Marmara and Mediterranean regions using regional mean temperature series. Türkeş (1996) worked with the area-averaged annual rainfall series during the 1930–93 period and pointed out that decreases were generally observed over Turkey, particularly in the Black Sea and Mediterranean regions. The application of a trend detection framework in Turkish precipitation data has resulted in the identification of some significant trends, especially January, February, and September precipitations and annual average precipitation (Partal and Kahya 2005, manuscript submitted to Hydrol. Processes). In this study a noticeable decrease in the annual mean precipitation was found mostly in the west of Turkey, south of Turkey, and the shore of the Black Sea region. Kadıoğlu (1997) examined trends in the mean annual temperature records during the 1939–89 period in 18 stations across Turkey and found insignificant increasing trends in the mean annual temperatures. He also indicated that a regional increase in mean minimum temperatures, which could be attributed to the urban heat island effect, appeared around 1955. His results are inconclusive for the existence of long-term trends. In contrast, Tayanç et al. (1997) found statistically significant cooling in mean temperatures mostly in northern Turkey and warming mostly in large urban locations. In the same context, Karaca et al. (1995) showed the urban heat island intensity in Istanbul, although it is surrounded by the Black Sea and the Marmara Sea. Kahya and Kalaycı (2004) applied four nonparametric trend tests for monthly mean flow data. As a result of their study, basins located in western Turkey, in general, exhibit downward trend, significant at the 0.05 or lower level, whereas basins located in eastern Turkey show no trend. In most cases, four tests provided the same conclusion about trend existence.

To stress the importance of trend analysis of hydrologic variables (streamflow as the most attracting variable), in a watershed that is assumed not to be exposed to anthropogenic influences, the following explanations based on the work of Zhang et al. (2001) are presented. Under certain geomorphic conditions, the nature of rivers reflects the integrated watershed response to climatic forcing. This critical point was previously noted by Cayan and Peterson (1989) and Kahya and Dracup (1993) in searching for teleconnections between the surface hydroclimatic variables and the large-scale atmospheric circulation. Since the geomorphologic evolution of watershed is quite slow in comparison with climate change, the detectable changes in the hydrologic regimes of stable, unregulated watersheds may be considered as the reflection of changes in climate. Consequently, hydrologic variables might be used as indicators to detect and monitor climate change.

In light of the review of major trend studies covering Turkey and the fact of streamflow being a privileged variable as stated above, a study regarding streamflow trend analysis in the geography of Turkey seemed to be an important necessity. The objective of this investigation is to document trend characteristics of Turkish daily mean streamflow data for evidence of climate change. Trend analysis detection was carried out using a parametric test and a nonparametric test. The tests were adopted for annual maximum, mean, and low flows separately, and the results were analyzed on the regional scale for the whole country.

2. Data

In this study the daily flow data compiled by the General Directorate of Electrical Power Resources Survey and Development Administration (EIE) throughout Turkey were used. Turkey is divided into 26 hydrologic basins (Fig. 1a). In most hydroclimatologic studies, a completely homogeneous dataset has been rarely used. Thus, the common practice in most cases is to put forward reasonable criteria for the condition of homogeneity. For example, Lins (1985) included streamflow stations on watercourses where diversion amounts have been less than 10% of the mean flow and storage capacity amounted to less than 10% of the mean annual runoff. To comply with the homogeneity condition, a total of 96 streamflow gauging stations distributed over 24 hydrologic regions (Fig. 1b), where there was no reported regulation or diversion upstream, have been selected among more than 300 stations. The homogeneity was confirmed by checking out the data for man-made changes, such as jumps due to relocation of station, regulation, or diversion due to the presence of dams or weirs. The drainage areas of the selected river flow stations show a variation between 166 and 25 516 km2. The record length changes between 25 and 66 yr.

3. Techniques for trend detection

a. Parametric test

Assuming that two random variables X and Y have bivariate normal distribution as joint probability distribution, the t statistic, defined depending on rx,y, has t distribution as sampling distribution with a degree of freedom N − 2, where N is the number of elements in the sample, and the t statistic is computed as
i1525-7541-6-3-280-eq1
The t statistic is used to check the Ho: ρX,Y = 0 hypothesis against Ho: ρX,Y ≠ 0. If the computed t value falls within the confidence interval corresponding to α significance level, the ρX,Y = 0 hypothesis will be accepted. Otherwise the ρX,Y = 0 hypothesis will be rejected and the ρX,Y ≠ 0 hypothesis will be accepted. In the study presented, the X and Y variables represent the yearly flow statistic (maximum, mean, or minimum) and time (year), respectively.

b. Mann–Kendall test

This technique, based on the Kendall tau statistic, τ, has been widely used to test for randomness against trend in climatological time series (Zhang et al. 2001). In this test, the null hypothesis Ho states that the deseasonalized data (x1, . . . , xn) are a sample of n independent and identically distributed random variables (Yu et al. 1993). The alternative hypothesis H1 of a two-sided test is that the distribution of xk and xj are not identical for all k, j ≤ n with kj. The test statistic S calculated with Eqs. (1) and (2) below has mean zero and variance of S, computed by Var(S) = [n(n −1)(2n + 5) − Σtt(t − 1)(2t + 5)]/18,
i1525-7541-6-3-280-e1
i1525-7541-6-3-280-e2
and is asymptotically normal (Hirsch and Slack 1984), where t is the extent of any given tie, and Σt denotes the summation over all ties. For the cases in which n is larger than 10, the standard normal variate z is computed by using the following equation (Douglas et al. 2000):
i1525-7541-6-3-280-e3
Thus, in a two-sided test for trend, the Ho should be accepted if |z| ≤ zα/2 at the α level of significance. A positive value of S indicates an “upward trend” and a negative value indicates a “downward trend.”

4. Method of the trend analysis

The trend detection work was done in three steps, that is, the analysis of annual maximum flows, mean flows, and low flows. Once the maximum, mean, and low flows for each water year of the whole historical record were computed, they were plotted with respect to time (year as time interval). The trend line on a plot was established, and the correlation between time and flow statistic (maximum, mean, or low flow) was computed. The parametric test (t test) was applied at the 5% significance level. Then the nonparametric test (Mann–Kendall test) was employed at the same significance level.

5. Trend analysis of maximum flows

As defined, annual maximum flow is the maximum flow measured in the water year under consideration. In this part of the study trends in the annual maximum flows belonging to 96 different stations in 24 hydrologic regions of Turkey were investigated. The variation interval of the statistics for the annual maximum flows of these stations are presented in Table 1. The coefficient of variation, cυ, changes between 0.25 and 1.79. The maximum of annual maximums, xmax, shows a variation between 50 and 3591 m3 s−1. The drainage area, observation length, mean, x, and standard deviation, sx, of the annual maximum flows and the maximum of the maximum flows are presented in Cigizoglu et al. (2002). The parametric and nonparametric trend test results are summarized below.

a. Results of parametric test

In the trend analysis, trends have been detected at 14 of the total 96 stations at the 5% significance level. Three of these stations display positive trend (stations 1222, 1331, and 1418), suggesting increase in maximum flow, and 11 are of decreasing (negative) type (stations 328, 407, 509, 725, 1102, 1223, 1224, 1401, 1412, 2215, and 2218). Among the total 24 hydrologic regions no trend in maximum flows was detected in 12 regions (hydrologic regions: 6, 8, 9, 10, 15, 16, 18, 20, 23, 24, 25, and 26; Fig. 1).

b. Nonparametric test results

The Mann–Kendall test found trends at 15 flow stations. Only four stations had positive trend (stations 1222, 1331, 1418, and 2228) whereas the remaining trends were of decreasing type (stations 105, 328, 407, 601, 1102, 1223, 1224, 1401, 1412, 2215, and 2218). Differing from the parametric test, the Mann–Kendall test found in one more station positive trend (station 2228). On the other hand, in two stations the Mann–Kendall test detected decreasing trend in contrast to the parametric test (stations 105 and 601) whereas only the parametric test could point to a negative trend in one station (station 509). The flow stations with trend are presented in Table 2 and illustrated in Fig. 2.

The stations where both tests display trend existence in maximum flows are situated in northwestern and northern parts of Turkey (Fig. 2). The differences between the parametric test and the Mann–Kendall test can be explained with the marginal probability distribution assumption in the parametric test. In applying the parametric test the variable is assumed to be represented with normal distribution. Since the annual maximum flows are skewed, another probability distribution such as generalized extreme value distribution should be considered for these series (Önöz and Bayazit 1995).

6. Trend analysis of annual mean flows

Annual mean flow is the average of all the measured flow values for a specific water year. In this part of the study the trend existence at 96 flow stations in the Turkish drainage basins was investigated. The variation interval of basic statistics for the annual mean flows of these stations are presented in Table 3. The coefficient of variation changes between 0.11 and 0.96 whereas the skewness has a minimum −2.05 and a maximum 3.34. The parametric test and Mann–Kendall test are applied to the annual mean flows, and the trend existence was investigated at the 5% significance level.

a. Parametric test results

Trend analysis with the parametric test displayed trend existence at 27 flow stations. The stations with trend are plotted in Fig. 3. All of the detected trends are negative (decreasing trend). If we analyze the test results on the basin basis, all of the stations in three drainage basins (basins 6, 8, and 17) show negative trend. Basins 10 and 11 show the same behavior, but each of these two basins are represented only with a single station in the study. In contrast, the parametric test could not find any trend in eight hydrologic regions (regions 9, 14, 15, 20, 22, 23, 24, and 26).

b. Mann–Kendall test results

Similar to the parametric test, the Mann–Kendall test displayed trend at 27 flow stations (Fig. 4). All of the trends are negative. Differing from the parametric test, no trend was found at two stations (stations 523 and 1314), but trend was detected at two different stations (stations 311 and 2505) in contrast. According to the Mann–Kendall test, no trend was encountered in nine hydrologic regions (regions 9, 13, 14, 15, 20, 22, 23, 24, and 26). On the other hand, three regions had trend in all of the flow stations (regions 8, 16, and 17). The test pointed to negative trend existence also in four other hydrologic regions (regions 6, 10, 11, and 25), which were represented in the study only with one flow station. The regions with trend are the northwestern part (Marmara region), western part (significant part of Aegean region), southern part (significant part of Mediterranean region), western part of Middle Anatolia (Sakarya basin in particular), and some parts of eastern and southeastern Anatolia in Turkey.

It is seen that trend test results provided by two tests are quite close to each other. This can be explained with the annual mean flows having a distribution close to normal distribution. The skewness of annual mean flows is close to zero, in general verifying the symmetric distribution assumption.

7. Trend analysis of annual low flows

In this part of the study the trend analysis of 1-day and 7-day low flows was the focus. As a definition annual 1-day low flow is the lowest daily mean flow along a water year. Seven-day low flow is the lowest of the consecutive 7-day mean values within a water year. Both parametric and Mann–Kendall tests were employed for annual low flows in trend detection work.

a. Parametric test results

A total of 96 flow measurement stations belonging to 24 hydrologic regions were included in low-flow trend analysis. Application of the parametric test to 1-day low flows resulted in trend detection in 44 flow stations throughout the country (Fig. 5). Except for 4 stations (stations 106, 515, 1331, and 2620), all of the 40 stations displayed decreasing trend. For the 7-day low flows, trend was found at 43 stations. Among them, only three showed increasing trend (Fig. 6). It was seen that the number of flow stations where trend was observed both for 1-day and 7-day low flows was 39.

In nine hydrologic basins (basins 3, 4, 5, 6, 8, 10, 11, 17, and 21) all the flow stations showed trend in 1-day and 7-day low flows. However, it should be noted that each of the hydrologic basins 4, 10, and 11 are represented with a single station in the study. Five of the eight stations in hydrologic basin 12 displayed trend existence in both 1-day and 7-day low flows.

On the regional basis, trend existence in low flows were detected intensively in western and southern parts of Turkey. Central and eastern regions displayed trend existence only in some parts.

b. Mann–Kendall test results

The parametric test was also employed for low flows in order to find the regions with trend. Accordingly, the Mann–Kendall test pointed out 42 stations with trend existence both in 1-day and 7-day low flows. A total of 46 trends were detected for 1-day low flows, whereas this value was 45 for 7-day low flows (Figs. 7 and 8). Except for five stations (stations 106, 1331, 1517, 2304, and 2620), all the trends were of decreasing type for 1-day low flows. Only four trends were positive for 7-day low flows (stations 106, 1517, 2304, and 2620). The parametric and Mann–Kendall tests did not agree for all flow stations in trend detection. The differences between the two tests are presented in Tables 4 and 5 for 1-day and 7-day low flows, respectively.

On the basin scale the Mann–Kendall test showed trend existence in all stations for eight regions (3, 4, 5, 6, 8, 10, 11, and 16) for both 1-day and 7-day low flows. Similar to parametric test results, trends were widely spread in the western, southern, and eastern parts of Turkey whereas only some parts of the central region showed trend existence.

8. Trends in precipitation data

It is investigated how the trends in the streamflows of Turkish rivers are related to the trends in precipitation. A number of studies have been performed analyzing trends in Turkish precipitation data. Toros et al. (1994) have examined seasonal and annual rainfall data of the western part of Anatolia using 68 stations for the period 1930–92. They found a decrease in rainfall after 1982 and argued that this decrease was not due to climatic change but that it was evidence only of a fluctuation in rainfall.

Türkeş (1996) has analyzed the annual and seasonal rainfall data of 91 stations for the period 1930–93 and reported a decreasing trend over Turkey as a whole and particularly over the Black Sea and Mediterranean regions. Türkeş (1996), analyzing the same data, found that the time series plots for all of Turkey indicated a slightly downward trend in annual rainfall. The normalized annual rainfall series of 17 stations showed a significant trend (at the 5% level) in the mean, and 15 of them were decreasing trends. Significant decreasing trends were the major characteristics of the Mediterranean region. Many of the significant downward trends appeared to have occurred as a result of abrupt decreases during the last 20–25 yr of the study period. Abrupt decreases and drought-dominated conditions in annual rainfall over most of Turkey could be associated generally with the persistence of drought-favoring anticyclonic weather types over the central and eastern Mediterranean basin (Türkeş 1990).

Recently, Partal and Kahya (2005), manuscript submitted to Hydrol. Processes) investigated the precipitation data of 96 stations in Turkey for the period 1929–93. Monthly, seasonal, and annual data were subjected to trend detection analysis using Mann–Kendall, sequential Mann–Kendall, and Sen’s t tests. The tests were repeated using the index time series for the regional trend analysis of the seven clusters in Turkey. It was found that the majority of the annual precipitation data had negative trends at the 5% significance level in western and southern Turkey and some of the Black Sea region, and also at some stations in northeast and eastern Turkey. Tests with the index series showed that the trends found at the stations were related to the regional trends.

Trends of Turkish precipitation data detected in the above studies are in general parallel to those found in this study in relation with the streamflow data. In both cases, significant negative trends were detected in western and southern Turkey, and in some parts of central and eastern Turkey. This shows that the downward trends of the maximum, mean, and low flows can be explained in terms of the climate change resulting in significantly reduced precipitation.

Potter (1991) indicated that there are often large trends in streamflow basins affected by changing land use (in particular changing agricultural practices and land-use changes associated with urbanization), but for which there are no dams and no diversion. However, as mentioned above, the Turkish trends in both precipitation and streamflow are parallel to each other. Besides, both the General Directorate of Electrical Power Resources Survey of Turkey, the organization that provided the data used in this study, and the General Directorate of State Water Works of Turkey stated that noticeable agricultural practices and land-use changes were not observed during the data periods considered for this study.

9. Discussion and conclusions

The presented work covered the trend investigation of different flow statistics in Turkey. Trend existence was more common in mean and low flows compared with maximum flows. Except for a few flow series, trend is generally of decreasing type. Considering a time period between 25 and 60 yr, statistically significant decrease was detected mainly in annual mean and low flows (also in maximum flows of some flow series) on the western, central, eastern, and southern parts of Turkey. Results were further evaluated considering all statistics together. The flow series that displayed trend on all of the three statistics (annual mean, 1-day, and 7-day low flows) were investigated using both parametric and Mann–Kendall tests together. Accordingly, parametric and Mann–Kendall tests found decreasing trend in 21 and 22 stations, respectively (Figs. 9 and 10) for all these statistics. A final examination was carried out considering all statistics (annual maximum, mean, 1-day, and 7-day low flows). In this case both parametric and Mann–Kendall tests detected six flow series with trend existence for all statistics Figs. (11 and 12).

The decreasing trend in annual maximum, mean, and low flows carries significance in water resources engineering for different reasons. The decrease in low flows may indicate the increase of the dry periods within a year in these rivers. This is especially true for summer months when irrigation is quite significant. Another problem would be the increase in the contaminant concentration in the case of wastewater discharge to the river. The decrease in low flows is especially important for the location of the water treatment facility, the quantity of irrigation, and drinking water. The changes in low-flow statistics also affect the minimum water quantity released by dams downstream for sustainable protection of ecological cycles.

The decrease in annual mean flows, on the other hand, plays a dominant role in the determination of water reservoir capacity and in the reservoir management works afterward. It can be said that the negative trend in mean flows affects the dimensions and hence the construction cost of the dam and can be also considered as a limiting factor for the water quantity withdrawn from the reservoir. A trend in maximum flows, however, is significant in particular for spillway design. As is well known, the flood flows are considered for finding the crest height of a dam. Similarly, flood flows are also significant for the design of flood protection structures in the rivers.

Trends detected in the flow data of Turkish rivers are in general parallel to those found in precipitation data by other authors. Therefore, it may be concluded that the recent climate change has caused the downward trends in the precipitation and runoff in certain regions of Turkey.

Acknowledgments

The authors would like to present their thanks to the General Directorate of Electrical Power Resources Survey and Development Administration of Turkey for the submission of river flow data and related information for the drainage basins. The financial support of the Science Foundation of Istanbul Technical University during the whole research study is also deeply appreciated by the authors.

REFERENCES

  • Bloomfield, P., 1992: Trends in global temperatures. Climatic Change, 21 , 116.

  • Burn, H. B., , and Elnur M. A. H. , 2002: Detection of hydrologic trends and variability. J. Hydrol., 255 , 107122.

  • Cayan, D. R., , and Peterson D. H. , 1989: The influence of North Pacific atmospheric circulation on streamflow in the West. Geophys. Monogr., No. 55, Amer. Geophys. Union, 375–397.

    • Crossref
    • Export Citation
  • Cigizoglu, H. K., , Bayazıt M. , , Önöz B. , , Yıldız M. , , and Malkoç Y. , 2002: Trends in the maxımum, mean and low flows of Turkish rivers (in Turkish). Istanbul Technical University (ITU) Research Foundation Project Report, 265 pp.

  • Douglas, E. M., , Vogel R. M. , , and Kroll C. N. , 2000: Trends in floods and low flows in the United States: Impact of spatial correlation. J. Hydrol., 240 , 90105.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Groisman, P. Y., , and Easterling D. R. , 1994: Variability and trends in total precipitation and snowfall over the United States and Canada. J. Climate, 7 , 184205.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Guttman, N. B., , Wallis J. R. , , and Hosking J. R. M. , 1992: Regional temporal trends of precipitation quantiles in the U.S. Research Rep. RC 18453 (80774), IBM Research Division, Yorktown Heights, NY.

  • Hirsch, R. M., , and Slack J. R. , 1984: A nonparametric trend test for seasonal data with serial dependence. Water Resour. Res., 20 , 727732.

  • Hubbard, E. F., , Landwehr J. M. , , and Barker A. R. , 1997: Temporal variability in the hydrologic regimes of the United States. FRIEND’97—Regional Hydrology: Concepts and Models for Sustainable Water Resource Management, A. Gustard, Ed., IAHS Publication 246.

    • Search Google Scholar
    • Export Citation
  • Houghton, J. T., , Ding Y. , , Griggs D. J. , , Noguer M. , , van der Linden P. J. , , and Xiaosu D. , 2001: Climate Change 2001: The Scientific Basis. Cambridge University Press, 944 pp.

    • Search Google Scholar
    • Export Citation
  • Kadioğlu, M., 1997: Trends in surface air temperature data over Turkey. Int. J. Climatol., 17 , 511520.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Kahya, E., , and Dracup J. A. , 1993: US streamflows patterns in relation to the El Nino/Southern Oscillation. Water Resour. Res., 29 , 24912503.

  • Kahya, E., , and Kalaycı S. , 2004: Trend analysis of streamflow in Turkey. J. Hydrol., 289 , 128144.

  • Karl, T. R., , and Knight R. W. , 1998: Secular trends of precipitation amount, frequency, and intensity in the Unites States. Bull Amer. Meteor. Soc., 79 , 231241.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Lettenmaier, D. P., , Wood E. F. , , and Wallis J. R. , 1994: Hydro-climatological trends in the continental United States, 1948–88. J. Climate, 7 , 586607.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Lins, H. F., 1985: Streamflow variability in the United States: 1931–1978. J. Climate Appl. Climatol., 24 , 463471.

  • Lins, H. F., , and Slack J. R. , 1999: Streamflow trends in the United States. Geophys. Res. Lett., 26 , 227230.

  • Olsen, J. R., , Stedinger J. R. , , Matalas N. C. , , and Stakhiv E. Z. , 1999: Climate variability and flood frequency estimation for the upper Mississippi and lower Missouri rivers. J. Amer. Water Resour. Assoc., 35 , 15091520.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Önöz, B., , and Bayazit M. , 1995: Best-fit distributions of largest available flood samples. J. Hydrol., 167 , 195208.

  • Potter, K. W., 1991: Hydrological impacts of changing land management practices in a moderate-sized agricultural catchment. Water Resour. Res., 27 , 845855.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Tayanç, M., , Karaca M. , , and Yenigün O. , 1997: Annual and seasonal air temperature trend patterns of climate change and urbanization effects in relation to air pollutants in Turkey. J. Geophys. Res., 102 , D2,. 19091919.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Toros, H. D., , Deniz A. , , and Karan H. , 1994: Statistical analysis of western Anatolia precipitation (in Turkish). Proc. National Hydrometeorology Symp., Istanbul, Turkey, 185–209.

  • Türkeş, M., 1990: Arid regions and significant dry years in Turkey (in Turkish). Ph.D. thesis, Istanbul University Marine Sciences and Geography Institute, 195 pp.

  • Türkeş, M., 1996: Spatial and temporal analysis of annual rainfall variations in Turkey. Int. J. Climatol., 16 , 10571076.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Türkeş, M., , Utku M. S. , , and Kılıç G. , 1995: Variations and trends in annual mean air temperatures in Turkey with respect to climatic variability. Int. J. Climatol., 15 , 557569.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Vinnikov, K. Y., , Groisman P. Y. , , and Lugina K. M. , 1990: Empirical data on contemporary global climate changes (temperature and precipitation). J. Climate, 3 , 662677.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Yu, Y. S., , Zou S. , , and Whittemore D. , 1993: Non-parametric trend analysis of water quality data of rivers in Kansas. J. Hydrol., 150 , 6180.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Zhang, X., , Harvey K. D. , , Hogg W. D. , , and Yuzyk T. R. , 2001: Trends in Canadian streamflow. Water Resour. Res., 37 , 987998.

Fig. 1.
Fig. 1.

(a) Map showing the hydrologic regions of Turkey. (b) Map showing the flow gauging stations in Turkey.

Citation: Journal of Hydrometeorology 6, 3; 10.1175/JHM412.1

Fig. 2.
Fig. 2.

Flow stations where both parametric and nonparametric tests found trend on annual maximum flows (up arrow: positive trend; down arrow: negative trend).

Citation: Journal of Hydrometeorology 6, 3; 10.1175/JHM412.1

Fig. 3.
Fig. 3.

Flow stations where the t test found trend on annual mean flows (up arrow: positive trend; down arrow: negative trend).

Citation: Journal of Hydrometeorology 6, 3; 10.1175/JHM412.1

Fig. 4.
Fig. 4.

Flow stations where the Mann–Kendall test found trend on annual mean flows (up arrow: positive trend; down arrow: negative trend).

Citation: Journal of Hydrometeorology 6, 3; 10.1175/JHM412.1

Fig. 5.
Fig. 5.

Flow stations where the parametric test found trend on 1-day annual low flows (up arrow: positive trend; down arrow: negative trend).

Citation: Journal of Hydrometeorology 6, 3; 10.1175/JHM412.1

Fig. 6.
Fig. 6.

Flow stations where the parametric test found trend on 7-day annual low flows (up arrow: positive trend; down arrow: negative trend).

Citation: Journal of Hydrometeorology 6, 3; 10.1175/JHM412.1

Fig. 7.
Fig. 7.

Flow stations where the Mann–Kendall test found trend on 1-day annual low flows (up arrow: positive trend; down arrow: negative trend).

Citation: Journal of Hydrometeorology 6, 3; 10.1175/JHM412.1

Fig. 8.
Fig. 8.

Flow stations where Mann–Kendall test found trend on 7-day annual low flows (up arrow: positive trend, down arrow: negative trend).

Citation: Journal of Hydrometeorology 6, 3; 10.1175/JHM412.1

Fig. 9.
Fig. 9.

Flow stations where parametric test found trend on annual mean, 1-day and 7-day annual low flows (up arrow: positive trend, down arrow: negative trend).

Citation: Journal of Hydrometeorology 6, 3; 10.1175/JHM412.1

Fig. 10.
Fig. 10.

Flow stations where Mann–Kendall test found trend on annual mean, 1-day and 7-day annual low flows (up arrow: positive trend, down arrow: negative trend).

Citation: Journal of Hydrometeorology 6, 3; 10.1175/JHM412.1

Fig. 11.
Fig. 11.

Flow stations where parametric test found trend on annual maximum, annual mean, 1-day and 7-day annual low flows (up arrow: positive trend, down arrow: negative trend).

Citation: Journal of Hydrometeorology 6, 3; 10.1175/JHM412.1

Fig. 12.
Fig. 12.

Flow stations where Mann–Kendall test found trend on annual maximum, annual mean, 1-day and 7-day annual low flows (up arrow: positive trend, down arrow: negative trend).

Citation: Journal of Hydrometeorology 6, 3; 10.1175/JHM412.1

Table 1.

The variation interval of statistical parameters for annual maximum flows (x: mean; sx: standard deviation; cυ: coefficient of variation; xmax: maximum of annual maximums).

Table 1.
Table 2.

List of the flow stations that have trends on annual maximum flows (↑: positive trend; ↓: negative trend).

Table 2.
Table 3.

The variation interval of statistical parameters for annual mean flows (x: mean; sx: standard deviation; cυ: coefficient of variation; csx: skewness).

Table 3.
Table 4.

List of the flow stations where parametric and Mann–Kendall tests provided different results for 1-day minimum flows (↑: positive trend; ↓: negative trend).

Table 4.
Table 5.

List of the flow stations where parametric and Mann–Kendall tests provided different results for 7-day minimum flows (↑: positive trend; ↓: negative trend).

Table 5.
Save