https://orcid.org/0000-0002-1575-0387
Abstract
Climate change (CC) will increase the intensity of extreme phenomena such as drought and flood in arid and semi-arid regions. This will cause the water supply of these areas to become very difficult in times of crisis. This study identifies sub-basins with high flood potential in the baseline period (1982–2005) and the future period (2025–2048) in the Hablehroud basin, north-central Iran. It uses the soil and water assessment tool (SWAT) and 23 coupled model intercomparison project 5 (CMIP5) general circulation models (GCMs). It estimates the instantaneous peak flow (IPF) and uses a flood index (FI) to determine the contributions of each sub-basin to the floods. The rainfall of the basin will increase by 11.5% under representative concentration pathway (RCP) 4.5 and 12.6% under RCP 8.5. The minimum daily temperature (Tmin) of the basin will increase by 0.8 °C under RCP 4.5 and 1.1 °C under RCP 8.5 in the future period. In addition, the maximum daily temperature (Tmax) will rise by 1 °C under RCP 4.5 and 1.2 °C under RCP 8.5. Moreover, basin runoff will increase by 6.4% under RCP 4.5 and 11.6% under RCP 8.5. The results indicate that the central and southern sub-basins made the most significant contribution to floods in the baseline period, while the eastern sub-basins will make the most considerable contribution to future floods.
HIGHLIGHTS
Using the SWAT model to simulate daily rainfall-runoff and determine instantaneous peak flow (IPF), and 23 CMIP5 GCMs to predict future climatic parameters.
Using a flood index (FI) to determine the contributions of each sub-basin to floods.
The integration of GCMs through the k-nearest neighbors (KNN) algorithm and evaluation of climate change (CC) and sub-basin contributions to floods.
INTRODUCTION
Floods are serious natural disasters that can impose economic and social losses. Flood damage has increased in recent decades (Botzen et al. 2019). This indicates the increased frequency and intensity of floods (Adib et al. 2019). Climate change alters the average quantities of climatic variables, such as the temperature and rainfall in different regions (Khazaei et al. 2019). According to the Intergovernmental Panel on Climate Change (IPCC), extreme events such as intensive floods are growing fast. Heavy floods caused by CC have imposed numerous damage across the world in recent decades. These impacts have been larger in arid climates (Vaghefi et al. 2019). Therefore, it is necessary to identify and prioritize flood-prone regions in terms of flood control, planning, and the comprehensive management of basins (Saghafian et al. 2008). In other words, different parts of a basin have different effects on the flow during a flood. Some regions intensify instantaneous floods due to soil and hydrological conditions, and the flow significantly rises in such regions (Khazaei et al. 2012). The soil and water assessment tool (SWAT) has been used as a semi-distributed model for determination of CC effects on the value of runoff (Zolin & Rodrigues 2015). General circulation models (GCMs) can accurately simulate all the climatic processes at the global or continental scale (Adib et al. 2021a, 2021b). To prevent and control floods, it is necessary to evaluate their sources (i.e., sub-basins) so that proper managerial measures could be implemented (i.e., reducing impermeable regions, increasing permeability, and reviving land uses to minimize runoff) (Kundzewicz & Takeuchi 1999). Saghafian & Khosroshahi (2005) introduced the unit flood response method (UFRM) to identify flood-inducing regions. The regions with a high potential of inducing floods are identified so that modification can be implemented at hazardous levels. Details of other research on the subject of the current study are given below.
Almasi & Soltani (2017) studied CC effects on flood frequency (FF) in the Bazoft watershed, central Iran, using four CMIP3 GCMs in the baseline period of 1971–2000, the future period of 2020–2049, and the far future period of 2071–2100. They used the distributed WetSpa model and estimated the daily instantaneous peak flow (IPF) through the Sangal, Fill-Steiner, and Fuller methods. They concluded that the instantaneous peak flow would reduce in all future periods.
Maghsood et al. (2019) studied influences of CC on the FF and flood source areas in the Talar watershed using 20 GCMs based on scenarios representative concentration pathway (RCP) 2.6 and RCP 8.5. They applied SWAT and estimated the daily IPF by the Sangal, Fill-Steiner, Fuller, and slope-based methods. The instantaneous peak was concluded to increase in the future period.
Babaeian et al. (2021) introduced four different adaptation pathways to reduce the uncertainty of GCMs for the agricultural management of the Hablehroud basin. They applied five GCMs under RCP 4.5 and RCP 8.5. Furthermore, they integrated SWAT and adaptation pathways. It was found that changes in the cropping patterns induced the highest uncertainties.
Esmaeili-Gisavandani et al. (2021) employed five continuous rainfall-runoff models, including SWAT, IHACRES, HBV, SMA, and AWBM, for determination of runoff in the Hablehroud River. They showed that SWAT had the highest accuracy in the hydrological simulation of the Hablehroud basin.
Lotfirad et al. (2021) distinguished the impacts of CC on hydrological drought in the Hablehroud basin. They used 23 GCMs for climate projecting, the k-nearest neighbors (KNN) method for combining GCMs, and the IHACRES model for simulating the runoff. Their results indicate occurrence of short- and medium-term droughts in the Hablehroud basin in the future.
The purpose of this research is to estimate IPF in the future period. Lotfirad et al. (2021) focused on hydrological drought, while Esmaeili-Gisavandani et al. (2021) examined rainfall-runoff models and Babaeian et al. (2021) determined changes in cropping patterns and crop planting dates in the future period. Also, this study utilized a daily SWAT model instead of the monthly SWAT model that was used by previous studies.
Overall, studying the impact of CC on flood risk using an integrated climate-hydrological model is necessary. The objective of this study is to estimate IPFs in the Hablehroud basin and to determine the contribution of different sub-basins in the flooding of the basin under the influence of CC. The potential impacts of CC on FF patterns and flood source area in the Hablehroud basin were evaluated too. The novelties of this study include the integration of GCMs through the KNN algorithm and the assessment of CC effects and sub-basin contributions to floods. This study was conducted in four steps:
- 1
Hydrological modeling at the daily scale using SWAT;
- 2
Climatic simulation by integrating 23 CMIP5 GCMs in the baseline and future periods under scenarios RCP 4.5 and RCP 8.5 and statistical downscaling;
- 3
Estimating IPF in the baseline and future periods and comparing flood frequencies; and
- 4
Calculating the flood index (FI) for each sub-basin in the baseline and future periods.
MATERIALS AND METHODS
Case study
The Hablehroud basin lies in the north of Iran, with an arid and semi-arid climate (Figure 1). The Hablehroud River basin suffers significant flood damage every year. These floods occur at time intervals in spring, summer, and autumn, with different sources and flows (Lotfirad et al. 2021). The Hablehroud River feeds the Garmsar Plain as one of the largest agricultural plains in Iran (Esmaeili-Gisavandani et al. 2021). Indiscriminate flooding of this river during planting and harvesting seasons causes more damage than other seasons (Babaeian et al. 2021). In the Hablehroud basin, the river length is 119.5 km and basin area is 3,261 km2, with average annual minimum/maximum daily temperatures (Tmin/Tmax), rainfall, and runoff of 6.4 °C, 14.8 °C, 2,356 mm, and 7.5 m3/s, respectively (Lotfirad et al. 2021).
Data analysis
This study used data of a number of climatic variables at the daily scale, including rainfall, Tmin, Tmax, wind speed, relative humidity, and sunshine hours, during 1982–2005. Rainfall and flow discharge data were prepared from the Iranian Ministry of Energy and other meteorological data such as temperature were obtained from the Iran Meteorological Organization (IMO).
The attribute table for land use in the Hablehroud basin
| Class Name | Area (km2) | Percentage | Type |
|---|---|---|---|
| URBN | 3.27 | 0.10 | Urban |
| AGRL | 40.94 | 1.26 | Agricultural land-generic |
| BARR | 837.89 | 25.74 | Barren |
| GRAS | 2,307.61 | 70.88 | Grassland |
| ORCD | 28.31 | 0.87 | Orchard |
| FRST | 37.82 | 1.16 | Forest-mixed |
| Class Name | Area (km2) | Percentage | Type |
|---|---|---|---|
| URBN | 3.27 | 0.10 | Urban |
| AGRL | 40.94 | 1.26 | Agricultural land-generic |
| BARR | 837.89 | 25.74 | Barren |
| GRAS | 2,307.61 | 70.88 | Grassland |
| ORCD | 28.31 | 0.87 | Orchard |
| FRST | 37.82 | 1.16 | Forest-mixed |
Location of the Hablehroud basin.
(a) Land use, (b) Slope and (c) Soil map of the Hablehroud basin.
Evaluation of Climate Change
Weighting the GCM outputs
Schematic of applied methodology in this study.
and 
are the absolute error of GCMs for temperature and rainfall, respectively. 
and 
are the mean temperature and precipitation from 1982 to 2005. Indices B, m, G and O are the baseline period, the month, the GCM and observed data, respectively.
and 
are the weights of model 
in month 
for precipitation and temperature, respectively.
and 
are the monthly mean of temperature and rainfall for future period. 
and 
are changes of temperature and rainfall (Zareian et al. 2015).Stochastic downscaling
The GCM outputs represented only the temperature and rainfall variations due to the large scale of the GCMs (Doulabian et al. 2021). Hence, weather generator downscaling techniques needed to be adopted to generate daily meteorological data. The base of GCM is semi-empirical distribution functions that could generate future dry and wet periods, and generated data were validated by the KS, t, and F tests (Semenov et al. 1998). For downscaling, this study used LARS-WG as a meteorological generator. After calculating the changes of temperature and precipitation between the future period and the base period (calculated by selected GCM by weighting method), the scenario file was imported into LARS-WG software, and meteorological data were generated.
In the other words, the Tmin, Tmax, and rainfall changes under scenarios RCP 4.5 and RCP 8.5 were introduced to the LARS-WG model, generating daily data based on the observational data of the five regions (Figure 1(a)) in the baseline (1982–2005) and future (2025–2048) periods. Another advantage of ensemble 23 GCMs is the improving quality of the generated meteorological data by the LARS-WG model. Different GCMs considered different climatic parameters in the ocean and atmosphere and they underestimate or overestimate meteorological phenomena such as temperature and precipitation. The ensemble of different GCMs by a weighting method and by introducing the Tmin, Tmax, and precipitation changes under scenarios RCP 4.5 and RCP 8.5 to the LARS-WG model fix a problem of the LARS-WG model (the LARS-WG model generates more ‘flatter data’ and the generated flood extremes may be smaller than the observed data). The results of this study demonstrate this matter (Figures 4, 6, 8 and 9).
SWAT model
The SWAT model is a semi-distributed and continuous hydrological model to simulate streamflow, sediments, and agricultural chemical yields (Srinivasan et al. 1998). The smallest spatial unit is the hydrologic response unit (HRU) and water balance components are determined separately for each HRU, then routed for each sub-basin and the entire basin (Gassman et al. 2007).
Execution of SWAT model
To initially configure the model, the Arc SWAT 2012 extension in ArcGIS V.10.3 was utilized. Digital elevation layers were applied, drawing the stream network and dividing the basin with a total area of 3,234 km2 into 17 sub-basins. The HRUs were created by superposing the land use maps and developing three slope classes (including 0–20% for flat surfaces, 20–40% for medium slope surfaces, and >40% for sloped surfaces). To reduce the number of HRUs, thresholds of 10% (use) and 5% (slope and soil) were applied (Figure 2). Moreover, crops were split into the dominant crops of the region (with codes), including agricultural crops (i.e., wheat, barley, alfalfa, and potato) and horticultural crops (i.e., apple, pistachio, and walnut). Agricultural data of the Hablehroud basin were prepared from the database of the Ministry of Agriculture Jihad. Irrigated wheat, barley, potatoes, apples, walnuts and pistachios are the main products of the Hablehroud basin. Information related to fertilizer rate, irrigation rate, planting, harvesting and irrigation efficiency was entered into the SWAT model from the sub-basin data module of the model. Groundwater is used for irrigate agricultural products in this basin, with an irrigation efficiency of 0.45.
To perform a more realistic simulation, agriculture management data were incorporated into the model. Finally, SWAT was executed at the daily scale for a 24-year period, with three years (1982–1984) considered for warming up the model. The period of calibration of the SWAT model is 1985–2001 and the period of validation is 2002–2005.
SWAT calibrating and evaluating
Calibrating and evaluating the SWAT model was accomplished by using the SWAT-CUP Premium software. SWAT-CUP Premium can perform calibration and validation through the SWAT Parameter Estimator (SPE) and PSO programs. The present study adopted the SPE algorithm to perform the sensitivity analysis of the parameters and calibrate and validate the SWAT. The runoff records of the Bonkuh hydrometric station were utilized for the calibration and validation of the SWAT model. A calibration period of 1985–2001 and a validation period of 2002–2005 were considered. At least 70% of the total time period should be selected for calibration and the rest for verification. Different states were evaluated (in these cases, the calibration time period was more than 70% of the time period) and the calibration time period with the highest performance criteria (NSE, R2, RMSE) was chosen as the best calibration time period.
The performance evaluation (accuracy testing) of the SWAT model was performed using the coefficient of determination (R2), Nash–Sutcliffe coefficient (NSE), ARILCI-factor, and PCI-factor. Larger R2 and NSE values (closer to 1) represent greater consistency between the simulation values and observational data (Nash & Sutcliffe 1970). The PCI-factor is the percentage of data restricted in the estimate uncertainty level of 95% (that is, the degree at which data are restricted at a level of 95%) and can vary up to 100%, at which the entire observed data are ideally restricted at 95%. The ARILCI-factor represents the strength of the model and is calculated based on the ratio between the 95% band thickness to the standard deviation of the observed data. The ARILCI-factor should be ideally close to 1 (<1).
Flood frequency and instantaneous peak flow estimation
An analysis of the FF and determining the return period are essential in the management of flood control projects (Silveira et al. 2000). FF analysis is typically performed based on the annual peak flow. Considering the characteristics of the Hablehroud basin, this study obtained the peak flow by the empirical Fuller (Fuller 1914), Sangal (Sangal 1983), Fill-Steiner (Chen et al. 2017), and slope-based (Chen et al. 2017) methods. These methods calculate IPF through daily flow data. The best IPF estimation method was selected by performance criteria, such as NSE, RMSE, and R2, in the baseline period (Moriasi et al. 2007; Farajpanah et al. 2020). FF was estimated for return periods of 2–200 years based on the SWAT simulations of runoff in the Hablehroud basin and sub-basins.
Flood index in sub-basins
is the IPF difference between the sub-basins and the basin, while A is the sub-basin area (km2) (Saghafian et al. 2008).Figure 3 illustrates the methods used in the current study to identify flood-prone sub-basins in the Hablehroud basin in the north of Iran.
RESULTS AND DISCUSSION
CC based on GCM outputs
Monthly long-term average (a) Tmin (b) Tmax and (c) rainfall in baseline and future.
Monthly long-term average (a) Tmin (b) Tmax and (c) rainfall in baseline and future.
By evaluating changes of Tmin, Tmax and rainfall in the future periods at different sub-basins, it is observed that the largest rises in Tmin and Tmax occurred to be 1.24 °C and 1.57 °C, respectively, in the south of the basin (sub-basin 1) under scenario RCP 4.5. According to the results, rainfall increased in the southeast (sub-basins 2, 4, and 5), while the west (sub-basins 1 and 3) experienced rainfall decreases.
SWAT calibration, validation, and sensitivity analysis
Sensitivity of SWAT parameters
| Rank | Name | Description | Min | Max | Fitted | t-stat | P-Value |
|---|---|---|---|---|---|---|---|
| 1 | v_CH_K2.rte | Effective hydraulic conductivity in main channel alluvium | 0 | 150 | 65.35 | 8.38 | 0.00 |
| 2 | r_CN2.mgt | SCS runoff curve number | −0.4 | 0.4 | 0.01 | −6.13 | 0.00 |
| 3 | v_RCHRG_DP.gw | Deep aquifer percolation fraction | 0 | 1 | 0.69 | 5.54 | 0.00 |
| 4 | v_GW_DELAY.gw | Groundwater delay (days) | 0 | 200 | 158.14 | 5.24 | 0.00 |
| 5 | v_ALPHA_BF.gw | Baseflow alpha factor (days) | 0 | 1 | 0.36 | −4.05 | 0.00 |
| 6 | v_SFTMP.bsn | [OPTIMAL] Snowfall temperature | −5 | 5 | 2.00 | −2.51 | 0.01 |
| 7 | v_SLSUBBSN.hru | Average slope length | 10 | 150 | 28.18 | 2.08 | 0.03 |
| 8 | r_SOL_AWC.sol | Available water capacity of the soil layer | −0.5 | 0.5 | −0.25 | 2.06 | 0.04 |
| 9 | v_MSK_CO2.bsn | Calibration coefficient used to control impact of the storage time constant for low flow | 0 | 10 | 0.29 | 1.99 | 0.04 |
| 10 | v_CH_N2.rte | Manning's ‘n’ value for the main channel | 0 | 0.3 | 0.04 | −1.63 | 0.10 |
| 11 | v_SHALLST.gw | Initial depth of water in the shallow aquifer (mm) | 1,000 | 3,000 | 2,133 | −1.21 | 0.22 |
| 12 | r_SOL_BD.sol | Moist bulk density | −0.4 | 0.4 | 0.12 | −1.15 | 0.25 |
| 13 | r_SOL_K.sol | Saturated hydraulic conductivity | −0.8 | 0.8 | 1.10 | −1.07 | 0.28 |
| 14 | v_OV_N.hru | Manning's ‘n’ value for overland flow | 0 | 0.8 | 0.47 | 1.03 | 0.30 |
| 15 | v_SMFMX.bsn | Maximum melt rate for snow during year | 0 | 10 | 5.11 | −0.99 | 0.32 |
| 16 | v_MSK_CO1.bsn | Calibration coefficient used to control impact of the storage time constant for normal flow | 0 | 10 | 2.92 | −0.89 | 0.37 |
| 17 | v_GW_REVAP.gw | Groundwater ‘revap’ coefficient | 0.02 | 0.2 | 0.14 | 0.86 | 0.38 |
| 18 | v_ESCO.hru | Soil evaporation compensation factor | 0 | 1 | 0.23 | −0.73 | 0.46 |
| Rank | Name | Description | Min | Max | Fitted | t-stat | P-Value |
|---|---|---|---|---|---|---|---|
| 1 | v_CH_K2.rte | Effective hydraulic conductivity in main channel alluvium | 0 | 150 | 65.35 | 8.38 | 0.00 |
| 2 | r_CN2.mgt | SCS runoff curve number | −0.4 | 0.4 | 0.01 | −6.13 | 0.00 |
| 3 | v_RCHRG_DP.gw | Deep aquifer percolation fraction | 0 | 1 | 0.69 | 5.54 | 0.00 |
| 4 | v_GW_DELAY.gw | Groundwater delay (days) | 0 | 200 | 158.14 | 5.24 | 0.00 |
| 5 | v_ALPHA_BF.gw | Baseflow alpha factor (days) | 0 | 1 | 0.36 | −4.05 | 0.00 |
| 6 | v_SFTMP.bsn | [OPTIMAL] Snowfall temperature | −5 | 5 | 2.00 | −2.51 | 0.01 |
| 7 | v_SLSUBBSN.hru | Average slope length | 10 | 150 | 28.18 | 2.08 | 0.03 |
| 8 | r_SOL_AWC.sol | Available water capacity of the soil layer | −0.5 | 0.5 | −0.25 | 2.06 | 0.04 |
| 9 | v_MSK_CO2.bsn | Calibration coefficient used to control impact of the storage time constant for low flow | 0 | 10 | 0.29 | 1.99 | 0.04 |
| 10 | v_CH_N2.rte | Manning's ‘n’ value for the main channel | 0 | 0.3 | 0.04 | −1.63 | 0.10 |
| 11 | v_SHALLST.gw | Initial depth of water in the shallow aquifer (mm) | 1,000 | 3,000 | 2,133 | −1.21 | 0.22 |
| 12 | r_SOL_BD.sol | Moist bulk density | −0.4 | 0.4 | 0.12 | −1.15 | 0.25 |
| 13 | r_SOL_K.sol | Saturated hydraulic conductivity | −0.8 | 0.8 | 1.10 | −1.07 | 0.28 |
| 14 | v_OV_N.hru | Manning's ‘n’ value for overland flow | 0 | 0.8 | 0.47 | 1.03 | 0.30 |
| 15 | v_SMFMX.bsn | Maximum melt rate for snow during year | 0 | 10 | 5.11 | −0.99 | 0.32 |
| 16 | v_MSK_CO1.bsn | Calibration coefficient used to control impact of the storage time constant for normal flow | 0 | 10 | 2.92 | −0.89 | 0.37 |
| 17 | v_GW_REVAP.gw | Groundwater ‘revap’ coefficient | 0.02 | 0.2 | 0.14 | 0.86 | 0.38 |
| 18 | v_ESCO.hru | Soil evaporation compensation factor | 0 | 1 | 0.23 | −0.73 | 0.46 |
r and v are codes that determine the variations of the parameter; v indicates that the parameter should be set to a new value, while r suggests the multiplication of the parameter by the new value.
Comparison between observed and simulated streamflow. Please refer to the online version of this paper to see this figure in colour: http://dx.doi.org/10.2166/ws.2022.271.
Comparison between observed and simulated streamflow. Please refer to the online version of this paper to see this figure in colour: http://dx.doi.org/10.2166/ws.2022.271.
According to Moriasi et al. (2007), SWAT has good performance in the simulation of flow in a basin and can be used to project runoff in the future period. Moreover, Esmaeili-Gisavandani et al. (2021) stated that the SWAT model with NSE = 0.75 had higher performance than other rainfall-runoff models in the Hablehroud basin. Therefore, to simulate flow in the future period of 2025–2048, the projected Tmin, Tmax, and rainfall based on scenarios RCP 4.5 and RCP 8.5 were introduced to the SWAT model.
Table 2 presents the sensitive analysis of the SWAT model parameters. Eighteen (18) parameters were ranked by sensitive analysis procedure. Table 2 shows 18 parameters in order of their effect on IPF. Parameters 1 to 6 have a very high effect on IPF (at a significance level of 99%, 
and P-value < 0.01), parameters 7 to 9 have a high effect on IPF (at a significance level of 95%, 
and P-value < 0.05) and parameter 10 has a relatively high effect on IPF at a significance level of 90%, 
and P-value < 0.1).
Figure 5 illustrates the comparison between observed and simulated daily flow discharge by the SWAT model. The quantitative values of NSE, ARILCI-factor, and PCI-factor show the high quality of the simulated daily flow discharge by the SWAT model. These values are unprecedented compared to the results obtained by previous studies. Figure 5 shows that the increasing and decreasing trends of the observed and simulated flow discharges are the same. Due to the large number of daily flow discharges, a number of simulated daily flow discharges are greater than observed daily flow discharges and a number of simulated values are less than observed values.
Estimation of IPF
To estimate the IPF in the Hablehroud basin, empirical methods, including the Fuller, Sangal, slope-based, and Fill-Steiner methods, were applied. According to Table 3, the Fill-Steiner method had acceptable performance in the estimation of the IPF consistent with the IPF recorded at the Bonkuh station. Therefore, the IPF of the future period was estimated by the Fill-Steiner method. Based on the Kolmogorov–Smirnov goodness of fit test module in EasyFit, the log-Pearson III distribution had the highest fitness with the IPF of the baseline period. Also, the least square error method illustrated that the log-Pearson III is the best empirical probability distribution for estimating the annual peak flow. Due to the high coefficient of skewness of the annual peak flow data, in most watersheds of the world the log-Pearson III governs probability distribution of IPF (Bobée 1975).
Results of IPF estimation methods from daily discharge
| Index | Fuller | Sangal | Slope-based | Fill-Steiner |
|---|---|---|---|---|
| R2 | 0.52 | 0.46 | 0.48 | 0.65 |
| NSE | 0.50 | 0.34 | 0.14 | 0.58 |
| RMSE (m3/s) | 59.70 | 68.54 | 78.45 | 54.81 |
| Index | Fuller | Sangal | Slope-based | Fill-Steiner |
|---|---|---|---|---|
| R2 | 0.52 | 0.46 | 0.48 | 0.65 |
| NSE | 0.50 | 0.34 | 0.14 | 0.58 |
| RMSE (m3/s) | 59.70 | 68.54 | 78.45 | 54.81 |
IPF in a return period of 2 to 200 years in the baseline and future periods at Bonkuh station.
IPF in a return period of 2 to 200 years in the baseline and future periods at Bonkuh station.
Impacts of climate change on IPF
Average IPF in the baseline period and percentage changes in IPF in the future period compared to the base period, under two scenarios: RCP 4.5 and RCP 8.5.
Average IPF in the baseline period and percentage changes in IPF in the future period compared to the base period, under two scenarios: RCP 4.5 and RCP 8.5.
Long-term monthly runoff of the Hablehroud basin in the baseline and future periods.
Long-term monthly runoff of the Hablehroud basin in the baseline and future periods.
Figure 8 shows that monthly runoff will increase in June, July, August AND September. In these months, temperature will increase while changes of rainfall is negligible in the future periods (Figure 4). An increase in temperature causes drying of the soil and reduction of vegetation. This matter will reduce soil permeability and will increase IPF. In recent years (2001, 2002, 2012, 2014, 2015, 2017, 2018 and 2019), summer rains have caused large and destructive floods in the north of Iran. IPFs and the number of these floods are increasing and this fact will increase monthly runoff of the Hablehroud basin in summer.
Impacts of climate change on flood index in sub-basins
The effect of CC on the flooding index of sub-basins in the return period of 2 and 200 years. Very Low (<0.1), Low (0.1–0.2), Moderate (0.2–0.4), High (0.4–0.6), Very High (>0.6).
The effect of CC on the flooding index of sub-basins in the return period of 2 and 200 years. Very Low (<0.1), Low (0.1–0.2), Moderate (0.2–0.4), High (0.4–0.6), Very High (>0.6).
CONCLUSION
This study investigated the CC effects on floods and flood potential in the Hablehroud basin as the most important basin in the north of Iran. To study CC effects on the temperature and rainfall in the basin, 23 CMIP5 GCMs were executed under scenarios RCP 4.5 and RCP 8.5 for the future period of 2025–2048. Since Esmaeili-Gisavandani et al. (2021) suggested that SWAT was the best model in the runoff estimation of the Hablehroud basin, the present study adopted SWAT to simulate runoff in the Hablehroud basin at the daily scale. Tmin and Tmax were projected to increase in all months in the future period.
The obtained results are similar to results obtained by Lotfirad et al. (2021) in this basin. These results would increase water demand of crops and evapotranspiration in the Garmsar Plain downstream of the Hablehroud basin.
Furthermore, it was observed that IPF would increase in the future under RCP 4.5 and RCP 8.5 compared with the baseline period. Through the SWAT simulation of the daily flow in the baseline period, the IPF of the Hablehroud basin was calculated using empirical methods. The Fill-Steiner method outperformed the other techniques in the estimation of IPF (Almasi & Soltani 2017). Therefore, the IPF of the future period was estimated by the Fill-Steiner method. According to the monthly flow variations of the basin in the baseline and future periods, IPF would be larger in the future period than in baseline period (under RCP 4.5 and RCP 8.5). The largest IPF was projected to occur in spring (from March to May) in the future period due to the temperature rise and snow melting in the Hablehroud basin. The IPF was calculated for the baseline and future periods at the return periods of 2–200 years in EasyFit. The future IPF of the Hablehroud basin showed an increasing trend compared to the baseline period at return periods of 2–200 years.
The central and southern sub-basins were more prone to floods in the baseline period, while the eastern and southern sub-basins had higher flood proneness in the future period. This matter is completely consistent with the climatic modeling results suggesting increased rainfall in the eastern sub-basins in the future period.
The results of the baseline and future periods (under RCP 4.5 and RCP 8.5) indicate that sub-basins in the basin outlet (i.e., sub-basins 15, 14, and 16), sub-basins that lie in the middle of the basin (i.e., 9, 7, and 8), and upstream sub-basins (i.e., 2 and 3) have a soil of hydrological group D, large slopes, and bare lands with poor-to-medium cover and thus have significant flood potential. This finding is consistent with Maghsood et al. (2019) and Saghafian & Khosroshahi (2005). Therefore, it is necessary to implement watershed management in these sub-basins.
ACKNOWLEDGEMENTS
The authors are thankful to the anonymous reviewers and the editor for their constructive comments and insightful suggestions which helped us to improve the overall quality of the manuscript. Also, we would like to thank the Iran Meteorological Organization, Ministry of Agriculture-Jahad, and the Ministry of Energy of Iran for providing the data.
DATA AVAILABILITY STATEMENT
Data cannot be made publicly available; readers should contact the corresponding author for details.
%20cropped.png?versionId=5956)