Impact of climate change on surface runoff: a case study of the Darabad River, northeast of Iran

Climate change is one of the major challenges affecting natural ecosystems and various aspects of human life. The effects of global warming on the hydrology and water cycle in nature are very serious, and the quantitative recognition of these effects creates more readiness to deal with its consequences. In the present study, the 2006 – 2100 period is predicted based on the statistical downscaling model (SDSM). Finally, the effects of climate change on the hydrological conditions in the Darabad watershed are simulated using the soil and water assessment tool (SWAT) model. The SWAT model calibration is done based on the SUFI-2 algorithm, and the effective and optimal parameter is identi ﬁ ed. The results of the study, while con ﬁ rming the ef ﬁ ciency of both SDSM in climate simulations and SWAT in hydrological simulation, showed that the increase in precipitation and temperature is probably in future climate conditions for the 2010 – 2040 period. The surface ﬂ ow and runoff at the watershed area during the observation period (1970 – 2010) is 0.29 m 3 /s, but this value for the predicted period with regard to climate change in the RCP 2.6, RCP 4.5, and RCP 8.5 scenarios is equal to 0.43, 0.44, and 0.45 m 3 /s. The results of research, while highlighting the importance of effects of climate change, make it essential to apply them for proper management in order to adapt to climate change in the future policies of the Darabad watershed management. under RCP 2.6, RCP 4.5, and RCP 8.5 scenarios is increased by 45, 49, and 50%, respectively.


INTRODUCTION
Today, numerous studies have been conducted in Iran and around the world regarding the potential impacts of climate change on water resources, including the impact on water quantity, hydrology, and water demand. Using global data available from the last century, it was found that the global runoff is increased by 4% with the earth temperature rising by 1% (Vera et al. ; Narsimlu et al. ).
The increase in temperature, severe droughts, rise in sea levels, exacerbation of climate phenomena, the melt of mountain glaciers, and decrease in polar ice glaciers are among the effects of climate change. Climate change leads to dramatic changes in hydrosphere, biosphere, and ecosphere, with hydrological consequences such as water stresses.
Water stresses are caused by a set of factors, such as rising temperatures, droughts, and changes in water consumption pattern (Raziei et al. ). Meanwhile, the temperature has a greater role in creating water stress than the other climate elements due to its effects on the hydrological cycle (precipitation, evapotranspiration, and interception) (Huo & Li ). Therefore, with the continuation of global climate change and its impact on water resources, it is expected to affect more than 4 billion inhabitants of the earth. In fact, the climate of the planet is changing, and global warming is taking place (Narsimlu et al. ; Klein ). According to the report, over the last century, the average annual temperature of the planet has increased 0.3-0.6 C due to greenhouse gas emissions and is expected to increase about 1-3.5 C by 2100 (Houghton et al. ).
The process of climate change, especially temperature and precipitation changes, is the most important issue in the field of environmental sciences. This phenomenon is of increasing importance due to its scientific and applied dimensions (environmental and socioeconomic effects), as human systems dependent on the climate elements such as agriculture, industry and the like are designed based on the stability and sustainability of the climate (Huo & Li ). Also, the increase in air temperature due to climate change will change the precipitation from snow to rain, and change the amount and intensity of runoff, ratio of snow to rain, and amount of water stored in the snow mass. As a result, the amount of precipitation as snow in the middle and low heights of mountainous regions is reduced and increased in the highlands, which ultimately affects the peak daily flow rate (Rahimi et al. ).

The Intergovernmental Panel on Climate Change
(IPCC) suggests that in recent decades, climate change has globally changed the hydrological characteristics, so that precipitation and surface flow are higher in the upper and middle latitudes and are lower in the lower latitudes, and the probability of confrontation with extreme climatic events such as floods and droughts will be increased (Narsimlu et al. ).
In order to investigate the effects of climate change on different systems in upcoming periods, the future climate variables should be initially simulated. There are several methods for simulating climate variables in the upcoming periods, and the most valid method is the use of climate outputs of atmosphere-ocean general circulation models (AOGCM). In these models, efforts have been made to simulate the processes that are effective on the climate and the climate situation to be predicted for the upcoming years  during the upcoming period, the amount of precipitation is decreased and instead, the intensity of precipitation in the upcoming period, and finally, the average monthly precipitation shows an increasing trend.
In general, by studying the effects of climate change, we conclude that the climate change performance on the runoff greatly varies, so that some cases show the precipitation and runoff decreases and some others show the precipitation and runoff increases. However, in all reports, there is an evident increase in temperature in the upcoming period. In general, in this research, the study area is located upstream of the Tehran watershed, and the study of climate and its effects on runoff are very important. This is especially manifested when there is a large population downstream of the watershed. In fact, flood and drought are more likely to happen in these situations.
Generally, the objectives of the research are as follows: (1) evaluation of the performance of CanEMS2 climate change model and SDSM, (2) evaluation of the performance of the SWAT model in the precipitation-runoff simulation,

Datasets and input data
To simulate rainfall-runoff and climate change in this study, we need daily time series of climate data and maps. Daily data include rainfall, maximum temperature, minimum temperature, relative humidity, wind speed, flow rate, and amount of sunlight. Table 1 and Figure 1 show the features of the stations used. In order to check and control the input data into the precipitation-runoff model in terms of accuracy, precision, and adequacy, we began by forming the time series of the precipitation data, maximum and minimum temperatures, wind speed, and relative humidity and then correcting and reconstructing the data gaps and outliers using the methodology proposed by the American Water Resources Association (AWRA) (Ahmadi et al. a, b).
Using the AWRA-proposed methodology, the mean, upper bound, lower bound, and standard deviation of the data were evaluated. The coefficient of determination of the outliers was calculated using the tables provided by the AWRA at a significance level of 10% considering normal distribution.
The so-called run-test was performed to investigate the homogeneity of the time series data. For this purpose, the data points were sorted in ascending or descending order.
Then, the median was identified, and the state of each data point with respect to the median (larger or smaller than the median) was signified by a particular sign. Finally, the upper and lower bounds of the existing ranges are calculated using the related table (Ahmadi et al. a, b).
To prepare the precipitation-runoff model, it is necessary to use validated data. Any error in the input data of the model can be an important factor in the error in the estimated model and simulated flow parameters.
The required meteorological data are precipitation, minimum and maximum daily temperature, radiation, The emission scenarios are used to predict the concentration of greenhouse gases in the atmosphere. The IPCC used the RCP scenario as representatives of various trajectories of greenhouse gas concentrations for compiling the fifth report. In this study, the RCP (2.6, 4.5, and 8.5) scenarios are cited. The RCP 8.5 scenario encompasses the highest rate of increased greenhouse gases and the resulting radiative forcing, which will be progressed in this scenario without adopting any climate consequence mitigation and prevention policies, while the RCP 2.5 scenario includes the lowest rate of increased greenhouse gases and radiative forcing (Mirdashtvan et al. ).
The output of the general circulation model is downscaled by two statistical and dynamic methods (Samadi et al. ). In this research, the SDSM as a statistical method is used for the downscaling. The SDSM is an auxiliary tool for assessing the effects of local climate change which is developed by Wilby et al. (). This model uses  independent observational data to assess the relationship between local downscaled variables and atmospheric upscaled variables (Wilby & Dawson ).
In order to simulate the future-period precipitation and maximum and minimum temperatures, one should use the SDSM to develop multivariate regression equations relating the mentioned variables to the 26 parameters of the NCEP.

Error and uncertainty analysis
In order to determine the reliability of the simulated data, the mean and two statistical indicators, namely NSE (Equation (1)) and coefficient of determination (R 2 ) (Equation (2)) and mean bias error (MBE) (Equation (3)

RESULTS
Error and uncertainty analysis of climate model and hydrologic model In the simulation of the present and future maximum and minimum temperatures, the air temperature 2 m was among the most important parameters contributing to the simulation of the maximum and minimum temperatures. By examining the error of the SDSM in the CanESM2 scenario for the precipitation parameter (Figure 3(a)   Following a trial and error approach for selecting the effective parameters (Table 3) Based on Table 4, it can be stipulated that fairly good results were obtained during the calibration and evaluation periods across the study area. Moreover, the value of P-factor was found to be larger than 0.5, indicating the appropriate performance of the model in simulating the precipitation-runoff data. By examining the form (Figure 4(a)) in the calibration period, as can be seen, the model has a proper function and it can be said that only in the maximum and minimum values, it has made errors and, as can be seen in Figure 4( Table 4, it can be stipulated that fairly good results were obtained during the calibration and evaluation periods across the study area. Moreover, the value of P-factor was found to be larger than 0.5, indicating appropriate performance of the model in simulating the precipitation-runoff data.

Precipitation and temperature modeling in upcoming period
The minimum and maximum precipitation and temperature were simulated using the SDSM under three RCP (2.6, 4.5, and 8.5) scenarios for the upscaled CanESM2 model in the three periods, 2010-2040, 2041-2070, and 2071-2100, and the results are presented in Figure 5 and Table 5.
In addition, Figure 5 shows the simulation for each station in the assessment period (historical data of the    summer and the rise in the fall, and it shows different behaviors in spring and winter, as shown in Table 5. Most changes are observed in all periods in the fall and winter seasons, and the variations in each three periods and each scenario can be seen in Figure 4 and Table 5. Reviewing the maximum temperature and minimum temperature in Figure 6 and Table 6 shows that the highest increasing trend of temperature occurs in the winter and summer temperatures. In all periods, the highest temperature rise is during the period 2071-2100. The changes in all seasons for all the periods and all the scenarios can be seen in Table 6.

Runoff modeling in the upcoming period
To simulate the runoff in the upcoming period, after selecting the effective parameters for the runoff simulation in the SWAT_CUP model and entering the SWAT model, the    Table 7. In general, it can be stated that there is an increase in precipitation in the spring and summer and there is a decrease in the runoff in summer.

DISCUSSION
In general, the goals were accomplished in this study. The purpose of this study was to investigate the performance of SDSM and CanESM2 model, evaluate the performance of the SWAT model in the precipitation and runoff simulation, and finally, to study the climate change process and its effect on runoff. In the next section, the questions asked in the introduction are answered, respectively.
In general, the changes were considered regarding the precipitation, maximum temperature, and minimum temp- increase in Northern Tehran station in RCP 2.6 and 4.5 scenarios. In general, in the areas upstream of the study area, which includes the Fashem, Roodbar Ghasran, and Roodak stations, an increase in precipitation is observed between 6 and 39%, while in the downstream areas, there is an increase of 1-52%. Table 8  results of other studies. In general, it can be stated that in the period 2010-2100, the maximum temperature will vary from 1 to 6 C, and the detailed specifications can be seen in Table 9.
By reviewing the runoff variations in the upcoming period (2010-2040) under the three RCP 2.6, RCP 4.5, and RCP 8.5 scenarios, an increasing trend is found equal to 0.14, 0.15, and 0.16 in the runoff of the upcoming period, respectively, so that the highest changes in all three scenarios were observed in the spring. In fact, it can be stated that in the spring, winter, and summer, the increase in runoff was observed, and in the fall, there was a decrease in the runoff. In general, it can be stated that in the period 2010-2040 in three RCP 2.6, RCP 4.5, and RCP 8.5

CONCLUSION
It should be noted that the Darabad study area is located upstream of the big Tehran watershed in a mountainous area. This region has snowfall at heights during winter.
The results of this study showed that the minimum tempera-