Abstract

It is essential to assess streamflow response to climate and land-use change in catchment basins that serve cities and their surrounding areas. This study used the Distributed Hydrology Soil Vegetation Model (DHSVM) to simulate streamflow under different climate and land-use change scenarios in the Dashi River catchment, China. The most sensitive soil parameters were maximum infiltration, porosity, field capacity, and wilting point, while the most sensitive vegetation parameters were leaf area index (LAI) and vegetation height. The suitability of the DHSVM model was aligned with Nash–Sutcliffe model efficiency coefficients (NSE) greater than 0.41 and 0.84 at daily and monthly scales, respectively. Streamflow increased/decreased with increasing/decreasing precipitation, while it decreased with increasing air temperature. Furthermore, streamflow decreased with the increase in forestland due to higher water consumption, especially during summer. Results from this study could help us to better understand streamflow response to changes in climate and land use.

INTRODUCTION

The combined effect of climate and land-use change has dramatically altered hydrological processes (Niraula et al. 2015). On the one hand, climate change/variability effectuate temperature increases, precipitation changes, and evapotranspiration alterations, thereby affecting the volume, peak rate, and timing of streamflow (Novotny & Stefan 2007). On the other hand, anthropogenic activities cause land-use change to occur, which alters rainfall interception, evapotranspiration, and the hydraulic conductivity of surface soil, which leads to changes in surface streamflow, stream discharge, and sediment transportation (Yan et al. 2013). In recent years, extreme hydrological events have generally exhibited an increasing trend that corresponds to the intensification of global warming and anthropogenic activities (Goswami et al. 2006), which have led to an increase in the frequency of extreme hydrological catastrophes, such as flood and drought, e.g., Europe (Madsen et al. 2014), and the Haihe basin of China (Zhang et al. 2015a). In particular, extreme hydrological events have enhanced the risk of flood and drought in catchment basins that serve cities and their surrounding areas. This is due to the expansion of impermeable areas that have undermined the water storage capacities of catchments (Peng et al. 2015).

In order to reduce such risks, it is essential to better understand the potential effects of climate and land-use change on streamflow at a basin scale (Bellu et al. 2016). Considerable attention has been paid to the assessment of separate or combined impacts of climate and land-use change on streamflow (Wang et al. 2013a). For example, Jiang & Wang (2016) compared the impact of climatic and anthropogenic effects in the Liao River Basin and concluded that the anthropogenic effect had a far greater contribution (>56.6%) on streamflow variability compared to climate change (<43.4%).

Distributed hydrological models have been widely used to estimate the contribution of climate and land-use effects on streamflow (e.g., Karlsson et al. 2016). Given their rapid hydrological response to climate and land-use change, mountainous river systems are subject to higher flood risk (e.g., the July 2012 Beijing flood that resulted in significant damage and loss in the region (Wang et al. 2013b)). Accordingly, the objectives of this study are: (i) to explore parameter sensitivity and assess the performance of the Distributed Hydrology Soil Vegetation Model (DHSVM) in a small mountainous catchment (drainage basin) that serves a city and its surrounding area; and (ii) to estimate streamflow response to various climate and land-use change scenarios in the Dashi River catchment, China. The results from this study could help us to better understand streamflow response to changes in climate and land use, and subsequently address flood risk in catchment basins that serve cities and their surrounding areas.

STUDY SITE AND METHODOLOGY

Study site

The Dashi River is a tributary of the Juma River in the Haihe River basin, which originates from the Xiayunling National Forest Park in the Fangshan District, southwest Beijing, China. The middle and upper Dashi River catchment areas are mountainous, while the lower catchment area is plain. The total length of the Dashi River is 129 km, and the catchment area is 1,280 km2. The Dashi River catchment belongs to a semi-humid continental monsoon climate, which is hot and rainy in summer and cold and dry in winter. The mean annual temperature range is between 9 and 11 °C, while the mean annual precipitation (from 1956 to 2004) is 587.6 mm, which is concentrated in summer (from June to August). We selected the upper catchment, a 664 km2 mountainous area above the hydrological station, as the study site (Figure 1).

Figure 1

Geographical location of the Dashi River catchment, China. The meteorological station (signified by ▴) and the hydrological station (signified by ▪) are also shown in the figure.

Figure 1

Geographical location of the Dashi River catchment, China. The meteorological station (signified by ▴) and the hydrological station (signified by ▪) are also shown in the figure.

Being a tributary, the hydrological system of the Dashi River has undergone significant changes, impacted by climate change/variability and anthropogenic activities that have taken place throughout the past couple of decades in the Haihe River basin (Lei et al. 2014). Although streamflow has exhibited a statistically significant decreasing trend in the Dashi River in the past couple of decades, the Dashi River catchment was impacted by a severe rainstorm on July 21, 2012 (known as the July 2012 Beijing flood). This was the strongest rainstorm recorded in northern China since detailed meteorological records began in 1951, and also caused the greatest number of casualties and economic loss recorded (Su et al. 2015).

The distributed hydrology soil vegetation model and its application

Model description

The Distributed Hydrology Soil Vegetation Model (DHSVM) is a physically based distributed hydrologic model developed for mountainous catchments (Wigmosta et al. 1994). The DHSVM is designed to resolve the water and energy balance of each grid cell and requires both spatially distributed variables (soil, vegetation, elevation, and geology) and climate forcing (precipitation, temperature, wind speed, solar radiation, and relative humidity). This model has already been successfully and widely utilized in various research disciplines, such as hydrologic analysis and modeling (Wigmosta et al. 1994; Safeeq & Fares 2012), as well as in studies related to climate change effects on water resources (Cuo et al. 2009; Dickerson-Lange & Mitchell 2014) and land-use change impacts on hydrological processes (Cuo et al. 2008; Zhang et al. 2011).

The DHSVM was originally designed for application in temperate mountainous forest catchments located in the northwestern region of the United States of America. Basically, the model accounts for topographic and vegetation effects in each grid cell, while representing physical processes, such as the land surface energy balance, unsaturated soil moisture movement, saturation overland flow, snow accumulation and melt, and water table recharge and streamflow. The DHSVM uses a two-layer canopy representation for evapotranspiration, which is estimated by the Penman–Monteith equation. The model also includes a two-layer energy-balance model for snow accumulation and melt. The multilayer soil column in each pixel represents a series of soil moisture reservoirs, and lateral saturated subsurface flow takes place in the deepest soil layer. Runoff generation is represented by saturation excess and infiltration excess mechanisms. Stream segment storage and propagation are computed using linear-reservoir routing (Wigmosta et al. 1994).

Model application

To build the DHSVM for use in the Dashi River catchment, we compiled a model database using meteorological data (e.g., precipitation, wind speed, maximum and minimum temperatures) and topographic data (consisting of a digital elevation map as well as land use and soil texture). We obtained meteorological data from 1961 to 2012, including precipitation, temperature, and wind speed, for DHSVM calibration and validation, from the Xiayunling weather station (Figure 1). This model requires a digital elevation model (DEM) as well as soil texture and vegetation types to simulate hydrological cycles. We used the Arcinfo Geographical Information System (GIS) product line to process all necessary maps for this study. We generated a 100 m resolution DEM from 90 m resolution USGS/NASA Shuttle Radar Topography Mission (SRTM) data. For the Dashi River catchment, the soil map (100 m resolution) was adapted from a world soil database (Harmonized World Soil Database). We generated the vegetation map (100 m resolution) based on the land-cover data of Beijing in 1980 provided by the Data Center for Resources and Environmental Sciences (RESDC), Chinese Academy of Sciences.

Sensitivity analysis

One-factor-at-a-time (OFAT) has been widely used to diagnose land-surface models, and has proven to be a useful approach in doing so (Cuo et al. 2011). We used this method to analyze parameter sensitivity of the DHSVM, and examined such parameters using bias and the root-mean-square error (RMSE). The simulation period for sensitivity analysis was from January 1, 1977, to December 31, 1979. Soil parameters and vegetation parameters are provided in Tables 1 and 2, respectively. 
formula
(1)
 
formula
(2)
where is the simulated streamflow value at time step i before adjustment; is the simulated streamflow value after adjustment; n is the simulation time.
Table 1

Soil parameters in the DHSVM used for sensitivity analysis

Parameters Abbrev. Units Range/Mean ± SD Source 
1. Lateral conductivity Kls m/s 3.55 × 107 ± 1.48 × 107 Meyer et al. (1997)  
2. Exponential decrease EDR m−1 1–5 Model default 
3. Maximum infiltration Imax m/s 3.55 × 107 ± 1.48 × 107 Meyer et al. (1997)  
4. Surface albedo α – 0.1–0.2 Model default 
5. Porosity m3/m3 0.410 ± 0.0513 Meyer et al. (1997)  
6. Pore size distribution PSD – 0.275 ± 0.0834 Meyer et al. (1997)  
7. Bubbling pressure hb 0.507 ± 0.305 Meyer et al. (1997)  
8. Field capacity fc m3/m3 0.299 ± 0.0623 Meyer et al. (1997)  
9. Wilting point wp m3/m3 0.165 ± 0.0344 Meyer et al. (1997)  
10. Bulk density ρb kg/m3 1,000–2,000 Model default 
11. Vertical conductivity Kvs m/s 3.55 × 107 ± 1.48 × 107 Meyer et al. (1997)  
12. Thermal conductivity TC W/mK 6–8 Model default 
13. Thermal capacity VTC J/m31 × 106–2 × 106 Model default 
Parameters Abbrev. Units Range/Mean ± SD Source 
1. Lateral conductivity Kls m/s 3.55 × 107 ± 1.48 × 107 Meyer et al. (1997)  
2. Exponential decrease EDR m−1 1–5 Model default 
3. Maximum infiltration Imax m/s 3.55 × 107 ± 1.48 × 107 Meyer et al. (1997)  
4. Surface albedo α – 0.1–0.2 Model default 
5. Porosity m3/m3 0.410 ± 0.0513 Meyer et al. (1997)  
6. Pore size distribution PSD – 0.275 ± 0.0834 Meyer et al. (1997)  
7. Bubbling pressure hb 0.507 ± 0.305 Meyer et al. (1997)  
8. Field capacity fc m3/m3 0.299 ± 0.0623 Meyer et al. (1997)  
9. Wilting point wp m3/m3 0.165 ± 0.0344 Meyer et al. (1997)  
10. Bulk density ρb kg/m3 1,000–2,000 Model default 
11. Vertical conductivity Kvs m/s 3.55 × 107 ± 1.48 × 107 Meyer et al. (1997)  
12. Thermal conductivity TC W/mK 6–8 Model default 
13. Thermal capacity VTC J/m31 × 106–2 × 106 Model default 
Table 2

Vegetation parameters in the DHSVM used for sensitivity analysis

Parameters Abbrev. Units Range/Mean ± SD Source 
1. Fractional coverage FC m2/m2 0.6–0.9 Field observation 
2. Hemi fract coverage HFC m2/m2 0.6–0.9 Field observation 
3. Trunk space TS m/m 0.4–0.6 Field observation 
4. Aerodynamic attenuation AA – 0.1–3.5 Kang et al. (2008)  
5. Radiation attenuation RA – 0.1–0.5 Kang et al. (2008)  
6. Max snow int capacity MSIC m/m 0.055 ± 0.05 Breuer et al. (2003)  
7. Snow interception Eff SIE – 0–1 Model default 
8. Mass release drip ratio MRDR m/m 0–1 Model default 
9. Overstory vegetation height ho 12–17 Field observation 
10. Understorey vegetation height hu 1–3 Field observation 
11. Overstorey monthly LAI LAIo m2/m2 5.4 ± 1.7 Breuer et al. (2003)  
12. Understory monthly LAI LAIu m2/m2 6.2 ± 3.8 Breuer et al. (2003)  
13. Maximum resistance Rs max s/m 1,000–3,000 Du et al. (2014)  
14. Minimum resistance Rs min s/m 300–680 Du et al. (2014)  
15. Moisture threshold MT m3/m3 0.165 ± 0.0344 Meyer et al. (1997)  
16. Vapor pressure deficit VPD Pa 600–5,000 Kang et al. (2008)  
17. Rpc RPC W/m2 0.01–0.95 Cuo et al. (2011
18. Overstorey monthly Alb αo – 0.27 ± 0.05 Breuer et al. (2003)  
19. Understorey monthly Alb αu – 0.27 ± 0.07 Breuer et al. (2003)  
20. Root zone depths RZD 0.5–0.7 Field observation 
Parameters Abbrev. Units Range/Mean ± SD Source 
1. Fractional coverage FC m2/m2 0.6–0.9 Field observation 
2. Hemi fract coverage HFC m2/m2 0.6–0.9 Field observation 
3. Trunk space TS m/m 0.4–0.6 Field observation 
4. Aerodynamic attenuation AA – 0.1–3.5 Kang et al. (2008)  
5. Radiation attenuation RA – 0.1–0.5 Kang et al. (2008)  
6. Max snow int capacity MSIC m/m 0.055 ± 0.05 Breuer et al. (2003)  
7. Snow interception Eff SIE – 0–1 Model default 
8. Mass release drip ratio MRDR m/m 0–1 Model default 
9. Overstory vegetation height ho 12–17 Field observation 
10. Understorey vegetation height hu 1–3 Field observation 
11. Overstorey monthly LAI LAIo m2/m2 5.4 ± 1.7 Breuer et al. (2003)  
12. Understory monthly LAI LAIu m2/m2 6.2 ± 3.8 Breuer et al. (2003)  
13. Maximum resistance Rs max s/m 1,000–3,000 Du et al. (2014)  
14. Minimum resistance Rs min s/m 300–680 Du et al. (2014)  
15. Moisture threshold MT m3/m3 0.165 ± 0.0344 Meyer et al. (1997)  
16. Vapor pressure deficit VPD Pa 600–5,000 Kang et al. (2008)  
17. Rpc RPC W/m2 0.01–0.95 Cuo et al. (2011
18. Overstorey monthly Alb αo – 0.27 ± 0.05 Breuer et al. (2003)  
19. Understorey monthly Alb αu – 0.27 ± 0.07 Breuer et al. (2003)  
20. Root zone depths RZD 0.5–0.7 Field observation 

Parameters 1–8, 11 and 18 were only used for forestland.

The model was first run using mean values of all soil and vegetation parameters to derive the reference time series of daily streamflow. For experimental runs, we increased or decreased all parameter values by 1 standard deviation (SD), while we maintained the remaining parameter values at their specified mean values, except when a decrease of 1 SD resulted in a negative value. Parameter means and SD were taken from the literature. Additionally, for parameters for which only the range was known, we calculated means and SD under the assumption of a uniform distribution (Cuo et al. 2011): 
formula
(3)
 
formula
(4)
where a and b are the lower and upper boundaries of the range, respectively.

Calibration and validation

We used the period from January 1, 1973, to December 31, 1975, for calibration, and we used the period from January 1, 1977, to December 31, 1979, for validation. The DHSVM was calibrated and validated based on continuous streamflow measurements, conducted at a hydrological station downstream of the Dashi River catchment (Figure 1).

The model was run at a 3 h time step and the simulated streamflow was converted to daily and monthly values to evaluate model performance. Nash–Sutcliffe model efficiency coefficients (NSE) (Nash & Sutcliffe 1970) and coefficients of determination (R2) were used to evaluate model performance for our simulations. The NSE was defined as follows: 
formula
(5)
where is the measured data; is the model simulation output; and is the mean of measured data.

Climate and land-use change scenarios

In order to assess streamflow response to climate and land-use change, we established eight climate change scenarios and four land-use change scenarios for validation. We also used validated simulation results as a reference baseline. We used eight climate change scenarios to assess impacts of climate change on streamflow with precipitation changes of +10% (scenario 1), +20% (scenario 2), −10% (scenario 3), and −20% (scenario 4) when air temperature ranged within that found under natural conditions, and with increases in air temperature of 1 °C (scenario 5), 2 °C (scenario 6), 3 °C (scenario 7), and 4 °C (scenario 8) when precipitation ranged within that found under natural conditions.

The dominant land-use types in the Dashi River catchment are forestland (84%), grassland (9%), and farmland (5%), which account for approximately 98% of the entire area. Therefore, we established four scenarios to determine how land-use adjustments impact streamflow in the Dashi River catchment. These four different land-use scenarios are as follows: scenario 9, grassland converted from farmland (forestland, 84%; grassland, 14%; others, 2%); scenario 10, forestland converted from farmland (forestland, 89%; grassland, 9%; others, 2%); scenario 11, grassland converted from forestland (grassland, 93%; farmland, 5%; others, 2%); and scenario 12, forestland converted from grassland (forestland, 93%; farmland, 5%; others, 2%).

RESULTS

Parameter sensitivity and model validation

To determine optimal model parameters, we applied sensitivity analyses to 13 soil parameters and 20 vegetation parameters of the DHSVM. Among these parameters, the most sensitive soil parameters were maximum infiltration, porosity, field capacity, and wilting point, while the most sensitive vegetation parameters were LAI and vegetation height (Figure 2). The DHSVM was manually calibrated on the basis of daily and monthly streamflow by adjusting these aforementioned sensitive parameters.

Figure 2

Parameter analysis results from the Dashi River catchment. (a) and (b) represent the bias and RMSE of soil parameters; (c) and (d) represent the bias and RMSE of vegetation parameters in forestland; (e) and (f) represent the bias and RMSE of vegetation parameters in grassland.

Figure 2

Parameter analysis results from the Dashi River catchment. (a) and (b) represent the bias and RMSE of soil parameters; (c) and (d) represent the bias and RMSE of vegetation parameters in forestland; (e) and (f) represent the bias and RMSE of vegetation parameters in grassland.

For both the calibration and validation periods, simulated daily and monthly streamflow agreed reasonably well with the observed streamflow from both time steps (Figure 3). The NSE values for daily streamflow simulations during the calibration and validation periods were 0.62 and 0.41, respectively, while the NSE values for monthly simulations during the calibration and validation periods were 0.79 and 0.84, respectively. The R2 of the monthly time step reached 0.84 and 0.89 for the calibration and validation periods, respectively (Table 3).

Table 3

Prediction accuracy for daily and monthly streamflow defined by the Nash–Sutcliffe model efficiency (NSE) coefficients and the coefficients of determination (R2)

Statistical Index Calibration (1973–1975)
 
Validation (1961–1966, 1977–1979)
 
Daily streamflow Monthly streamflow Daily streamflow Monthly streamflow 
NSE 0.62 0.79 0.42 0.83 
R2 0.73 0.84 0.77 0.89 
Statistical Index Calibration (1973–1975)
 
Validation (1961–1966, 1977–1979)
 
Daily streamflow Monthly streamflow Daily streamflow Monthly streamflow 
NSE 0.62 0.79 0.42 0.83 
R2 0.73 0.84 0.77 0.89 
Figure 3

Observed and simulated streamflow during the calibration (from January 1, 1973 to December 31, 1975) and validation (from January 1, 1977 to December 31, 1979) periods: (a) daily streamflow; (b) monthly streamflow.

Figure 3

Observed and simulated streamflow during the calibration (from January 1, 1973 to December 31, 1975) and validation (from January 1, 1977 to December 31, 1979) periods: (a) daily streamflow; (b) monthly streamflow.

Streamflow response to climate change

Figure 4 shows streamflow response to changes in the precipitation and air temperature scenarios. As expected, mean annual streamflow kept pace with increases or decreases in precipitation, while mean annual streamflow exhibited decreasing trends along with increases in air temperature. These findings could also result from changes in the relative percentage of annual average streamflow as well as from changes in the maximum and minimal baseline rates (deduced from Figure 4 and shown in Table 4).

Table 4

Predicted relative percentage changes in annual average streamflow and the maximum and minimum change rates of the baseline under the different climate scenarios

Scenarios
 
Change in mean annual streamflow (%) Maximum change of streamflow (%) Minimum change of streamflow (%) 
Precipitation change 32.73 54.28 9.47 
68.18 122.50 17.92 
−28.69 −39.32 −9.82 
−52.17 −64.48 −19.73 
Air temperature change −2.79 −8.67 −1.33 
−4.80 (2.01) −16.35 −2.28 
−5.93 (1.13) −23.31 −2.27 
−6.12 (0.19) −29.44 −1.07 
Scenarios
 
Change in mean annual streamflow (%) Maximum change of streamflow (%) Minimum change of streamflow (%) 
Precipitation change 32.73 54.28 9.47 
68.18 122.50 17.92 
−28.69 −39.32 −9.82 
−52.17 −64.48 −19.73 
Air temperature change −2.79 −8.67 −1.33 
−4.80 (2.01) −16.35 −2.28 
−5.93 (1.13) −23.31 −2.27 
−6.12 (0.19) −29.44 −1.07 

Values in parentheses are relative percentage changes in annual average streamflow for the same temperature increase between scenarios.

Figure 4

Simulated streamflow for different climate change scenarios: (a) streamflow changes under precipitation change scenarios 1 to 4, which correspond to precipitation changes of +10%, +20%, −10%, and −20%; and (b) streamflow changes under air temperature scenarios, where scenario 5 to 8 correspond to air temperature increases of 1 °C, 2 °C, 3 °C, and 4 °C.

Figure 4

Simulated streamflow for different climate change scenarios: (a) streamflow changes under precipitation change scenarios 1 to 4, which correspond to precipitation changes of +10%, +20%, −10%, and −20%; and (b) streamflow changes under air temperature scenarios, where scenario 5 to 8 correspond to air temperature increases of 1 °C, 2 °C, 3 °C, and 4 °C.

Increases in precipitation contributed significantly to changes in streamflow. For example, annual streamflow increased by approximately 32.73% and 68.18% for precipitation increases of 10% and 20%, respectively, while annual streamflow decreased by approximately 28.69% and 52.17% for precipitation decreases of 10% and 20%, respectively. Maximum changes in streamflow yielded similar results. Furthermore, an increase in air temperature will cause a decrease in streamflow due to its complex effect on hydrological processes. It is interesting to note that we found a decrease in streamflow sensitivity with an increase in air temperature. As shown in Table 4, a 1 °C increase in temperature will cause a 2.79% decrease in streamflow, while the same temperature increase will only cause −2.01%, −1.13%, and −0.19% increases in streamflow between scenarios 5 and 6, scenarios 6 and 7, and scenarios 7 and 8, respectively.

Streamflow response to land-use change

Figure 5 shows streamflow response to the four land-use change scenarios. Streamflow showed a significant increase (P < 0.05) when forestland was converted to grassland (i.e., scenario 11), especially during the rainy season (from June to August); streamflow showed a decreasing trend when farmland or grassland was converted to forestland (i.e., scenarios 10 or 12); streamflow showed a slightly increasing trend when farmland was converted to grassland (i.e., scenario 9). Furthermore, changes in monthly streamflow of the different scenarios (see Figure 5(b)) showed that: streamflow varied slightly in scenario 9 due to the similar hydrological regulations for farmland and grassland; streamflow showed similar decreasing trends in scenarios 10 and 12 when farmland or grassland was converted to forestland, which resulted in an annual streamflow decrease of 3.45% and 6.45%, respectively, with a maximum reduction in July for both scenarios (i.e., 7.30% and 13.84%, respectively); streamflow increased within a range from 12.75% (June) to 67.08% (July) with an average of 35.81% when forestland was converted to grassland (scenario 11).

Figure 5

Simulated streamflow for the different land-use change scenarios determined by the DHSVM in the Dashi River catchment: (a) is the simulated average monthly streamflow, and (b) is the streamflow change rate.

Figure 5

Simulated streamflow for the different land-use change scenarios determined by the DHSVM in the Dashi River catchment: (a) is the simulated average monthly streamflow, and (b) is the streamflow change rate.

DISCUSSION

Streamflow response to climate change

As shown in Figure 5, streamflow increased with increasing precipitation, while streamflow decreased with increasing air temperature. As reported by Bao et al. (2012), additive effects of decreasing precipitation and increasing air temperature have contributed to a decrease in streamflow in the Haihe basin. Results from sensitivity analysis showed that streamflow changes are more sensitive to precipitation increases compared to precipitation decreases. For example, streamflow can increase by 32.73% or decrease by 28.69% with a 10% increase or decrease in precipitation, respectively. Streamflow response to changes in precipitation is consistent with findings from Qi et al. (2009). It is interesting to note that the increasing rate of streamflow decreased with an increase in air temperature. For example, streamflow decreased by 2.01% under a 1 °C temperature increase compared to baseline conditions, while streamflow decreased by 1.13% under a 2 °C temperature increase compared to a 1 °C temperature increase. This is primarily because streamflow will decrease with an increase in evaporation when the range of other meteorological variables are within normal conditions. In fact, streamflow changes that arise from increasing air temperature reflect the complex interactions between meteorological variables. In particular, explanations derived from the ‘evaporation paradox’ have enhanced our overall understanding of hydrological mechanisms related to increasing air temperature (Brutsaert & Parlange 1998; Liu & McVicar 2012; Szilagyi et al. 2017).

Furthermore, an increase in extreme hydrological events is the result of climate change (Jena et al. 2014), which will increase flood risk in the study area due to concentrated precipitation distribution effects (see Figure 5). Lei et al. (2014) concluded that precipitation was the primary driver for changes in river streamflow over the past century in northern China, whereas temperature had a negative impact on streamflow, and the effect of temperature was clearly weaker than precipitation. In addition to the temporal and spatial variability of precipitation, a shift from snowfall to rainfall could also result in a decrease in streamflow (Berghuijs et al. 2014). In our study, we did not explain the effect of snowfall on streamflow, first because precipitation was concentrated in summer and second because of the small scale of the study area.

Streamflow response to land-use change

Changes in vegetation type have a profound impact on the water balance of catchments, and such changes control the response of streamflow at a basin scale (Gentine et al. 2012; Liu et al. 2016), as verified by our results (see Figure 5). For example, streamflow increased by an average of 35.81% when forestland was converted to grassland. This is because changes in vegetation characteristics resulting from land-use change, such as leaf area and rooting depth, inevitably alter the partitioning of precipitation into streamflow and evapotranspiration (Zhang et al. 2001; Liu et al. 2016). Sensitivity analysis showed that the LAI and vegetation height were the two key parameters that regulated streamflow (as shown in Table 2). Similar vegetation characteristics between the different land-use types (such as farmland and grassland) produced similar streamflow response. Compared to other land-use types, forests with higher leaf area and deeper root depth can effectively reduce streamflow. As reported by Liu & McVicar (2012), this is likely the result of changes in land use, associated with increased perennial vegetation cover and increased root depth (which is consistent with findings by Zhang et al. (2001), Zhang et al. (2015b), and Liu et al. (2016)). Land-use changes have been confirmed to be one of the most effective measures in controlling streamflow (Ozdemir & Elbasi 2015; Berghuijs et al. 2016; Pourmokhtarian et al. 2017). For example, Isik et al. (2013) modeled the effects of changing land-use conditions on daily streamflow, and they predicted increased flow under an urban scenario and a lower average streamflow in forest-dominated scenarios. In China, several policies have been implemented to protect and restore the ecological environment, such as the ‘Grain for Green’ program, which has dramatically altered vegetation patterns in the Loess Plateau as well as related hydrological processes (Liang et al. 2015). Presently, China's ‘sponge cities’ have garnered considerable attention in improving the quality of urban ecological environments and reducing flood risk (e.g., Xia et al. 2017). The regulation of land-use change on streamflow will help address climate change impacts and subsequently reduce flood risks in urban areas.

CONCLUSIONS

This study investigated streamflow response to changes in climate and land-use type within a mountainous catchment that serves a city and its surrounding area, namely, the Dashi River catchment. The DHSVM was used in the study area and run under various climate and land-use change scenarios. Several conclusions can be drawn from our results:

  1. We examined parameter sensitivity of the DHSVM in this area. The most sensitive soil parameters were maximum infiltration, porosity, field capacity, and wilting point, while the most sensitive vegetation parameters were leaf area index and vegetation height. The DHSVM was good at simulating streamflow with NSE and R2 of 0.83 and 0.89 at a monthly time scale, respectively.

  2. Streamflow was more sensitive to increasing precipitation than decreasing precipitation. For air temperature, streamflow decreased with an increase in air temperature, and streamflow gradients in conjunction with an increase in air temperature exhibited a decreasing trend; namely, a 1 °C increase in temperature caused a 2.01% decrease in streamflow compared to baseline conditions, while a 2 °C increase in temperature caused a 1.13% decrease in streamflow compared to a 1 °C increase in temperature.

  3. Streamflow decreased with an increase in forestland area, while streamflow increased with a decrease in forestland area, especially during the summer season (from June to August). Changes in grassland and farmland land-use types exhibited similar patterns in regulating streamflow.

  4. The results indicated that changes with climate changes, especially increasing extreme precipitation events, would increase the flood risk. Consequently, the regulation of land-use types would help to reduce flood risks and alleviate climate change impacts on streamflow.

ACKNOWLEDGEMENTS

This study was supported by the National Key Basic Research and Development Project (Grant No. 2017YFC0404505), the National Natural Science Foundation of China (no. 51439001, 51579008, and 41771042), the Interdisciplinary Research Funds of Beijing Normal University, and the Open Research Fund Program of the Beijing Climate Change Response Research and Education Center (Beijing University of Civil Engineering and Architecture). We would like to thank the National Meteorological Information Center, China Meteorological Administration, for providing the meteorological data used in this study.

REFERENCES

REFERENCES
Bao
Z. X.
,
Zhang
H. Y.
,
Wang
G. Q.
,
Fu
G. B.
,
He
R. M.
,
Yan
X. L.
,
Jin
J. L.
,
Liu
Y. L.
&
Zhang
A.
2012
Attribution for decreasing streamflow of the Haihe River basin, northern China: climate variability or human activities?
J. Hydrol.
460–461
,
117
129
.
Berghuijs
W. R.
,
Woods
R. A.
&
Hrachowitz
M.
2014
A precipitation shift from snow towards rain leads to a decrease in streamflow
.
Nat. Clim. Change
4
,
583
586
.
Berghuijs
W. R.
,
Hartmann
A.
&
Woods
R. A.
2016
Streamflow sensitivity to water storage changes across Europe
.
Geophys. Res. Lett.
43
,
1980
1987
.
Breuer
L.
,
Eckhardt
K.
&
Frede
H. G.
2003
Plant parameter values for models in temperate climates
.
Ecol. Model.
169
,
237
293
.
Brutsaert
W.
&
Parlange
M. B.
1998
Hydrologic cycle explains the evaporation paradox
.
Nature
396
,
30
.
doi:10.1038/23845
.
Cuo
L.
,
Lettenmaier
D. P.
,
Mattheussen
B. V.
,
Storck
P.
&
Wiley
M.
2008
Hydrologic prediction for urban watersheds with the distributed hydrology-soil-vegetation model
.
Hydrol. Process.
22
,
4205
4213
.
Cuo
L.
,
Lettenmaier
D. P.
,
Alberti
M.
&
Richey
J. E.
2009
Effects of a century of land cover and climate change on the hydrology of the Puget Sound basin
.
Hydrol. Process.
23
,
907
933
.
Gentine
P.
,
D'Odorico
P.
,
Lintner
B. R.
,
Sivandran
G.
&
Salvucci
G.
2012
Interdependence of climate, soil, and vegetation as constrained by the Budyko curve
.
Geophys. Res. Lett.
39
(
19
),
L19404
.
http://dx.doi.org/10.1029/2012GL053492
.
Goswami
B. N.
,
Venugopal
V.
,
Sengupta
D.
,
Madhusoodanan
M. S.
&
Xavier
P. K.
2006
Increasing trend of extreme rain events over India in a warming environment
.
Science
314
,
1442
1445
.
Kang
L. L.
,
Wang
S. R.
&
Gu
J. Q.
2008
The simulation test of the distributed hydrological model DHSVM on the runoff change of Lanjiang River basin
.
J. Trop. Meteorol.
24
,
176
182
.
(in Chinese)
.
Karlsson
I. B.
,
Sonnenborg
T. O.
,
Refsgaard
J. C.
,
Trolle
D.
,
Borgesen
C. D.
,
Olesen
J. E.
,
Jeppesen
E.
&
Jensen
K. H.
2016
Combined effects of climate models, hydrological model structures and land use scenarios on hydrological impacts of climate change
.
J. Hydrol.
535
,
301
317
.
Liang
W.
,
Bai
D.
,
Wang
F. Y.
,
Fu
B. J.
,
Yan
J. P.
,
Wang
S.
,
Yang
Y. T.
,
Long
D.
&
Feng
M.
2015
Quantifying the impacts of climate change and ecological restoration on streamflow changes based on a Budyko hydrological model in China's Loess Plateau
.
Water Resour. Res.
51
(
8
),
6500
6519
.
http://dx.doi.org/10.1002/2014WR016589
.
Liu
Q.
,
McVicar
T. R.
,
Yang
Z. F.
,
Donohue
R. J.
,
Liang
L. Q.
&
Yang
Y. T.
2016
The hydrological effects of varying vegetation characteristics in a temperate water-limited basin: development of the dynamic Budyko-Choudhury-Porporato (dBCP) model
.
J. Hydrol.
543
(
B
),
595
611
.
doi:10.1016/j.jhydrol.2016.10.035
.
Madsen
H.
,
Lawrence
D.
,
Lang
M.
,
Martinkova
M.
&
Kjeldsen
T. R.
2014
Review of trend analysis and climate change projections of extreme precipitation and floods in Europe
.
J. Hydrol.
519
,
3634
3650
.
Meyer
P. D.
,
Rockhold
M. L.
&
Gee
G. W.
1997
Uncertainty Analyses of Infiltration and Subsurface Flow and Transport for SDMP Sites
.
Office of Scientific & Technical Information Technical Reports, US Nuclear Regulatory Commission
,
Washington, DC
,
USA
.
Novotny
E. V.
&
Stefan
H. G.
2007
Stream flow in Minnesota: indicator of climate change
.
J. Hydrol.
334
,
319
333
.
Peng
X. D.
,
Shi
D. M.
,
Guo
H. Z.
,
Jiang
D.
,
Wang
S. S.
,
Li
Y. X.
&
Ding
W. B.
2015
Effect of urbanisation on the water retention function in the Three Gorges reservoir area, China
.
Catena
133
,
241
249
.
Pourmokhtarian
A.
,
Driscoll
C. T.
,
Campbell
J. L.
,
Hayhoe
K.
,
Stoner
A. M.
,
Adams
M. B.
,
Fernandez
I.
,
Burns
D.
,
Mitchell
M. J.
&
Shanley
J. B.
2017
Modeled ecohydrological responses to climate change at seven small watersheds in the northeastern U.S
.
Global Change Biology
23
(
2
),
840
856
.
Qi
S.
,
Sun
G.
,
Wang
Y.
,
McNulty
S. G.
&
Moore Myers
J. A.
2009
Streamflow response to climate and landuse changes in a coastal watershed in North Carolina
.
Trans. ASABE
52
(
3
),
739
749
.
Wang
K.
,
Wang
L.
,
Wei
Y. M.
&
Ye
M. S.
2013b
Beijing storm of July 21, 2012: observations and reflections
.
Nat. Hazards
67
,
969
974
.
Wigmosta
M. S.
,
Vail
L. W.
&
Lettenmaier
D. P.
1994
A distributed hydrology-vegetation model for complex terrain
.
Water Resour. Res.
30
,
1665
1679
.
Xia
J.
,
Zhang
Y. Y.
,
Xiong
L. H.
,
He
S.
,
Wang
L. F.
&
Yu
Z. B.
2017
Opportunities and challenges of the Sponge City construction related to urban water issues in China
.
Science China Earth Sciences
60
(
4
),
652
658
.
Zhang
L.
,
Dawes
W. R.
&
Walker
G. R.
2001
Response of mean annual evapotranspiration to vegetation changes at catchment scale
.
Water Resour. Res.
37
,
701
708
.
Zhang
W.
,
Pan
S. M.
,
Cao
L. G.
,
Cai
X.
,
Zhang
K. X.
,
Xu
Y. H.
&
Xu
W.
2015a
Changes in extreme climate events in eastern China during 1960–2013: a case study of the Huaihe River basin
.
Quatern. Int.
380
,
22
34
.