Impact assessment of climate change on hydro-climatic conditions of arid and semi-arid watersheds (case study: Zoshk-Abardeh watershed, Iran)

The hydrologic cycle in the river basins of semi-arid regions is severely influenced by climate change. The aim of this study is to assess the impact of climate change on the hydro-climatic condition in Zoshk-Abardeh watershed in eastern Iran. The Soil and Water Assessment Tool (SWAT) was calibrated using the Sequential Uncertainty Fitting – Version 2 (SUFI-2) algorithm to improve the simulation results of the runoff. The Model for Interdisciplinary Research on Climate-Earth System Models (MIROC-ESM) was used to investigate the effects of climate change on hydro-climatic components under the representative concentration pathway scenarios (RCPs: 2.6, 4.5, 6.0, and 8.5) and in near(2014–2042), mid(2042–2071), and far(2072–2100) futures. The temperature component under the RCP4.5 and RCP6.0 during the nearand mid-future intervals and the far-future period (for RCP6.0) indicated a significant rising trend. The rainfall parameter in all RCPs and future intervals showed an insignificant descending trend. Runoff alterations under the RCP4.5 amid the midto far-future intervals and under the RCP8.5 throughout the far-future period trailed a significant descending trend. The results determined that the temperature will track an upward tendency, while precipitation and runoff will follow a descending trend in this watershed by the end of the 21st


INTRODUCTION
Climatic paleontology evidences that climate change has always been present throughout the history of the planet, but the climatic changes of the last century have two distinct features, as compared with past climatic changes. First, human activities play a greater role in the nature of the current climate change. Second, the speed of recent climatic change is greater so that a considerable number of changes will be occurring in the Earth's atmosphere over the short term (IPCC ). The average surface temperature of the Earth is increasing, which is mainly the result of greenhouse gas emissions (Sartori ). are the secondary aspects of climate change. These changes are not simply made due to an increase in greenhouse gas concentrations, and they occur in response to global warming (changes in the concentration of aerosols directly affect clouds and precipitation; however, this effect is most severe on temperature). Thus, in many areas, the greatest impact of climate change on hydrology is likely due not only to the relatively small changes in the turbulent behavior of clouds and precipitation but also to the direct effect of temperature rise on the hydrology cycle. In this case, the amount of evapotranspiration from the surface of the soil and plants will necessarily increase significantly, while the size of the snow grains will become smaller and thaw earlier in the warm season (Fung et al. ). The increase in evaporation will undoubtedly drastically reduce the availability of water and provide the basis for the occurrence and intensification of drought effects. Runoff is a component of the hydrology cycle that is influenced by many parameters. Certainly, increasing the temperature and decreasing snowfall in the mountainous basins along with the evaporation of the water bodies will also lead to a decrease in the volume of runoff and water reserves and an intensification of the hydrological drought phenomenon.
Nowadays, global warming has significant effects on precipitation and runoff yield and water resources due to the increased concentration of greenhouse gases (Zhang et al. ). Rainfall as a key factor in changing the frequency and range of hydrological cycle has serious consequences in social, economic, and agricultural developments (Zhang et al. ). The average of climatic variables, especially the components of temperature, precipitation, and runoff in the annual or seasonal scale, plays a predominant role in the hydrological cycle. The climate change is usually investigated through the assessment of the average of these parameters, as well (Afshar et al. ).   Afshar et al. (), the annual climatic components of the Kashafrood basin in Iran were assessed in the historical and future periods by using the AR5 models (Afshar et al. ). According to the results, the precipitation component showed a significant decreasing change trend, and the average temperature also indicated a remarkable increase with a confidence level of 90, 99, and 99.9%. On the basis of the obtained results, the average temperature of the basin will increase from 0.56 to 3.3 C, and precipitation will decrease about 10.7% by the end of the 21st century.
The study watershed in this research (Zoshk-Abardeh) is one of the important sources of income for the regional villagers and has a high ecotourism potential in Khorasan Razavi Province, Iran. It is considered as an urban watershed and according to the historical evidence has a high flood potential, as well. Thus, this study aims to assess the impact of climate change on the most important hydroclimatological factors (rainfall, temperature, and runoff) affecting on the ecotourism potential of the watershed.

Study area and data set
The Zoshk-Abardeh Watershed as a sub-watershed of the Kashafrood basin, with an area of 9,225.9 ha, is located in the west of Mashhad in Khorasan Razavi Province, Iran the SUFI-2 algorithm can be narrowed down by identifying a range of parameters that reduce the total uncertainty of the output data. In order to find the optimal parameter uncertainties from prior ranges, the SUFI-2 combines calibration and uncertainty analysis with the minimum number of iterations and the smallest possible prediction uncertainty band (Abbaspour ). A set of parameter ranges is mapped for all sources of uncertainty (parameter, conceptual model, and forcing input). In the SWAT model, the parameters, uncertainties, and statistical analysis have to be computed through a proper likelihood function. In this study, Nash-Sutcliffe (NS) coefficient (Nash & Sutcliffe ) and coefficient of determination (R 2 ), as the likelihood (objective) functions, were employed to calibrate the SUFI-2 algorithm (see equations below):  The P-factor is the percentage of observed records that is bracketed by the 95% uncertainty estimation band (95PPU), while the r-factor based on Equation (3) is the average thickness of the 95PPU band (between the upper and lower boundaries) divided by the standard deviation of observational data (Abbaspour ), as follows: where N, Y Lower , Y Upper , and σ x are, respectively, the total number of the observed data, the lower and the upper limits of the 95PPU, and the standard deviation of the observed data. When 100% of the observed data is bracketed by the 95PPU, the maximum value of P-factor is 1. The lower value of the r-factor indicates a better performance of the model.
The sensitivity analysis was carried out by the SWAT-CUP tool to improve the understanding of the impact of sensitive parameters on runoff discharge. In SUFI-2, the parameter sensitivities are determined using a multiple regression approach that relates the Latin Hypercube-generated parameters to the objective function values using the below equation (Abbaspour ; Memarian et al.
These RCPs were included one mitigation scenario leading to a very low forcing level (RCP2.6), two medium stabiliz-  points. Based on Equation (5), the bias correction of the precipitation divides the projected precipitation by the observed values as follows: where GCM t and GCM p are, respectively, monthly temperature and precipitation of the GCMs output, OBS t and OBS p are, respectively, monthly observed temperature and precipitation, and F t and F p are, respectively, bias corrections for temperature and precipitation at the output points of the GCMs. At the second step, the bias correction is converted from low resolutions to higher resolutions by means of the linear interpolation method. At the final step (Equation where X i and X j are sequential data values, n is the length of time series, and the sgn(X i À X j ) function was defined according to the following equation: when n ! 8, the test statistic S is nearly normally distributed with the mean (E(S) ¼ 0). The variance statistic is computed by the following equation: var ¼ n(n À 1)(2n þ 5) À P m i¼1 t i (i)(i À 1)(2i þ 5) 18 (9) where t i is the number of ties for sample i. The test statistics Z C is computed via the following equation: where Z C follows a standard normal distribution. A positive (or negative) value of Z C indicates an upward (or downward) trend. A significance level of α is also employed for testing whether or not an upward or downward trend is uniform. If Z C appears greater than Zα/2, then the trend is  The hydrograph of monthly observed and simulated flow rates during calibration and validation periods, which is shown in Figure 5, has been used in order to evaluate the efficiency of the model in the base and peak discharge     to the study area (Shafiei et al. ). In this study, Hussein

CONCLUSIONS
The performance evaluation of the SWAT model for runoff simulation was done using the R 2 and NS metrics. The R 2 and NS coefficients during the calibration step were estimated to be 0.52 and 0.47, respectively, while these measures during the validation process were 0.46 and 0.42, respectively. Results showed that the SWAT performance for the simulation of monthly runoff in the Zoshk-   ations, but its trend was not statistically significant at the 5% level. In general, these results indicated that the amount of temperature would follow an increasing tendency, while precipitation and runoff volume would follow a decreasing movement in the Zoshk-Abardeh watershed by the end of the 21st century. The increasing drift of temperature, and in particular the minimum temperature, can affect the evapotranspiration rates as well as future snowfall in the region. Thus, the amount of runoff will be impacted and decreased. Consequently, these hydro-climatic changes will impact the watershed biodiversity. Finally, the investigation of climate change impacts on groundwater, land use, and land cover condition can be suggested for further researches in the Zoshk-Abardeh watershed, as well.