Abstract
This study examined the separate and combined impacts of future changes in climate and land use on streamflow, nitrate and ammonium in the Kor River Basin, southwest of Iran, using the representative concentration pathway 2.6 and 8.5 scenarios of the Fifth Assessment Report (AR5) of the Intergovernmental Panel on Climate Change (IPCC). Different land use and climate change scenarios were used and the streamflow, nitrate and ammonium in the future period (2020–2049) under these scenarios were simulated by Integrated Catchment Model for Nitrogen (INCA–N). Results indicated that climate change will increase streamflows and decrease nitrate and ammonium concentrations in summer and autumn. Land use changes were found to have a little impact on streamflows but a significant impact on water quality, particularly under an urban development scenario. Under combined scenarios, larger seasonal changes in streamflows and mixed changes of nitrate and ammonium concentrations were predicted.
INTRODUCTION
The positive and negative impacts of global climate change on natural and social environment are confirmed by the Intergovernmental Panel on Climate Change (IPCC) (Whitehead et al. 2015), with changing the runoff from catchments and altered streamflow in rivers. Climate change is predicted to alter chemical transformations and transport characteristics of water pollutants impacted by climate change in European catchments (Tu 2009; Whitehead et al. 2009). Many recent studies have investigated the impact of climate change on hydrology and water quality (Mimikou et al. 2000; Whitehead et al. 2006; Wilby et al. 2006; Cox & Whitehead 2009; Rehana & Mujumdar 2011; Jin et al. 2012; Crossman et al. 2013a; Li et al. 2016). These studies confirmed that streamflow and water quality variability are closely associated with climate change. For instance, Jin et al. (2012) indicated that during the 2030s, drier hydrological conditions will prevail in the River Thames catchment in southern England. The degree of dryness will be much more profound in the 2080s. Furthermore, their results showed that river NO3–N concentration will increase in winter and decrease in summer as a result of climate change.
In addition to climate change, land use activities such as conversion of natural landscapes for human use and different land management practices, have transformed a large proportion of the Earth's land surface. Deforestation, intense agricultural activities, and urban area expansion in a watershed can influence different processes, including infiltration, groundwater recharge, baseflow, and runoff (Fan & Shibata 2015). There are several studies on separate impacts of climate and land use change (Hadjikakou et al. 2011; Whitehead et al. 2011, 2015; Astaraie-Imani et al. 2012; Crossman et al. 2013b; Mitsova 2014; Pervez & Henebry 2015; Fan & Shibata 2015; Fernandes et al. 2016). Most of the previous studies found significant impact of land use changes on streamflow and water quality. For example, Fan & Shibata (2015) found that climate change scenarios have greater impact in increasing surface runoff, lateral flow, groundwater discharge, and water yield than land use change. They also confirmed that the sediment and nutrient loads were mainly supplied from agricultural land under land use in each climate change scenario, suggesting that land use change was responsible for nutrient and sediment load changes in the watershed. However, limited studies have analyzed the combined impacts of these two variables on streamflow and water quality (Tu 2009; Praskievicz & Chang 2011; Wilson & Weng 2011; Tong et al. 2012; Kim et al. 2013; El-Khoury et al. 2015; Mehdi et al. 2015; Whitehead et al. 2015).
Many researchers have used Coupled Model Intercomparison Project Phase 3 (CMIP3) under Special Report on Emission Scenarios (SRES) in the IPCC Fourth Assessment Report (AR4). Only a few studies focused on the CMIP5 models based on the Fifth Assessment Report (AR5) of the IPCC scenarios (Kim et al. 2013; Fernandes et al. 2016). In comparison to CMIP3, CMIP5 includes more comprehensive models and enhanced experiments. In addition, in CMIP5, higher-spatial-resolution models with a richer set of output fields is used. Compared to CMIP3, better documentation of the models and experiment conditions are provided for CMIP5 (Taylor et al. 2012). New scenarios, called representative concentration pathways (RCPs), are based on greenhouse gas concentrations and emissions pathways. The RCP 2.6 (IMAGE) is representative of scenarios that lead to very low greenhouse gas concentration levels. It is a ‘peak-and-decline’ scenario; its radiative forcing level first reaches a value of 3.1 W/m2 by mid-century and returns to 2.6 W/m2 by 2100. In order to reach such radiative forcing levels, greenhouse gas emissions are reduced substantially over time. The RCP 8.5 (MESSAGE) is characterized by increasing greenhouse gas emissions over time that lead to high greenhouse gas concentration levels. It is a ‘rising’ scenario; its radiative forcing level continuously rises to 8.5 W/m2 by 2100. The RCPs also included impacts caused by land use changes. In contrast, the SRES scenarios was based on forcing by greenhouse gas and aerosol from artificial climate change factors (Kim et al. 2013).
The Integrated Catchment model for Nitrogen (INCA–N) (Whitehead et al. 1998a, 1998b) was first developed for the assessment of multiple sources of nitrogen (N) in catchments. The model simulates flow, nitrate–N and ammonium–N concentrations in the soil, groundwater, and in stream water. INCA–N was first used for land use and atmospheric depositions effects on N fluxes in catchments (Whitehead et al. 1998a, 1998b, 2002; Wade et al. 2002). It has also been used to investigate the impacts of climate and land use change on hydrology and water quality (Whitehead et al. 2006, 2009, 2015; Wilby et al. 2006; Cox & Whitehead 2009; Hadjikakou et al. 2011; Jin et al. 2012; Crossman et al. 2013b, 2014).
In this paper, INCA–N was applied to the Kor River Basin (southwest of Iran) to examine separate and combined impacts of climate and land use change on streamflow, nitrate (NO3–N), and ammonium (NH4–N) concentrations. Different sets of climate and land use change scenarios have been evaluated to assess the impacts on both flow and water quality in the Kor River system.
METHODS
Study area
The Kor River is located in the Maharloo–Bakhtegan basin in the Fars province (southwest of Iran). The Maharloo–Bakhtegan Basin lies between latitudes 28°99′–31°25′ N and longitudes 51°82′–54°50′ E. It covers an area of about 31,874 km2 which is 1.9% of the total area of Iran. The Kor River originates from the Zagros Mountains and flows for 280 km before reaching Bakhtegan Lake. The total drainage catchment of the Kor River is 9,700 km2. Average annual temperatures and precipitation in the basin are 17 °C and 240 mm, respectively. Doroudzan dam was built in the Kor River for optimum use of water. The reservoir of the dam is around 960 Mm3 and is used for producing electric energy as well as to supply water for farms and industries like petrochemical plants in the capital of Fars province, Shiraz, and other areas. The Kor River watershed is an agricultural dominated catchment. The majority of the inhabitants are employed in agriculture, with water from the river being predominantly used for irrigation, to supply industry, urban centers, and intensive agriculture (Fakhraei 2009; Mohsenipour et al. 2013). The Kor River starting from downstream of Doroudzan reservoir to Polekhan hydrometric station (1,700 km2) has been selected as the key study area. The mean annual flow (1967–2012) was 46 m3 s−1 at Polekhan. Seasonally, high flows normally occur in the winter and early spring (January to April) and low flow occurs in the summer and late autumn (June to November). The study area, meteorological, discharge and water quality monitoring stations are presented in Figure 1. The period for the average annual temperature and precipitation in Doroudzan meteorological station was 1988–2015. The concentrations of NO3–N and NH4–N were determined using spectrophotometry method (EPA methods 4500 NO3–N and NH4–N). Streamflow in the gauge station (Polekhan) was measured by using stage–discharge relation which involves obtaining a continuous record of stage, making periodic discharge measurements, establishing and maintaining a relation between the stage and discharge, and applying the stage–discharge relation to the stage record to obtain a continuous record of discharge.
INCA–N model
INCA is a physical, process-based model which simulates flow, nitrate–N, and ammonium–N concentrations in soil, groundwater, and stream water (Whitehead et al. 1998a, 1998b). Allowing the user to specify the spatial nature of a river basin, to alter reach length, rate coefficients, land use, velocity–flow relationships, and to vary input pollutant from point sources, diffuse land sources, and diffuse atmospheric sources are the most important features of the model. Multi-branched stream network modeling is supported in the revised version. The model solves a series of interconnected differential equations by using fourth-order Runge–Kutta integrations method, simultaneously (Whitehead et al. 2015).
Figure 2 shows the INCA–N model main flow paths and processes. INCA simulates hydrology flow pathways in the surface and subsurface systems, and tracks fluxes of solutes/pollutants on a daily step in both land and stream portions of watersheds. The branched version of INCA simulates nutrient dynamics in multi-branch stream networks (Whitehead et al. 2015). The model uses a mass balance for the watershed by considering all inputs and outputs, with a daily time step. The key processes of nitrate–nitrogen (NO3–N) and ammonium–nitrogen (NH4–N) input, transformation, and removal in the soil zone are shown in Figure 3. Plant uptake of NO3 and NH4, nitrification, denitrification, mineralization, and immobilization within each land use type within each sub-catchment are modeled by INCA. In Figure 3, the soil reactive zone is assumed to leach water to the deeper groundwater zone and the river. Another assumption is that no biological reactions occur in the groundwater zone and a mass balance of NH4 and NO3 is adequate. The split between the volume of water stored in the soil and the groundwater is calculated using Base Flow Index. The index attempts to estimate the proportions of water in a stream derived from surface and deeper groundwater sources. Additionally, the index represents a practical method for calculating split and is based on the analysis of observed river flows in the UK (Wade et al. 2002).
The denitrification and nitrification are the main processes involved for nitrogen mass alteration in a reach. In a river reach, as shown in Figure 4, the sources of NO3–N and NH4–N input are the previous upstream reach, the soil and groundwater zones, and the direct effluent discharges. Nitrogen alterations in the stream are controlled both by temperature and residence time along with the appropriate rate coefficients (Whitehead et al. 2015). More detailed information of model components and equations is available in Whitehead et al. (1998a, 1998b) and Wade et al. (2002).
Model setup
In order to model Kor and its main tributaries (Sivand, Maein, and Ghadamghah), the river system is divided into nine reaches, with sub-catchments from the Droudzan reservoir outlet to the Polekhan discharge station (Figure 1 and Table 1). The Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) 30 m resolution global digital elevation model from United States Geological Survey (USGS) Global Data Explorer (http://gdex.cr.usgs.gov/gdex/) (NASA JPL 2009) was used to delineate the watershed boundaries for each reach, with six reaches covering the Kor River, and three reaches its main tributaries (Sivand, Maein, and Ghadamghah). Reaches and sub-catchments were selected, taking into account the location of flow gauging and water quality monitoring sites.
Reach No. . | River . | Reach length (m) . | Area (km2) . | % Land use . | Flow gauge . | WQ observation . | ||||
---|---|---|---|---|---|---|---|---|---|---|
Urban . | Arable . | Grassland . | Forest . | Rock . | ||||||
1 | Kor | 7,937 | 26.68 | 0.76 | 26.75 | 68.90 | 3.59 | 0 | ||
2 | Kor | 22,675 | 67.63 | 1.33 | 67.11 | 31.41 | 0.15 | 0 | Y | |
3 | Kor | 38,085 | 427.19 | 1.26 | 54.85 | 16.52 | 25.51 | 1.86 | Y | |
4 | Kor | 4,027 | 146.01 | 0.32 | 82.42 | 6.22 | 0 | 11.04 | Y | |
5 | Kor | 7,380 | 27.25 | 4.20 | 76.80 | 17.75 | 0 | 1.25 | Y | |
6 | Kor | 1,759 | 2.17 | 0 | 87.12 | 0 | 0 | 12.88 | Y | Y |
7 | Sivand | 41,528 | 183.32 | 5.19 | 64.9 | 16.20 | 13.71 | 0 | ||
8 | Maein | 13,015 | 584.60 | 0.45 | 14.97 | 68.23 | 16.35 | 0 | ||
9 | Ghadamghah | 17,747 | 242.28 | 0.52 | 15.03 | 74.06 | 10.39 | 0 |
Reach No. . | River . | Reach length (m) . | Area (km2) . | % Land use . | Flow gauge . | WQ observation . | ||||
---|---|---|---|---|---|---|---|---|---|---|
Urban . | Arable . | Grassland . | Forest . | Rock . | ||||||
1 | Kor | 7,937 | 26.68 | 0.76 | 26.75 | 68.90 | 3.59 | 0 | ||
2 | Kor | 22,675 | 67.63 | 1.33 | 67.11 | 31.41 | 0.15 | 0 | Y | |
3 | Kor | 38,085 | 427.19 | 1.26 | 54.85 | 16.52 | 25.51 | 1.86 | Y | |
4 | Kor | 4,027 | 146.01 | 0.32 | 82.42 | 6.22 | 0 | 11.04 | Y | |
5 | Kor | 7,380 | 27.25 | 4.20 | 76.80 | 17.75 | 0 | 1.25 | Y | |
6 | Kor | 1,759 | 2.17 | 0 | 87.12 | 0 | 0 | 12.88 | Y | Y |
7 | Sivand | 41,528 | 183.32 | 5.19 | 64.9 | 16.20 | 13.71 | 0 | ||
8 | Maein | 13,015 | 584.60 | 0.45 | 14.97 | 68.23 | 16.35 | 0 | ||
9 | Ghadamghah | 17,747 | 242.28 | 0.52 | 15.03 | 74.06 | 10.39 | 0 |
Land use data were obtained from European Space Agency Globcover Portal (http://due.esrin.esa.int/page_globcover.php). Twenty-two classes of Globcover 2009 were reclassified into five groups in order to limit the parameter input for INCA. The reclassified land use groups and details of each group percentages in each reach are illustrated in Figure 5 and Table 1.
INCA requires daily time series of soil moisture deficit (SMD), hydrologically effective rainfall (HER), air temperature, and precipitation as well as growing season, timing and quantity of fertilizer application, and locations of point sources and effluent concentrations as inputs. Daily time series of SMD and HER have been derived using a simple excel spreadsheet model (Limbrick 2002) for the INCA calibration and climate scenario applications, respectively (Figure 6). Growing season and fertilizer applications were obtained from a previous study in the area (Fakhraei 2009) and applied to the INCA model as other required inputs.
Climate change scenarios
The outputs of the fourth generation of The Community Climate System Model (CCSM4.0) based on CMIP5 under two RCP scenarios (RCP 2.6 and 8.5) were obtained from the Center for Environmental Data Analysis. The obtained data were precipitation, minimum and maximum air temperatures, and average temperatures for historical (1976–2005) and future periods (2020–2049). The delta change technique (Goodarzi et al. 2014) was then used in order to obtain the values of ΔT and ΔP. These values were used in LARS–WG to generate future time series of temperature and precipitation for two meteorological stations (Doroudzan and Takhtejamshid). Finally, future climate scenarios SMD and HER were calculated from the projected time series of precipitation and temperature.
Land use change scenarios
Based on previous research, agricultural land use in the area in 2008 has expanded by 60% since 1956 (Royan Consulting Engineers 2009). Three future land use change scenarios (LC) were set up based on this rate of change: (1) constant (LC) in which land use does not change: the current land data (2009) will be used for the future period (baseline); (2) Agricultural development (ALC): agricultural land use will increase 60% relative to 2009 land use in sub-catchments; (3) Urban development: urban land use will be doubled (ULC) relative to baseline land use. The distributions of the baseline and two land use change scenarios in nine sub-catchments are illustrated in Figure 7. Land use distribution is performed as follows: (1) In reaches where grassland land cover exists, in order to apply ALC and ULC scenarios, grassland is converted to agriculture and urban land uses. Other land cover types are not changed; (2) In reaches with low or no grassland land cover values, forest land use was converted to agriculture and urban land uses and other land cover types are not changed. These scenarios are then applied to INCA–N in order to evaluate the impacts of land use on streamflow, nitrate, and ammonium.
Combined scenarios
In this paper, four combined scenarios are utilized:
- 1.
High climate change (RCP 8.5) – Agricultural development (HCC–ALC)
- 2.
Low climate change (RCP 2.6) – Agricultural development (LCC–ALC)
- 3.
High climate change (RCP 8.5) – Urban development (HCC–ULC)
- 4.
Low climate change (RCP 2.6) – Urban development (LCC–ULC).
RESULTS AND DISCUSSION
Calibration and validation result
Streamflow was calibrated and validated at Polekhan discharge station (KOR06) for the periods of 2003–2008 and 2009–2012, respectively. In order to evaluate the model's goodness of fit, the coefficient of determination, or R2, plus the Nash–Sutcliffe model efficiency (NSE) (Nash & Sutcliffe 1970) were used. Streamflow of the Kor River system is simulated efficiently with a good match between modeled and observed flow data for both calibration and validation periods with R2 and NSE of 0.83 and 0.54, respectively, for the validation data set (Table 2). Observed and simulated time series of streamflow in the calibration and validation periods at Polekhan discharge station are presented in Figure 8, indicating that the model fit is good in both data sets. Generally, the complete hydrograph is simulated well and the simulations captured the dynamics of Kor River hydrology, although there are some points where significant differences existed between simulated and observed values. These differences could be attributed to the lack of knowledge of abstraction, illegal water uses in the area, and observation errors.
Reach No. . | Calibration and validation (2003–2012) . | |||||
---|---|---|---|---|---|---|
Flow . | NO3–N . | NH4–N . | ||||
R2 . | NSE . | R2 . | NSE . | R2 . | NSE . | |
1 | n.a.a | n.a. | n.a. | n.a. | n.a. | n.a. |
2 | n.a. | n.a. | 0.65 | 0.64 | 0.67 | 0.26 |
3 | n.a. | n.a. | 0.56 | 0.53 | 0.74 | 0.16 |
4 | n.a. | n.a. | 0.75 | 0.48 | 0.74 | 0.65 |
5 | n.a. | n.a. | 0.74 | 0.73 | 0.74 | 0.33 |
6 | 0.83 | 0.54 | 0.73 | 0.63 | 0.58 | 0.11 |
Reach No. . | Calibration and validation (2003–2012) . | |||||
---|---|---|---|---|---|---|
Flow . | NO3–N . | NH4–N . | ||||
R2 . | NSE . | R2 . | NSE . | R2 . | NSE . | |
1 | n.a.a | n.a. | n.a. | n.a. | n.a. | n.a. |
2 | n.a. | n.a. | 0.65 | 0.64 | 0.67 | 0.26 |
3 | n.a. | n.a. | 0.56 | 0.53 | 0.74 | 0.16 |
4 | n.a. | n.a. | 0.75 | 0.48 | 0.74 | 0.65 |
5 | n.a. | n.a. | 0.74 | 0.73 | 0.74 | 0.33 |
6 | 0.83 | 0.54 | 0.73 | 0.63 | 0.58 | 0.11 |
aNo data available.
In addition to streamflow calibration, it is necessary to calibrate water quality. The water quality data were limited in the catchment as data were collected monthly by previous researchers in the catchment. However, water quality data were only collected in five reaches of the Kor River. Table 2 lists the R2 and NSE values for all reaches where NO3–N and NH4–N measurements were available during the 2003–2012 period. The INCA–N simulated NO3–N and NH4–N concentrations agree well with the limited observed values such as reach KOR06 with R2 and NSE of 0.73 and 0.63 for NO3–N and 0.58 and 0.11 for NH4–N. Observed and modeled NO3–N and NH4–N concentrations at Polekhan water quality monitoring station (KOR06) are shown in Figure 9. Generally, NO3–N and NH4–N concentrations in the river are low in the summer and autumn and then increase over the winter period. High rates of instream denitrification may cause the lower concentrations of nitrate in the summer. Higher temperatures and longer residence times enhance denitrification in summer months which allows more time for microbiological processes to occur (Jin et al. 2012).
Impacts of climate change
The future climate projections (2020–2049) based on two RCP scenarios (RCP 2.6 and RCP 8.5) were compared to baseline climatology (1988–2015) for the Doroudzan station (Figure 10). The future projections show increases in average temperature in all months except in January for RCP 8.5, and are higher in magnitude, particularly in the spring (Figure 10(b)). Average temperature increases are 0.65 and 0.74 °C under RCP 8.5 in the spring. The maximum increase in average monthly temperature is 0.9 °C under RCP 8.5 in July. Precipitation shows a significant increase during the autumn under both scenarios and a slight decrease in other seasons under RCP 8.5, plus a decrease in summer under RCP 2.6. The maximum increase and decrease in precipitation were 421% and 85% under RCP 2.6 in September and August, respectively (Figure 10(a)).
The simulated results of climate change impacts on streamflow, nitrate and ammonium concentrations are indicated in Figure 11. Figure 11(a) shows the changes in mean monthly, seasonal, and annual streamflows for the future period (2020–2049) relative to the baseline simulations (2003–2012) according to RCP 2.6 and RCP 8.5 climate change scenarios. Compared with the baseline, streamflows decreases from January to May and increases from May to November under two future scenarios. In particular, the streamflow increases by 43% and 45% in October and decreases by 13% and 7% in March under RCP 2.6 and 8.5, respectively. Decrease in streamflows from January to May can be explained by projected decrease in precipitations and increase in mean temperatures in this period. The increase in streamflows from June to October can be attributed to the increase in precipitations during the period. In a box-and-whisker plot (Figure 11(a)), changes in seasonal and annual streamflows are illustrated. The seasonal streamflow shows the impacts of future climate change more clearly than monthly streamflow (Kim et al. 2013). The 50th percentile indicate flow reductions in spring and autumn and increases in summer and winter. The ranges of change in summer and spring flow are higher than those in other seasons. The central estimate (50th percentile) of annual flow under RCP 2.6 was higher than those in baseline and RCP 8.5. This can be explained by the higher precipitations under RCP 2.6 than under baseline and RCP 8.5.
The decrease in streamflows in spring and increase in autumn are also reported by previous climate change impact studies (Abbaspour et al. 2009; Amiri & Eslamian 2010). Amiri & Eslamian (2010) noted that in dry provinces of Iran such as Fars, precipitation will increase in fall and decrease in spring. These changes in Fars province were attributed to increase and decrease in fall and spring streamflows, respectively. They also reported that in the southwestern parts of Iran, runoff volume increases during winter and decreases during spring. However, northern parts of Iran (wet regions) will experience increase in discharge in wet seasons and decrease in summer from July to September which cause more frequent and large intensity floods (Abbaspour et al. 2009; Zarghami et al. 2011; Azari et al. 2016).
The changes in simulated monthly and seasonal NO3–N and NH4–N concentrations for future period (2020–2049) relative to baseline simulations (2003–2012) are illustrated in Figure 11(b) and 11(c). The increase in nitrate concentrations occurs in the spring and winter (December–May) then decreases in the summer and fall (June–November) under two RCP scenarios. Seasonally, maximum increase in nitrate concentrations are 87% and 57% in the winter under RCP 2.6 and RCP 8.5, respectively. NO3–N reductions in the fall are about 20% under two scenarios (Figure 11(b)). The increase in NH4–N concentrations occurs similarly to NO3–N in the spring and winter (December–June) then decreases in the summer and autumn (July–November). Maximum increase in ammonium concentrations are 67% and 63% in the winter under RCP 2.6 and RCP 8.5, respectively. Maximum decrease in NH4–N concentrations are 40% and 34% in the autumn under two future scenarios, respectively (Figure 11(c)).
Lower NO3–N and NH4–N concentrations in summer and fall reflect enhanced dilution due to increased river flows. In addition, lower NO3–N and NH4–N concentrations can be attributed to greater amounts of denitrification and nitrification, particularly in summer. With warmer water temperatures and slower water movement in the summer, greater amounts of NO3–N and NH4–N are lost from denitrification and nitrification. The difference in changes of NO3–N and NH4–N concentrations between the two scenarios can be explained by the changes in streamflow predictions. For instance, streamflow decreases during the late winter and spring months under the two scenarios causing NO3–N and NH4–N concentrations to increase due to lower dilution potentials. Jin et al. (2012) and Whitehead et al. (2015) confirmed that lower NO3–N and NH4–N concentrations are related to enhanced denitrification and nitrification processes and increased river flows. In contrast, higher NO3–N and NH4–N concentrations in spring and winter might reflect low flow conditions in these seasons. Streamflows are predicted to decrease in spring and winter which leads to higher NO3–N and NH4–N concentrations due to lower dilution potentials (Hadjikakou et al. 2011). Jin et al. (2012) noted that higher NO3–N concentrations in the winter months might reflect greater flushing from the soil.
Impacts of land use change
The effects of land use change on streamflow, nitrate and ammonium concentrations are presented in Figure 12. In terms of streamflow, there seems to be little difference between scenarios and baseline. This can be explained by less sensitivity of streamflow to land use changes compared to climate changes. Additionally, the Kor River Basin is an agricultural dominated watershed and urban land uses are limited throughout the basin. Therefore, with increasing urban land uses (even with doubling urban land use area) no significant changes in streamflows were projected. Tu (2009) and Tong et al. (2012) confirmed that streamflow is more sensitive to climate changes than to land use changes. However, the water quality is affected by land use change scenarios, particularly by urban development scenario. Average monthly nitrate and ammonium concentrations decrease by 17% under urban development scenario. Agricultural development scenario shows only 5% reductions in average monthly NO3–N and NH4–N concentrations.
Impacts of combined climate and land use change
In addition to evaluation of the separate impacts of climate change and land use change, the combined effects of these two variables are examined and presented in Figure 13. The results indicated that streamflows increase from May to November to 46% under HCC–ALC and HCC–ULC and to 43% under LCC–ALC and LCC–ULC. The ranges of change in streamflow from December to April are −9% to 3% under HCC–ALC and HCC–ULC and −15% to 14% under LCC–ALC and LCC–ULC (Figure 13(a)). The streamflow increases by 32–36% in summer and autumn under high climate change scenario (RCP 8.5) combined with land use change scenarios and by 28–38% under low climate change scenario (RCP 2.6) combined with land use change scenarios, respectively. The reductions in streamflow are 1–5% under HCC–ALC and HCC–ULC and 0.1–7% under LCC–ALC and LCC–ULC, respectively.
The magnitude and direction of changes in streamflow are almost similar to climate change scenarios, suggesting that land use changes have little influence on streamflow in the Kor River Basin. This supports the fact that the streamflow is more sensitive to climate change rather than land use change, as confirmed by previous studies (Tu 2009; Praskievicz & Chang 2011). However, there is a difference in magnitude of impacts on streamflow in this study (Praskievicz & Chang 2011; Tong et al. 2012; Kim et al. 2013) where the land use impacts are significant when compared to climate change.
Combined impacts of climate and land use changes on NO3–N and NH4–N are illustrated in Figure 13(b) and 13(c). In terms of nitrate, concentrations increase in spring and winter (December to May) to 100–132% under high climate change scenario (RCP 8.5) combined with land use change scenarios and to 112–160% under low climate change scenario (RCP 2.6) combined with land use change scenarios. In summer and autumn (June–November) ranges of change are −30% to 49% under HCC–ALC and HCC–ULC and −35% to 35% under LCC–ALC and LCC–ULC, respectively (Figure 13(b)). Ammonium concentrations (NH4–N) show almost similar increase and decrease patterns of NO3–N with an exception in summer where NH4–N shows increases under three combined scenarios (HCC–ALC, HCC–ULC, and LCC–ALC) and a slight decrease under LCC–ULC (Figure 13(c)). Maximum increase in NH4–N concentrations is 109% in winter under LCC–ALC. Maximum decrease in NH4–N concentrations is 40% in autumn under LCC–ULC.
The combined impacts of climate and land use change on nitrate and ammonium indicate that NO3–N and NH4–N are both sensitive to climate change and land use change. For example, the increase in nitrate concentration during spring and winter (December–May) is affected by both climate and land use changes. The increase is enhanced when climate and land use changes both change than change individually, even though land use changes show reductions in NO3–N concentrations. In contrast, during summer and autumn, when the direction of the changes in both individual stressors are opposite, the changes in these months are mixed; both increase and decrease can be found in these monthly and seasonal values when affected by both climate and land use change. For instance, during the autumn, nitrate concentrations show increase (LCC–ALC and HCC–ALC) and decrease (LCC–ULC and HCC–ULC). This increase and decrease are affected by decrease under climate change scenarios and increase under land use change scenarios (increase under ALC and decrease under ULC). These trend of changes under combined scenarios are reported by previous studies (Tu 2009; Tong et al. 2012; El-Khoury et al. 2015; Mehdi et al. 2015).
Limitations and uncertainties
It should be noted that in this study we only used one climate change model (CCSM4.0) and two climate change scenarios (RCP 2.6 and RCP 8.5). Using more climate change models and scenarios will provide more clear and profound results on how climate and land use change impact water quantity and quality in the Kor River watershed. Furthermore, land use change scenarios were developed in this study based on historical rate of land use change, assuming that future rates affected by current rate of land use changes. Further research might be required in order to address more precise land use changes in the future. In addition to limitations and assumptions in this study, there are numbers of inherent uncertainties: the modeling uncertainties due to climate simulations, GCM structure, downscaling technique, choice of the hydrological model (and model structure), data input as well as calibration process (Mehdi et al. 2015). However, despite these limitations and uncertainties, we believe this study provides a relevant and useful indication of the direction and the relative magnitude of changes that may occur in the future in the Kor River Basin.
CONCLUSIONS
This study assessed the separate and combined impacts of climate and land use change on streamflow, nitrate and ammonium concentrations in the Kor River Basin, southwest of Iran. The INCA–N model was used to simulate streamflow, nitrate and ammonium concentrations. Calibration, validation for streamflow, nitrate and ammonium were generally satisfactory and confirmed that INCA–N could be applied to evaluate the responses of Kor River Basin to climate and land use changes. Two climate change scenarios (RCP 2.6 and RCP 8.5) based on CCSM4.0 climate model along with two land use change scenarios and four combined scenarios are established.
Results of climate change scenarios revealed that monthly streamflow decreases from January to May and increases from May to November under two future scenarios. In particular, the streamflow increases in October and decreases in March under RCP 2.6 and 8.5, respectively. Seasonal streamflow projections indicate flow reductions in spring and autumn and increases in summer and winter. This pattern of change in streamflow can be attributed to different precipitation and temperature changes under two future climate change scenarios: nitrate and ammonium concentrations in the spring and winter (December–May) then decreases in the summer and fall (June–November) under two RCP scenarios. These reductions in nitrate and ammonium concentrations might reflect enhanced dilution due to increased river flows and greater amounts of denitrification and nitrification in summer and fall. Increase in concentrations are related to lower dilution potentials. Land use change impact results indicated that streamflow is less sensitive to land use change. However, water quality is affected by land use change scenarios, particularly by urban development scenario. Combined impact results revealed that the streamflow is more sensitive to climate change rather than land use change. In terms of nitrate and ammonium concentrations, increases in spring and winter and decreases in summer and autumn under different combined scenarios are projected. The combined impacts of climate and land use change on nitrate and ammonium indicated that NO3–N and NH4–N are both sensitive to climate change and land use change.
Considering the limitations and uncertainties of this study, it can be concluded that, examining separate and combined impacts of climate and land use change are useful for water resource managers and decision-makers in order to prepare and adapt strategies to mitigate negative impacts of climate and land use changes in the watershed.
DISCLOSURE STATEMENT
No potential conflicts of interest were reported by the authors.