EN中文

Journal of Geographical Sciences >

2017 , Vol. 27 >Issue 7: 817 - 834

DOI: https://doi.org/10.1007/s11442-017-1408-2

Orginal Article

Applicability evaluation of the SWIM at river basins of the black soil region in Northeast China: A case study of the upper and middle Wuyuer River basin

  • YANG Zhiyuan , 1 ,
  • GAO Chao 2 ,
  • ZANG Shuying , 1 ,
  • YANG Xiuchun 3
Expand
  • 1. Key Laboratory of Remote Sensing Monitoring of Geographic Environment, Harbin Normal University, Harbin 150025, China
  • 2. College of Territorial Resources and Tourism, Anhui Normal University, Wuhu 241000, Anhui, China
  • 3. Institute of Agricultural Resources and Regional Planning, Chinese Academy of Agricultural Sciences, Beijing 100081, China

Author: Yang Zhiyuan, PhD Candidate, specialized in land use, regional environment modeling and evaluation. E-mail:

*Corresponding author: Zang Shuying, Professor, E-mail:

Received date: 2016-11-15

  Accepted date: 2017-01-20

  Online published: 2017-07-10

Supported by

National Natural Science Foundation of China, No.41571199, No.41571105

Natural Science Foundation of Heilongjiang Province, No.ZD201308

Copyright

Journal of Geographical Sciences, All Rights Reserved
Fold

Abstract

In this paper, we selected the middle and upper reaches of the Wuyuer River basin in the black soil region of Northeast China as the study area. We adopted the soil and water integrated model (SWIM) and evaluated the parameter sensitivity using partial correlation coefficient. We calibrated and validated our simulation results based on the daily runoff data from Yi’an hydrological station at the outlet of the river basin and the evaporation data recorded by various weather stations from 1961 to 1997. Following evaluation of the modeling data against the observed data, we present the applicability of SWIM in the river basin of the black soil region, and discuss the resulting errors and their probable causes. Results show that in the periods of calibration and validation, the Nash-Sutcliffe efficiency (NSE) coefficients of the monthly and daily runoffs were not less than 0.71 and 0.55, and the relative errors were less than 6.0%. Compared to daily runoffs, the simulation result of monthly runoffs was better. Additionally, the NSE coefficients of the potential monthly evaporation were not less than 0.81. Together, the results suggest that the calibrated SWIM can be utilized in various simulation analyses of runoffs on a monthly scale in the black soil region of Northeast China. On the contrary, the model had some limitations in simulating runoffs from snowmelt and frozen soil. Meanwhile, the stimulation data deviated from the measured data largely when applied to the years with spring and summer floods. The simulated annual runoffs were considerably higher than the measured data in the years with abrupt increases in annual precipitation. However, the model is capable of reproducing the changes in runoffs during flood seasons. In summary, this model can provide fundamental hydrological information for comprehensive management of the Wuyuer River basin water environment, and its application can be potentially extended to other river basins in the black soil region.

Key words： SWIM; Wuyuer River basin; sensitivity analysis; runoff; potential evapotranspiration; applicability evaluation

Cite this article

YANG Zhiyuan , GAO Chao , ZANG Shuying , YANG Xiuchun . Applicability evaluation of the SWIM at river basins of the black soil region in Northeast China: A case study of the upper and middle Wuyuer River basin[J]. Journal of Geographical Sciences, 2017 , 27(7) : 817 -834 . DOI: 10.1007/s11442-017-1408-2

1 Introduction

Presently, hydrological modelling is one of the primary methods used to study climate change and hydrological impacts of land use and land cover change (LUCC) (Gao et al., 2009). Distributed hydrological models have already become important tools in the research of hydrological processes and changes in river basins (Faramarzi et al., 2009). Based on physical parameters, a distributed ecohydrological model termed Soil and Water Integrated Model (SWIM) is a simulation tool that was developed from the Soil and Water Assessment Tool (SWAT) and MATSALU models by the Potsdam Institute for Climate Impact Research in Germany. This model integrates hydrology, vegetation, erosion and nutrient dynamics at the scale of river basin. In river basins between 100 km2 and 24000 km2, it can effectively describe the temporal and spatial variations in water balance components, soil nutrient cycling and transfer by runoff, processes related to vegetation or crop growth, characteristics of soil erosion and sediment transport dynamics, and the impacts of climate and land use changes on related processes. Compared with the SWAT model, SWIM places greater emphasis on the effects of land use and climate change on hydrological processes at the regional scale (Krysanova et al., 2011).
Since its development, SWIM has received much attention from many researchers (Stefanova et al., 2015; Conradt et al., 2013; Hesse et al., 2015; Hattermann et al., 2005; Krysanova et al., 2014; Krysanova et al., 2005; Ge et al., 2012; Wortmann et al., 2014; Tao et al., 2014; Huang et al., 2009). Domestic researchers have made many meaningful attempts to introduce SWIM. Gao et al. studied the optimization of the digital elevation model (DEM) resolution and the effects of different DEM resolutions on topographic parameters and runoff simulation in the Changtaiguan basin of the Huaihe River (Gao et al., 2012). The applicability of SWIM was also evaluated at different spatio-temporal scales and with different databases. The case study in the upper area of Bengbu hydrological station at the Huaihe River demonstrated that SWIM was more suitable for establishing the rainfall-runoff relationship of small watersheds with areas less than 1 × 104 km2, indicating that this model may be more appropriate for accurately simulating and depicting hydrological processes of small basins with comprehensive data (Gao et al., 2013; Zhang, 2015). In addition, SWIM was applied to simulate (Zhang et al., 2011; Zhang et al., 2015; Mo, 2008) and quantitatively analyze (Xu, 2011; Zhang 2011; Zhang, 2015) the hydrological impacts of climate and land use change.
The black soil region of Northeast China, also known as the black soil area of the cold region, is located at high latitudes. The long and cold winter is a distinctive climate feature, and the runoff formation is greatly influenced by temperature. In the runoff process, spring and summer floods are two obvious flood seasons. In terms of hydrological characteristics, the rivers here are also quite different from the rivers in non-cold regions. As an important production base for commodity grains in China, the black soil region has undergone years of high strength reclamation. With its unique climate and geomorphological conditions (Cui et al., 2008), the underlying surfaces have been greatly affected by natural changes and human activities. Consequently, water related eco-environmental problems such as water and soil erosion, and drought and flood disasters have become quite prominent (Liu and Liu, 2006). Therefore, from a water cycle and eco-hydrological perspective, there is an urgent need to propose scientific strategies for the construction and protection of the ecological environment. SWIM has been applied in the humid areas of southern China as well as the semi-arid and semi-humid areas of northwestern China. However, the applicability evaluation of SWIM has rarely been reported in the black soil region of Northeast China. In this paper, we selected the upper and middle reaches of the Wuyuer River basin as the study area, which is representative in terms of topography, soil, climate, soil and water conservation, and evaluated the applicability of SWIM in the black soil region of Northeast China. Our research can potentially provide a scientific basis for the promotion and application of the model, integrated water resource management, and drought and disaster relief.

2 Overview of the study area

The Wuyuer River, the largest inland river in Heilongjiang Province, is located in the western part of the province. It starts in the transitional zone between the western foot of the Xiao Hinggan Mountains, Bei’an City and Songnen Plain, passes through cities and counties including Bei’an, Keshan, Kedong, Baiquan, Yi’an and Fuyu, and eventually flows into Zhalong wetland. The landscape of Wuyuer River is high in the northeast and low in the southwest, the total length is 587 km, and the basin area is about 15,000 km2. The river basin, located in an area with prevailing continental monsoon, has a typical inland semi-arid climate. It is dry and cold in winter, warm and rainy in summer, and dry and windy with drastic temperature changes in spring and autumn. Precipitation is the primary source of runoff, followed by ice melt and snow. The annual precipitation is 496.7 mm and is mainly concentrated during the flood season from June to September. Precipitation during the flood season, with strong intensity and short duration, accounts for 80% of the total annual precipitation. The rise and fall of the landscape is relatively small. Although the slope is gentle, the surface of the slope is long. This leads to increased drainage area and strong scouring by runoff, resulting in severe water and soil erosion, and deterioration of the physicochemical properties of the black soil. The three representative hydrological stations are Bei’an, Yi’an and Long’anqiao. The upper and middle reaches, located between the source and Yi’an hydrological station (hereafter Yi’an Station), are the main runoff areas. Huge areas of riverside marshland have developed downstream of Yi’an Station, which is the diffuse area of the runoff from the upper and middle reaches. Zhalong wetland, on the list of wetlands of international importance, is located within this area. Considering the spatial heterogeneity of topography, vegetation, soil and precipitation, we chose the region above Yi’an Station in the Wuyuer River basin as the study area (Figure 1), which is located in the center of the black soil region in Heilongjiang. In addition, we selected Yi’an Station as the water outlet of the river basin. The drainage area of the study is 8296.33 km2, and the area of water and soil erosion is 5788.70 km2.
Figure 1 Sketch map of the geographic location and spatial division of the study area

3 Data and methods

3.1 Basic data

The input data for SWIM mainly included DEM, land use, soil, hydrological and meteorological data.
3.1.1 Acquisition and analysis of the spatial data
We adopted the Albers equal-area conic projection and the Krasovsky ellipsoid as the reference system. Because data with variations in spatial resolution would affect the accuracy of the model (Zhang, 2015), we considered the study area (Krysanova et al., 2011) and hence unified the spatial resolution as 90 m.
(1) DEM data. The DEM data were obtained from the Computer Network Information Center, Chinese Academy of Sciences Scientific Data Mirror Website (http://datamirror.csdb.cn). Because DEM resolution is associated with the river basin area, we used the 90 m spatial resolution based on the “thousand-million” rule suggested by Maidment (1996). We achieved a seamless mosaic of the overlapping rasters through coordinates conversion and data examination of the downloaded split DEM rasters. The mosaiced DEM raster and the coordinates of the water outlets were loaded into Mapwindow GIS, and thus the final DEM data of the study area were obtained.
(2) Land use data. We acquired the grid data with a 30 m resolution through interpreting the Thematic Mapper (TM) remote sensing data using human-computer interaction with an accuracy of above 90%. The land use data of the study area were generated according to the basin boundaries. We recollected the data points in ArcGIS at a resolution of 90 m in order to achieve a unified spatial resolution. With reference to the guidelines for SWIM (Krysanova et al., 2011) and combining the actual conditions of Wuyuer River basin, we reclassified the codes of land use data based on the categories of land use in SWIM.
(3) Soil data. The soil data were from the 1:1,000,000 scale China Soil Database of Institute of Soil Science, Chinese Academy of Sciences. We obtained the soil data of the study area according to the basin boundaries and converted the data into GRID format with each grid cell as 90 m × 90 m. Because our national soil type standards could not be directly applied to establish a SWIM soil database, we did a series of conversions on the soil geophysical parameters, including soil thickness, bulk density, porosity, available water capacity, field capacity and saturated hydraulic conductivity. We converted the soil particle size using the cubic spline interpolation method in Matlab software. We also estimated the soil organic matter using the SPAW software, which was developed by the United States Department of Agriculture and used to analyze soil water characteristics. In addition, we reclassified the 12 soil types within the basin and finally acquired the soil type map that could be recognized by SWIM.
3.1.2 Acquisition and processing of the attribute data
(1) The runoff data from Yi’an Station were obtained from the Chinese Hydrological Yearbooks and also provided by Heilongjiang Hydrology Bureau. According to the input data format of SWIM, we used m3/s as the measuring unit for daily runoff.
(2) The meteorological data were from the China Meteorological Data Center (http://data.cma.cn/). We obtained the daily monitoring data of six weather stations from 1975 to 1997. The daily monitoring data included the highest, average and lowest temperature, precipitation, relative humidity, atmospheric pressure, vapor pressure of water, sunshine hours, wind speed and atmospheric radiation. The six weather stations were Bei’an, Keshan, Yi’an, Fuyu, Hailun and Mingshui stations. Meanwhile, we collected the data of small evaporating dishes at Kedong weather station from 1986 to 1994. We also calculated the total daily solar radiation using sunshine hours. The above data were respectively arranged according to the input data formats of the model.

3.2 Research methods

3.2.1 Method to determine the spatial disaggregation and the drainage area threshold
SWIM uses a similar three-level disaggregation scheme as the MATSALU model, which includes basin, sub-basins and hydrologic response units (HRUs). The sub-basins were obtained with certain threshold (areas) based on the digital landscape, and then the HRUs were defined by overlaying the sub-basin, land use map and soil maps. A HRU has the same types of land use, soil and hydrological response characteristics. The model is operated independently on each HRU, and then the cross-sectional aggregate of basin outlet is formed through the confluence of river channels (Krysanova et al., 2011).
The drainage area threshold determined the density of the digital river network and the information of the extracted sub-basin, including the position of water outlet, the size and the distribution. Since different thresholds would result in different basin hydrological characteristics and change the corresponding simulation results, the principle to determine the best drainage area threshold is to extract a digital river network maximally close to the virtual river network. We used the constant threshold method according to the size of the study area and empirical experience on threshold determination (Zhang, 2015). The thresholds tested were 90 km2, 100 km2, 140 km2, 150 km2, 160 km2 and 170 km2. A hydrographic map at a 1:250,000 scale was imported into MapWindow GIS to facilitate the extraction of the digital river network, which could reflect the virtual situation. The model was run respectively with the obtained sub-basin, land use and soil maps. Meanwhile, the daily runoffs of Yi’an Station from 1961 to 1974 were simulated using preliminarily calibrated parameters. The simulation results were also compared with the measured data in the same period, and the correlation coefficient R and efficiency coefficient E were calculated (Table 1). Finally, the optimized drainage area was determined to be 150 km2, and subsequently, the digital river network was generated which included 27 sub-basins and 556 HRUs (Figure 1).
Table 1 The characteristic parameters of sub-basins and the simulation accuracy of runoffs with different drainage area thresholds
Drainage area threshold (km2) Number of sub-basin Number of hydrologic response unit (HRU) Average annual
runoff (m3/s)		 R value Efficiency
coefficient (E)
90 93 1206 29.88 0.834 0.39
100 37 653 29.27 0.896 0.43
140 29 577 27.69 0.898 0.47
150 27 556 27.73 0.899 0.48
160 27 556 27.67 0.898 0.48
170 25 525 29.30 0.898 0.46
3.2.2 Sensitivity analysis and applicability evaluation of the model
Sensitivity analysis is critical to the construction and application of hydrological models. It aims to determine the degree to which individual parameters influence the simulation results, this can then be used to remove unimportant parameters, reduce parameter dimensions and the impact of parameter uncertainty, and thus improve the application accuracy of the model (Song et al., 2015). Common methods used for sensitivity analysis in hydrological models include screening analysis, regression analysis, variance-based analysis and agent-based model. Among these methods, regression analysis can analyze the sensitivity of a single input parameter under the condition that all inputs simultaneously affect the output, and as such, it can describe the relationship between the input and output. In addition, it is also easy to use (Kong et al., 2011). In this study, we adopted the partial correlation coefficient between different input parameters and the corresponding Nash-Sutcliffe efficiency (NSE) coefficient (E) and the relative error of runoff (r) as the evaluation criteria of the sensitivity analysis, where the NSE coefficient and relative error of runoff represented the simulation accuracy. The partial correlation coefficient between the specific input parameter Xj and the output parameter Y is described as follows.
$Co{{r}_{{{X}_{j}}Y}}=\frac{\sum\limits_{i=1}^{m}{({{X}_{ij}}-{{{\bar{X}}}_{j}})({{Y}_{i}}-\bar{Y})}}{\sqrt[2]{\sum\limits_{i=1}^{m}{{{({{X}_{ij}}-{{{\bar{X}}}_{j}})}^{2}}}}\sqrt[2]{\sum\limits_{i=1}^{m}{{{({{Y}_{i}}-\bar{Y})}^{2}}}}}$(1)
where Xj is the input parameter, and Y is the relative error of runoff or NSE coefficient E. $\overline X_j$. is the average of input variable Xj, and $\overline Y$ is the average of output variable Y. $\bar{Y}=\sum\nolimits_{i}{{{Y}_{i}}/m,\ {{{\bar{X}}}_{j}}}=\sum\nolimits_{i}{{{X}_{ij}}/m},$i is the number of observations (i = 1,…, m). Xij is the single output obtained from running the model separately for i times with the input parameter Xj given a different value each time. $-1\leq Cor_{X_{j}Y}\leq 1$, the value $ Cor_{X_{j}Y}$ is considered as a positive correlation if it is higher than 0. If the value is less than 0, it is a negative correlation. The closer the absolute value of the partial correlation coefficient is to 1, the closer the relationship between the two factors. On the contrary, the closer the value is to 0, the less close the relationship between the two factors.
In this paper, the applicability of SWIM was evaluated with the NSE coefficient (E) of annual runoff and the relative error of annual runoff (r) (Equations 2 and 3).
The NSE coefficient (E) of annual runoff is calculated as follows.
$E=\text{1}-\frac{\sum\limits_{t}{{{\left( Qob{{s}_{i}}-Qsi{{m}_{i}} \right)}^{\text{2}}}}}{\sum\limits_{t}{{{\left( Qob{{s}_{i}}-\overline{Q}obs \right)}^{\text{2}}}}}$. (2)
where Qobs was the observed runoff. Qsim was the simulated runoff. $\overline{Q}obs$ was the average of the observed annual runoff of multiple years. t is the time. The closer the E value is to the maximum value 1, the higher the simulation accuracy.
The relative error of runoff (r) was calculated as follows.
$r=\frac{\sum{Qsim-\sum{Qobs}}}{\sum{Qobs}}\times \text{100 }\!\!%\!\!\text{ }$. (3)
The lower the r value, the higher the simulation accuracy. A positive r value signified that the simulated runoff was higher than the observed runoff, and the converse for a negative r value.

4 Simulation results and applicability analysis

4.1 Sensitivity analysis and parameter calibration

SWIM, which is a physically based distributed hydrological model, contains large numbers of coupled empirical and conceptual mathematical equations with many parameters. Parameter uncertainty can cause large discrepancies in the simulation results. However, it is rather difficult to simultaneously increase the accuracy of every parameter. Sensitivity analysis can help identify the important parameters that affect the simulation results, and thus help prevent indiscriminate adjustment of parameters. Additionally, it can improve the efficiency of parameter optimization and calibration.
In this study, the parameters used in the sensitivity analysis were mainly from the wipper.bsn file (Krysanova et al., 2011), the user manual of SWIM (Krysanova et al., 2011) and relevant publications (Xu, 2011; Zhang, 2011; Shu et al., 2008; Rougier et al., 2005). In consideration of the hydrological, climate and geographic characteristics of the black soil region, we removed attributes with low levels of sensitivity and focused on attributes with higher levels of sensitivity (Yang et al., 2004; Cheng et al., 2009). We adjusted a range of key parameters and manually assigned values through the artificial perturbation analysis method, and we also analyzed and validated the impact of changing specific parameters on the output results. Based on the simulated daily runoffs of Yi’an Station from 1961 to 1974, we calculated the partial correlation coefficient between an assigned parameter and the corresponding NSE coefficient (E) and the relative error of runoff (r), which both represented the simulation accuracy. Through sensitivity analysis, we determined eight key parameters that showed strong impacts on the simulation accuracy of the runoffs in the upper and middle reaches of the Wuyuer River (Figure 2 and Table 2).
Figure 2 Results of the sensitivity analysis
Table 2 Calibrated values of the key parameters
Parameter code Parameter
meaning		 Range of
value		 Parameter
value		 Parameter value of the
upstream Jinghe River
thc
CN		 Correction factor for potential evapotranspiration on sky emissivity SCS curve number 0.5-1.5
10-100		 0.7
CN=70		 1.6
-
bff Baseflow factor for basin 0.2-1.0 0.7 0.01
gwq0 Initial groundwater flow contribution to streamflow 0.01-1.0 0.03 0.5
abf0 Alpha factor for groundwater 0.001-1.0 0.001 0.001
roc2 Routing coefficient 1-100 roc2=1.5 roc2=0.5
roc4 Routing coefficient 1-100 roc4=3 roc4=2
sccor Correction factor for soil saturated conductivity 0.01-10 1.8 5
The calibration results of the eight key parameters that strongly influenced the simulation accuracy of runoffs are shown in Table 2. We also considered the impact of ice and snowmelt on runoffs and calibrated the snowfall and snowmelt parameters as well. As evident in Table 2, compared with the sensitive parameters in the upper reach of the Jinghe River basin of the yellow soil region in northwestern China, the sensitivity of SWIM parameters showed great variance both in the river basin of the black soil region and in the upper reaches of the Jinghe River basin, which is located in the transitional area of the semi-humid and semi-arid temperate zones (Zhang, 2011). This indicated that the sensitivity analysis was dependent on river basin structures such as land use, soil and climate, and the results reflected specific characteristics of the river basin.

4.2 Analysis of simulation results

4.2.1 Simulation results of runoffs
In the initial phase, many parameters with 0 value would greatly impact the simulation results. In order to reduce the error, it was necessary to reasonably estimate the initial values of the parameters. Therefore, the years from 1957 to 1960 were selected as the precursory period to determine the appropriate initial values. The years from 1961 to 1974 were selected as the calibration period for the parameters, and the years from 1975 to 1985 were selected as the validation period to evaluate the applicability of the model. The upper reach of the Wuyuer River basin has a history of reclamation spanning 100 years, and the land was primarily used for agriculture. Since the 1960s, the whole region entered the peak period for agricultural land development and hydraulic engineering construction. Based on the existing research data (Bai et al., 2007) and compared with the land use profile of the study area in 1980, we found that the water area, and residential and agricultural lands were relatively stable in the study area. Because information regarding land use and land cover prior to 1980 was hard to obtain, we used the land use data in 1980 as the input.
In the calibration period, the simulation results of runoffs at Yi’an Station are presented in Figures 3a and 3c. The NSE coefficients of the daily and monthly runoffs were 0.57 and 0.73, respectively, and the relative error of the total runoff (r) was 1.4%. In the validation period, the NSE coefficients of daily and monthly runoffs were 0.55 and 0.71, respectively, and the relative error of the total runoff (r) was 5.9% (Figures 3b and 3d). These results demonstrate that after calibration, the simulation results of both daily and monthly runoffs match very well with the curve of the observed data. In both calibration and validation periods, the simulation efficiencies of SWIM on the daily and monthly runoffs met the evaluation criteria. However, the simulation efficiency in daily runoffs was not very satisfactory. The simulation efficiency of SWIM in monthly runoffs was superior compared to daily runoffs in the study area.
Figure 3 The comparisons between the simulated and observed daily and monthly runoffs of Yi’an Station from 1961 to 1985
The curves of average monthly flow at Yi’an Station from 1961 to 1985 indicate that the Wuyuer River basin is located in a cold area, and the runoffs have two characteristic peaks that correspond to the spring and summer flood seasons (Figure 4a). From April to May, the runoffs increased significantly because of snowmelt and the thawing of frozen soil, resulting in the spring flood. Characterized by short duration and high intensity, precipitation concentrated in the period from June to September, which led to the summer flood. SWIM did not produce satisfactory simulation results in the years with both spring and summer floods. During the spring flood, the simulated runoff was less than the observed data. During the summer flood, the simulated runoff was greater than the observed data. Generally, the model can still reproduce the change of flux in the flood season. The analyses of average monthly flow and average annual runoff showed that SWIM was more suitable for the study area after calibration. It can be applied to various runoff-related simulation analyses on a monthly scale.
Figure 4 The change in average monthly flows and annual runoffs of Yi’an Station from 1961 to 1985
The possible factors that influenced the simulation efficiency of the daily runoffs at the Yi’an Station include the following:
(1) Selection of the validation period. From 1964 to 1973, Yi’an Station was in an obvious fluctuation period, and the low-flow and high-flow periods appeared alternately. The period from 1974 to 1982 was a distinct low-flow period (Figures 3 and 4). The low-flow and high-flow years were not evenly distributed. Abbaspour et al. reported that the best simulation effect requires even distribution of low-flow year and high-flow year in the calibration and validation periods (Abbaspoour et al., 2007).
(2) Model structural error. SWIM uses many mathematical equations to simplify the actual water cycle in the basin, and thus errors are inevitable in this process. The simulation results exhibited lower simulated peak values in the spring flood season and higher in the summer flood season than the observed data. Other researchers reported similar simulation results for runoffs in the study area using other hydrologic models (Guo et al., 2016; Feng et al., 2010). In addition, the difference in flood peak value between the simulated and observed data in certain years such as 1961 and 1962 was relatively large (Figure 4b). Due to the model structure, it is possible that the difference between the simulated and observed data in flood seasons originated from unclear depiction of the physical process of runoff production from melting of snow and seasonal frozen soil. Further, it may have originated from applying the general precipitation-runoff mechanism on precipitation marked by strong intensity and short duration. The reason for the large difference in annual flood peak value between the simulated and observed results in certain years was probably mainly due to the absence of a reservoir module in SWIM. To resolve these issues, we still need to conduct an in-depth study and perfect the model structure in the future. According to relevant hydraulic and hydrological references, the largest 7-day flood at Yi’an Station started on July 30, 1961 and July 26, 1962, respectively. During the late 1950s to the early 1960s, a large number of embankments and reservoirs were constructed in the upper and middle reaches of the Wuyuer River, including Wuyuer River Northern Dam, Xianfeng Reservoir, Bao’an Reservoir, Xinshuguang Reservoir and Hongwei Reservoir. They all had their gates opened and discharged floodwater into the main river channel during the flood period, and as such, the daily observed runoffs increased.
(3) Random errors resulting from the abnormally high simulation values of runoff in certain years. For example, the simulation values of the years 1966, 1968, 1974, 1980 and 1982 were greater than the observed results, which increased the overall error of the simulation results for the whole period. In order to analyze the source of these abnormally high simulation values of runoff, we conducted statistical analysis of precipitation at Yi’an weather station from 1961 to 1985. We found that both average annual and monthly precipitation values of the above years were higher than those from adjacent years. The increment of annual precipitation was 115.79 mm, and monthly precipitation was 10.61 mm. Meanwhile, the simulation data of the annual runoff in the above years were several folds higher than the observed data, which produced continuous influence on the simulation results of the subsequent years (Figure 4b). This result was consistent with outcomes reported by others (Tzyy-Woei, 2003; Song and Ma, 2007). Nevertheless, this discrepancy also indicated that the simulation accuracy of years with an abrupt increase in annual precipitation was not sufficient. As such, further adjustment and analysis of abnormally high precipitation values in certain years is necessary for runoff-related simulations.
4.2.2 Analysis and validation of the simulation results of potential evapotranspiration
In order to further validate the applicability and reliability of the calibrated physical parameters, observed data including daily precipitation and runoff at Yi’an Station from 1986 to 1997 were selected. The monthly potential evapotranspiration during this period was simulated based on the 1995 land use data, and the simulated value of the Yi’an Station sub-basin, located at the overall outlet of the basin, was compared with the monthly evaporation data measured with small evaporation dishes at weather stations. The detailed simulation results are presented in Figure 5a. The simulated data matched well with the measured data with a NSE coefficient (E) of 0.81.
Figure 5 Comparison between the simulated monthly potential evapotranspiration and the measured monthly evaporation of small evaporating dish at Yi’an and Kedong weather stations from 1986 to 1997
We validated the simulated runoff and potential evapotranspiration at the overall outlet. However, this did not guarantee simulation accuracies of water balance components such as the runoff in other areas of the basin. Therefore, the outputs from more sub-basin outlets are needed to validate the applicability of the model. The measured data from small evaporating dishes at Kedong weather station from 1986 to 1994 were compared with the simulation results of potential evapotranspiration at the sub-basin where Kedong weather station is located (Figure 5b). The resulting NSE coefficient (E) was 0.87. In summary, these results demonstrate that SWIM can reasonably represent the evapotranspiration capability in the upper and middle reaches of the Wuyuer River after calibration.
Based on the calculations from SWIM, the potential evapotranspiration in the upper and middle reaches of the Wuyuer River varied significantly during the year (Figure 6a). The annual distribution of monthly potential evapotranspiration was lower in winter and higher in summer. The average monthly potential evapotranspiration from 1986 to 1997 had a minimum of 9.70 mm, which occurred in January 1990. The highest average monthly potential evapotranspiration was 230.89 mm, which occurred in June 1988. The average annual potential evapotranspiration from 1986 to 1997 was 1150.31 mm, and the average annual precipitation was 520.42 mm. The amount of evapotranspiration was 2.21 times that of precipi- tation. The evapotranspiration mainly occurred between April and September, accounting for 81.25% of the annual amount. The largest evapotranspiration occurred in June. Evapotranspiration in December and January was the weakest, together accounting for 1.99% of the annual amount. Monthly evapotranspiration was always larger than monthly precipitation, indicating that this basin has a relatively dry climate and is drought prone.
Figure 6 The distribution of monthly potential evapotranspiration and the change in annual potential evapotranspiration over years from 1986 to 1997
The change in annual potential evapotranspiration in the study area was around 150 mm from 1986 to 1997 (Figure 6b). The highest annual average was 1222.77 mm in 1989, and the lowest was 1077.26 mm in 1988. The climate tendency rate of potential evapotranspiration was 3.86 mm/a, indicating that potential evapotranspiration was increasing in the study area. In terms of seasons, potential evapotranspiration in spring, summer, autumn and winter was 1.87, 0.81, 0.38 and 1.22 mm/a, respectively (Figure 6b). This indicated that potential evapotranspiration was increasing in all seasons. However, the trend of increase in summer and fall was relatively weak, and the weakest was found in autumn. Potential evapotranspiration was the highest in summer from June to August, the second highest in spring from March to May, the second lowest in autumn from September to November and the lowest in winter from December to the following February, which accounted for 32.0%, 45.4%, 18.3% and 4.4% of the total amount, respectively. Similar results were obtained via statistical analysis (Zhang et al., 2011). In the 12-year period, the average annual increment of potential evapotranspiration was 50 mm. The years with the obvious change included 1988, 1991, 1986 and 1989, where the most obvious change occurred in 1988 and 1991 (Figure 7). On the other hand, the spatial distribution of annual potential evapotranspiration was obvious. In general, it declined from the southwest and northeast towards the center of the study area. The high-value area appeared from the southwest of Keshan County to the northwest of Yi’an County, along with the majority of the area in Bei’an City. The low-value area appeared in the east of Keshan County and Kedong County. The change in evapotranspiration was most obvious in the southern part of Baiquan County.
Figure 7 The spatial distribution of potential evapotranspiration in the study area

5 Conclusions

In this study, we introduced SWIM and selected water balance components such as runoff and potential evapotranspiration to validate the simulation results from multiple stations, and with various parameters, for a preliminary evaluation of the applicability of SWIM at the river basin of the black soil region in Northeast China. Based on our analyses of the simulation results, we conclude the following.
(1) We used the daily precipitation and runoff from Yi’an Station, located at the basin outlet, to calibrate and validate the parameters, and we demonstrated that the simulated results of daily runoffs matched well with the observed data. During the calibration and validation period, the NSE coefficients of the monthly and daily runoffs were greater than 0.71 and 0.55, respectively, and the relative error was within 6.0%. The simulation results of runoffs were reliable. Moreover, the simulation results of monthly runoffs were better than those of daily runoffs.
(2) We simulated the monthly potential evapotranspiration at sub-basins where Yi’an and Kedong stations are located. We further compared the simulation results with the measured monthly evaporation from small evaporating dishes at weather stations. The NSE coefficients were above 0.81.
(3) The application of the model had uncertainties and limitations. First, the model had some limitations in simulating the runoff from snowmelt and frozen soil. The simulation results of runoffs from spring flood, produced by snowmelt and the thawing of frozen soil, were not ideal. The simulation results were all lower than the observed data. Second, the model could not adequately simulate the runoffs in years with both spring and summer floods, although it was able to essentially represent the flow change in the flood seasons. Lastly, regarding the years with abrupt increases in annual precipitation, the simulation results of annual runoffs became abnormal. The simulation results of annual runoff were several folds higher than the observed data. Meanwhile, the years with the abrupt increases in precipitation had continuous influence on the simulation results of the subsequent years, which directly led to the relatively large error for the whole period.
(4) Following calibration, SWIM can be applied in the various runoff-related simulations in the Wuyuer River basin on a monthly scale. Our simulation results have certain reference values. They not only provide the hydrological basis for integrated management of the Wuyuer River basin environment, but the applicability may be extended to other river basins in the black soil region.

6 Discussion

SWIM takes the spatio-temporal heterogeneity of basin factors such as climate and underlying surface into consideration, and has important physical and hydrological significance. As a tool for evaluating the effects of climate and land usage on basin hydrology, it has been widely used in simulating hydrology and water quality in basins of different scales in Europe. In this paper, we introduced SWIM to evaluate the initial applicability of the model using the upper and middle reaches of the Wuyuer River as the study area. The results indicate that it is feasible to apply SWIM to simulate the runoffs related to hydrological processes. The model has potential application values in analyzing the effect of climate and land use change on basin hydrology, evaluating basin drought and flood conditions, and managing water resources in the black soil region of Northeast China. It is therefore necessary to further investigate the applicability of SWIM in this region, as well as its application in water resource management.
SWIM abstracts complex hydrological phenomena and processes. In order to describe the actual hydrological processes, it uses mathematical and physical equations to describe in detail the surface and subsurface runoff processes within each individual HRU, and obtain the cross-sectional aggregate of basin outlet from the confluence of river channels. The model requires various input data including meteorology, hydrology, DEM, soil and land use, and it validates the simulation results with the measured data. This increases the uncertainty of simulations. Spatial input data, such as the DEM accuracy and the grid size, the number of sub-basins, the accuracy of the soil and land use data, the spatial distribution and density of weather stations, and the accurate description of related basin features, all determine the results of hydrological simulations (Fitzhugh and Mackay, 2001; Romanowicz et al., 2005). Existing research has shown that for simulation with hydrological models, finer spatial resolution may not generate higher simulation accuracy or the best simulation effect if input information is incomplete (Li et al., 2005). In order to increase the simulation accuracy of models, it is therefore critical to determine the optimal spatial input resolution, the optimal drainage area threshold and the accurate extraction of sub-basin parameter information that are suitable for the study area.
Simulation accuracy is not only directly related to the grid size of the spatial input data, but it is also determined by optimal model parameters. Therefore, optimal calibration of parameters becomes an important precondition for simulation accuracy. SWIM has numerous parameters, and it is difficult to achieve auto-calibration. In addition, the majority of the parameters have specific physical meanings and can reflect the spatial heterogeneity of the basin and the complex physical hydrological processes to some degree. This certainly increases the difficulties associated with determining optimal model parameters. Presently, most of the parameter calibrations of SWIM are primarily determined manually based on experiences. This means that the range of initial parameters is decided empirically first based on the physical meaning of parameters, the model is run and the outputs are calculated. Subsequently, parameter optimization is done by comparing the simulated data with the measured data. Finally, optimal parameter values are determined according to the constrained target. Optimal parameters may not be obtained due to the equifinality for different parameters, which is caused by more parameters and less information. Hence, parameter calibration should focus more on the combination of both manual and automatic calibration, although methods that incorporate auto-calibration with practical experiences need to be further explored.
Due to the presence of extensive frozen soil in the black soil region of Northeast China, its hydrological characteristics are notably different from the areas without frozen soil. Seasonal frozen soil has deep influence on upper soil moisture content, soil evaporation capacity and soil penetration, which then affect basin confluence, and thus consequently affecting runoffs (Wang et al., 2006). The hydrological processes modeled by SWIM are complex and can depict hydrological features of basins in relatively fine detail. However, the results of simulated daily runoff in this study showed that SWIM was not ideal for simulating spring flood runoffs produced by snowmelt and the thawing of frozen soil. The simulation values were all less than measured results, which may be closely related to the hydrological characteristics of the black soil region of Northeast China. This also indicated that the model still has some limitations in simulating runoffs from snowmelt and frozen soil, and as such, further investigation and refinement of the model structure are required.

The authors have declared that no competing interests exist.

References

Publishing order | Descend order by publishing year | Descend order by cited within
[1]
Abbaspour K, Vejdani M, Haghighat S, 2007. SWAT-CUP calibration and uncertainty programs for SWAT. In: Modsim 2007 International Congress on Modelling and Simulation, Modelling and Simulation Society of Australia and New Zealand, 2007: 1596-1602.Abstract EXTENDED ABSTRACT Distributed watershed models are increasingly being used to support decisions about alternative management strategies in the areas of landuse change, climate change, water allocation, and pollution control. For this reason it is important that these models pass through a careful calibration and uncertainty analysis. Furthermore, as calibration model parameters are always conditional in nature the meaning of a calibrated model, its domain of use, and its uncertainty should be clear to both the analyst and the decision maker. Large-scale distributed models are particularly difficult to calibrate and to interpret the calibration because of large model uncertainty, input uncertainty, and parameter non-uniqueness. To perform calibration and uncertainty analysis, in recent years many procedures have become available. As only one technique cannot be applied to all situations and different projects can benefit from different procedures, we have linked, for the time being, three programs to the hydrologic simulator Soil and Water Assessment Tools (SWAT) (Arnold et al., 1998) under the same platform, SWAT-CUP (SWAT Calibration Uncertainty Procedures). These procedures include: Generalized Likelihood Uncertainty Estimation (GLUE) (Beven and Binley, 1992), Parameter Solution (ParaSol) (van Griensven and Meixner, 2006), and Sequential Uncertainty FItting (SUFI-2) (Abbaspour, et al., 2007). In this paper we describe SWAT-CUP and the three procedures and provide an application example using SUFI-2. Inverse modelling (IM) has often been used to denote a calibration procedure which uses measured data to optimize an objective function for the purpose of finding the best parameters. In recent years IM has become a very popular method for calibration. IM is concerned with the problem of making inferences about physical systems from measured output variables of the model (e.g., river discharge, sediment concentration). This is attractive because direct measurement of parameters describing the physical system is time consuming, costly, tedious, and often has limited applicability. In large-scale distributed applications most parameters are almost impossible to measure as they are lumped and; hence, do not carry the same physical meaning as they did in their small-scale applications. For example, soil parameters such as hydraulic conductivity, bulk density, water storage capacity are but fitting parameters in the large scale. Because nearly all measurements are subject to some uncertainty and the models are only approximations, the inferences are usually statistical in nature. Furthermore, because one can only measure a limited number of (noisy) data and physical systems are usually modelled by continuum equations, no hydrological inverse problem is really uniquely solvable. In other words, if there is a single model that fits the measurements there will be many of them and a large number of parameter combinations can lead to acceptable modelling results. Our goal in inverse modelling is then to characterize the set of models, mainly through assigning distributions (uncertainties) to the parameters, which fit the data and satisfy our presumptions as well as other prior information. To make the parameter inferences quantitative, one must consider 1) the error in the measured data (driving variables such as rainfall and temperature), 2) the error in the measured variables used in model calibration (e.g., river discharges and sediment concentrations, nutrient loads, etc.), and 3) the error in the conceptual model (i.e., inclusion of all the physics in the model that contributes significantly to the data). The latter uncertainty could especially be large in large-scale watershed models.

DOI

[2]
Bai Shuying, Zhang Shuwen, Zhang Yangzhen, 2007. Analyzing dynamic process of land use change in Songnen Plain of China: A case study in Duerbote Mongolian autonomous county of Daqing city.Resources Science, 29(4): 164-169. (in Chinese)Today,many environmental problems facing people are related to land use/land cover change.It is very important to hold a general and objective viewpoint to the detailed dynamic process of land use/land cover change.In this paper,Duerbote Mongolian Autonomous County in Daqing City was selected as a case study area,five periods of land use data from 1954 to 2001 is used to study the land use change process during the past 50 years by recording every patch of land use type based on remote sensing and GIS technology.The patches of land use change in a certain period were classified into three types of land use conversion,namely,unchanged type,human disturbance type,and natural evolution type.The stability of land use tendency in the process of land use conversion was also analyzed.The results are as follows: 1) Different land use type has different stability of land use conversion.Water area,urban areas,residential areas and cultivated land are the most stable one;degenerated saline-alkali land and other hardly used land are in the second place;wetland,woodland and grassland are most unstable one.This indicates that once the land is reclaimed and used as cultivated land,they will not be abandoned in a long period.Water area and wetland couldn't easily be changed for other purpose;yet degraded saline-alkali land was with a little change because of its difficult improvement.Wetland and grassland had a good degree of transformation because of their unstable state where the conversion between grassland,wetland,water areas and saline-alkali land was mainly affected by precipitation;2) the stability of land use tendency also varies from different land use change types.If the conversion is from unchanged type or natural evolution type to human disturbance type,then,the patch of land use type tends stable,and there is only a remote possibility of land use change in the next period;on the contrary,if the conversion of land use change is from unchanged type or human disturbance type to natural evolution type,the patch of land use type tends unstable in the next period,which indicate land use conversion tends to be the type of maximizing its value of productivity.This method that distinguished the influence of people's behaviors on land use and natural change is helpful for showing the laws of regional land use/cover change.

DOI

[3]
Cheng Lei, Xu Zongxue, Luo Ruiet al., 2009. SWAT application in arid and semi-arid region: A case study in the Kuye river basin.Geographical Research, 28(1): 65-73. (in Chinese)The SWAT is a physically-based distributed hydrological model.Its runoff generation mechanism is more practical.The methods adopted in the process of runoff including surface runoff,interflow and groundwater flow are applicable to various conditions of climate and underlying surface.In this study,the SWAT(Version 2005) is applied to the Kuye River basin,one of typical watersheds with plentiful and coarse sand in the middle reaches of the Yellow River with arid and semi-arid climate.When the model in the Kuye River basin was developed,automatic calibration method within the SWAT Version 2005 was used to calibrate the model.According to the stream flow hydrograph at Wenjiachuan station,the mouth of the river,the parameters were further adjusted.Then,daily and monthly discharges from 1980 to 1985 have been simulated in the study area,and observed data series of three hydrological stations(Wenjiachuan,Xinmiao and Wangdaohengta) are used to evaluate the simulated discharge series of SWAT.The effects of physical mechanism for streamflow generation processes have been analyzed and discussed.The result shows that the relative error of water budget is about 10%-20%,while the Nash-Sutcliffe efficiency coefficient(Ens) is relatively low,with Ens of daily series being about 0.2 and monthly Ens around 0.6.Preliminary analysis of simulation results showed that the SWAT is not effective to simulate the discharge of interflow,baseflow and spring flood in the Kuye River basin.In addition,it is proposed that the SWAT runoff generation mechanism in arid and semi-arid regions need to be further investigated.

DOI

[4]
Conradt T, Wechsung F, Bronstert A, 2013. Three perceptions of the evapotranspiration landscape: Comparing spatial patterns from a distributed hydrological model, remotely sensed surface temperatures, and sub-basin water balances.Hydrology & Earth System Sciences, 17(7): 2947-2966.A problem encountered by many distributed hydrological modelling studies is high simulation errors at interior gauges when the model is only globally calibrated at the outlet. We simulated river runoff in the Elbe River basin in central Europe (148 268 km(2)) with the semi-distributed eco-hydrological model SWIM (Soil and Water Integrated Model). While global parameter optimisation led to Nash-Sutcliffe efficiencies of 0.9 at the main outlet gauge, comparisons with measured runoff series at interior points revealed large deviations. Therefore, we compared three different strategies for deriving sub-basin evapotranspiration: (1) modelled by SWIM without any spatial calibration, (2) derived from remotely sensed surface temperatures, and (3) calculated from long-term precipitation and discharge data. The results show certain consistencies between the modelled and the remote sensing based evapotranspiration rates, but there seems to be no correlation between remote sensing and water balance based estimations. Subsequent analyses for single sub-basins identify amongst others input weather data and systematic error amplification in inter-gauge discharge calculations as sources of uncertainty. The results encourage careful utilisation of different data sources for enhancements in distributed hydrological modelling.

DOI

[5]
Cui Ming, Zhang Xudong, Cai Qiangguoet al., 2008. Relationship between black soil development and climate change and geomorphological evolution in Northeast China.Geographical Research, 27(3): 527-535. (in Chinese)Soil on slopes of the gentle hilly black soil region in Northeast China, one of the most important bases of cash rice, degraded seriously after dozens of years of intensive cultivation. The thickness of soil humus layer becomes thinner and less fertile year after year. So it is very essential to deepen the researches of soil restoration and improvement after severe soil erosion. Analysis of main reasons for black soil degradation revealed that the cultivation activities halted the accumulation of organic matter and then baffled the soil development which cannot compensate the decrease of the thickness of soil humus layer caused by soil erosion. Soil developing process and conditions are the most important foundation for soil restoration. So the developing history of black soil and chernozem was reconstructed and the geomorphological and climatic factors, which were the key factors affecting the formation of black soil, were analyzed through studying both the formation time of the underlying strata and the local climate change history since the late Pleistocene. The conclusion is that black soil and chernozem formed in different periods, from early period of late Pleistocene and the beginning of Holocene respectively. The former period was warm and wet, while the latter period was warm and dry. And they formed in different places, the black soil was mainly distributed on the second and the third terraces and the chernozem on the first terrace of the Nenjiang River, which is lower than the distributing places of black soil and can accept more carbonate from the highland to form the characteristic illuvial layer. The processes of the soil formation were very slow, so it is hard to restore. These results provide important basis for forulating policies to improve the quality of soils in the region.

DOI

[6]
Faramarzi M, Abbaspour K C, Schulin Ret al., 2009. Modelling blue and green water resources availability in Iran.Hydrological Processes, 23(3): 486-501.Abstract Knowledge of the internal renewable water resources of a country is strategic information which is needed for long-term planning of a nation's water and food security, among many other needs. New modelling tools allow this quantification with high spatial and temporal resolution. In this study we used the program Soil and Water Assessment Tool (SWAT) in combination with the Sequential Uncertainty Fitting program (SUFI-2) to calibrate and validate a hydrologic model of Iran based on river discharges and wheat yield, taking into consideration dam operations and irrigation practices. Uncertainty analyses were also performed to assess the model performance. The results were quite satisfactory for most of the rivers across the country. We quantified all components of the water balance including blue water flow (water yield plus deep aquifer recharge), green water flow (actual and potential evapotranspiration) and green water storage (soil moisture) at sub-basin level with monthly time-steps. The spatially aggregated water resources and simulated yield compared well with the existing data. The study period was 1990 2002 for calibration and 1980 1989 for validation. The results show that irrigation practices have a significant impact on the water balances of the provinces with irrigated agriculture. Concerning the staple food crop in the country, 55% of irrigated wheat and 57% of rain-fed wheat are produced every year in water-scarce regions. The vulnerable situation of water resources availability has serious implications for the country's food security, and the looming impact of climate change could only worsen the situation. This study provides a strong basis for further studies concerning the water and food security and the water resources management strategies in the country and a unified approach for the analysis of blue and green water in other arid and semi-arid countries. Copyright 2008 John Wiley & Sons, Ltd.

DOI

[7]
Feng Xiaqing, Zhang Guangxin, Yin Xiongrui, 2010. Study on the hydrological response to climate change in Wuyuer river basin based on the SWAT model. Progress in Geography, 29(7): 827-832. (in Chinese)The Wuyur River Basin, located in the inland semi-arid region, is sensitive to climate change. The streamflow of Wuyur River is an important recharge source of Zhalong Wetland. The study on the streamflow change associated with the future climate change scenarios has a practical significance for the local socio-economic development and eco-environmental protection of Zhalong Wetland. With the distributed hydrological model of SWAT, the streamflow in the Wuyur River Basin was simulated, and the hydrological response to climate change was analyzed. The simulated results showed that the SWAT model could effectively simulate the streamflow change in the Wuyur River Basin. Especially, at the stations with large amounts of streamflow, the efficiency of simulation was satisfactory. The influence of climate change on streamflow was significant. The streamflow in the future climate change scenarios decreased gradually over time, and different hydrological stations had different streamflow change amplitude. Considering the decrease in streamflow, we need to apply a reasonable water resources allocation for the wetlands in the watershed to alleviate the adverse effects.

DOI

[8]
Fitzhugh T W, Mackay D S, 2001. Impacts of input parameter spatial aggregation on an agricultural nonpoint source pollution model.Journal of Hydrology, 236(1/2): 35-53.The accuracy of agricultural nonpoint source pollution models depends in part on how well model input parameters describe the relevant characteristics of the watershed. The spatial extent of input parameter aggregation has previously been shown to have a substantial impact on model output. This study investigates this problem using the Soil and Water Assessment Tool (SWAT), a distributed-parameter agricultural nonpoint source pollution model. The primary question addressed here is: how does the size or number of subwatersheds used to partition the watershed affect model output, and what are the processes responsible for model behavior? SWAT was run on the Pheasant Branch watershed in Dane County, WI, using eight watershed delineations, each with a different number of subwatersheds. Model runs were conducted for the period 1990 1996. Streamflow and outlet sediment predictions were not seriously affected by changes in subwatershed size. The lack of change in outlet sediment is due to the transport-limited nature of the Pheasant Branch watershed and the stable transport capacity of the lower part of the channel network. This research identifies the importance of channel parameters in determining the behavior of SWAT's outlet sediment predictions. Sediment generation estimates do change substantially, dropping by 44% between the coarsest and the finest watershed delineations. This change is primarily due to the sensitivity of the runoff term in the Modified Universal Soil Loss Equation to the area of hydrologic response units (HRUs). This sensitivity likely occurs because SWAT was implemented in this study with a very detailed set of HRUs. In order to provide some insight on the scaling behavior of the model two indexes were derived using the mathematics of the model. The indexes predicted SWAT scaling behavior from the data inputs without a need for running the model. Such indexes could be useful for model users by providing a direct way to evaluate alternative models directly within a geographic information systems framework.

DOI

[9]
Gao Chao, Jin Gaojie, 2012. Effects of DEM resolution on results of the SWIM hydrological model in the Changtaiguan basin.Geographical Research, 31(3): 399-408. (in Chinese)

[10]
Gao Chao, Liu Qing, Su Budaet al., 2013. The applicability assessment of hydrological models with different resolution and database in the Huaihe river basin, China.Journal of Natural Resources, 28(10): 1765-1777. (in Chinese)In this paper, the applicability of three hydrological models, including artificial neural network(ANN) model, Hydrologiska Byr ns Vattenbalansavdelning- D(HBV- D)model and Soil and Water Integrated Model(SWIM), are examined at different temporalspatial scales and databases in the Huaihe River basin which is above the Bengbu hydrological gauging station.It is found that ANN model only needs monthly data to build rainfall-runoff relationship and can obtain well simulation results, but HBV-D and SWIM models require data on daily scale such as daily precipitation, daily temperature and daily runoff. SWIM model even requires crop management data, nutrient data, soil erosion data etc. In addition, on spatial scale, the applicability of ANN model is adequate to large- scale basin, SWIM model may only be suitable for small- scale basin with an area of less than 10000 km2, and HBV- D model can apply to a basin of about 10000 km2.Furthermore, according to simulation results, ANN model can get better result for overall hydrological simulation, but it is not suitable for the hydrological and water resources research under climate change. Although their Nash-Sutcliffe coefficients are less than ANN model, the physically based distributed hydrological model, HBV- D model and the SWIM model are good tools to study impacts of climate change, which is significantly controlled by model structure.

[11]
Gao Chao, Zhai Jianqing, Tao Huiet al., 2009. Hydrological response to land use/land cover change in Chaohu Basin.Journal of Natural Resources, 24(10): 1794-1803. (in Chinese)土地利用/覆被变化的水文效应导致水资源量变化，显著影响流域生态和社会经济发展。利用DEM、土壤数据库、三期土地利用数据及流域周边六个国家基本气象站1964—2000年气象、流域出口断面裕溪闸水文站1964—1989年水文资料，依托德国PIK研究所HBV-D模型建立巢湖流域降水径流关系，分析流域土地覆被变化对径流量影响：①利用HBV-D模型模拟流域降水径流关系，率定后系统相对误差控制在3%左右，纳希效率系数达约83%，适合巢湖流域土地利用/覆被变化水文效应研究；②分析得出单位面积的农业用地、居民用地和水域影响径流深大小分别为-0.134469、0.074908和-0.0015244，即巢湖流域农业用地对径流影响程度要高于居民用地且为负，农业用地减少将增加径流量，居民用地增长利于径流量增加，水域对径流量影响相对较小。

DOI

[12]
Ge G, Xu C Y, Chen Det al., 2012. Spatial and temporal characteristics of actual evapotranspiration over Haihe River basin in China.Stochastic Environmental Research & Risk Assessment, 26(26): 1-15.AbstractSpatial and temporal characteristics of actual evapotranspiration over the Haihe River basin in China during 1960鈥2002 are estimated using the complementary relationship and the Thornthwaite water balance (WB) approaches. Firstly, the long-term water balance equation is used to validate and select the most suitable long-term average annual actual evapotranspiration equations for nine subbasins. Then, the most suitable method, the Pike equation, is used to calibrate parameters of the complementary relationship models and the WB model at each station. The results show that the advection aridity (AA) model more closely estimates actual evapotranspiration than does the Granger and Gray (GG) model especially considering the annual and summer evapotranspiration when compared with the WB model estimates. The results from the AA model and the WB model are then used to analyze spatial and temporal changing characteristics of the actual evapotranspiration over the basin. The analysis shows that the annual actual evapotranspirations during 1960 2002 exhibit similar decreasing trends in most parts of the Haihe River basin for the AA and WB models. Decreasing trends in annual precipitation and potential evapotranspiration, which directly affect water supply and the energy available for actual evapotranspiration respectively, jointly lead to the decrease in actual evapotranspiration in the basin. A weakening of the water cycle seems to have appeared, and as a consequence, the water supply capacity has been on the decrease, aggravating water shortage and restricting sustainable social and economic development in the region.

DOI

[13]
Guo Min, Fang Haiyan, Li Zhiying, 2016. SWAT model-based runoff simulation of Wuyuer river basin in the black soil region of northeast China.Research of Soil and Water Conservation, 23(4): 43-47. (in Chinese)

[14]
Hattermann F F, Wattenbach M, Krysanova Vet al., 2005. Runoff simulations on the macroscale with the ecohydrological model SWIM in the Elbe catchment-validation and uncertainty analysis.Hydrological Processes, 19(3): 693-714.This study presents an example where the hydrological processes of the ecohydrological model SWIM (Soil and Water Integrated Model) are thoroughly analysed. The model integrates hydrology, vegetation, erosion and nutrient dynamics. It is process-based and has to be calibrated. The hydrological validation of the model is of prime importance, because all other ecological processes are related to the water cycle. On the other hand, these ecological processes influence the water cycle in turn, and therefore they were considered in the modelling process and in the sensitivity and uncertainty analysis.The validation was multi-scale, multi-site and multi-criteria: the validation strategy followed a bottom-up approach in which the model was firstly calibrated for 12 mesoscale sub-basins, covering the main subregions of the German part of the Elbe basin, and the information gained from the mesoscale was then used to validate the model for the entire macroscale basin. Special attention was paid to the use of spatial information (maps of water table) to validate the model in addition to commonly used observations of water discharge at the basin outlet. One main result was that investigations in smaller catchments have to accompany macroscale model applications in order to understand the dominant hydrological processes in the different areas of the entire basin and at different scales.The validation was carried out in the German part of the Elbe river basin (80 258 km2). It is representative of semi-humid landscapes in Central Europe, where water availability during the summer season is a limiting factor for plant growth and crop yield.

DOI

[15]
Hesse C, Krysanova V, Stefanova Aet al., 2015. Assessment of climate change impacts on water quantity and quality of the multi-river Vistula Lagoon catchment.Hydrological Sciences Journal, 60(5): 1-22.

DOI

[16]
Huang S, Hesse C, Krysanova Vet al., 2009 From meso- to macro-scale dynamic water quality modelling for the assessment of land use change scenarios.Ecological Modelling, 220(19): 2543-2558.The implementation of the European Water Framework Directive requires reliable tools to predict the water quality situations in streams caused by planned land use changes at the scale of large regional river basins. This paper presents the results of modelling the in-stream nitrogen load and concentration within the macro-scale basin of the Saale river (24,167 km 2) using a semi-distributed process-based ecohydrological dynamic model SWIM (Soil and Water Integrated Model). The simulated load and concentration at the last gauge of the basin show that SWIM is capable to provide a satisfactory result for a large basin. The uncertainty analysis indicates the importance of realistic input data for agricultural management, and that the calibration of parameters can compensate the uncertainty in the input data to a certain extent. A hypothesis about the distributed nutrient retention parameters for macro-scale basins was tested aimed in improvement of the simulation results at the intermediate gauges and the outlet. To verify the hypothesis, the retention parameters were firstly proved to have a reasonable representation of the denitrification conditions in six meso-scale catchments. The area of the Saale region was classified depending on denitrification conditions in soil and groundwater into three classes (poor, neutral and good), and the distributed parameters were applied. However, the hypothesis about the usefulness of distributed retention parameters for macro-scale basins was not confirmed. Since the agricultural management is different in the sub-regions of the Saale basin, land use change scenarios were evaluated for two meso-scale subbasins of the Saale. The scenario results show that the optimal agricultural land use and management are essential for the reduction in nutrient load and improvement of water quality to meet the objectives of the European Water Framework Directive and in view of the regional development plans for future.

DOI

[17]
Kong Fanzhe, Song Xiaomeng, Zhan Cheshenget al., 2011. An efficient quantitative sensitivity analysis approach for hydrological model parameters using RSMSobol method.Acta Geographica Sinica, 66(9): 1270-1280. (in Chinese)Sensitivity analysis of hydrological models is a key step for model uncertainty quantification.It can identify the dominant parameters,reduce the model calibration uncertainty,and enhance the model optimization efficiency.However,how to effectively validate a model and identify the dominant parameters for a large-scale complex distributed hydrological model is a bottle-neck to achieve the parameters optimization.There are some shortcomings for classical approaches,e.g.time-consuming and high computation cost,to quantitatively assess the sensitivity of the multi-parameters complex hydrological model.For this reason,a new approach was applied in this paper,in which the support vector machine was used to construct the response surface(a surrogate model) at first.Then it integrated the SVM-based response surface with the Sobol method,i.e.the RSMSobol method,to achieve the quantification assessment of sensitivity for complex models.Taking the distributed time-variant gain model in the Huaihe River Basin as a case study,we selected three objective functions(i.e.water balance coefficient WB,Nash-Sutcliffe efficiency coefficient NS,and correlation coefficient RC) to assess the model as the output responses for sensitivity analysis.The results show that the RSMSobol method can not only achieve the quantification of the sensitivity,and also reduce the computational cost,with good accuracy compared to the classical approaches.

DOI

[18]
Krysanova V, Hattermann F, Huang Set al., 2015. Modelling climate and land use change impacts with SWIM: Lessons learnt from multiple applications. Hydrological Sciences Journal, 60(4): 606-635(30).

[19]
Krysanova V, Hattermann F, Wechsung F, 2005. Development of the ecohydrological model SWIM for regional impact studies and vulnerability assessment.Hydrological Processes, 19(3): 763-783.Abstract In this paper the ecohydrological model SWIM developed for regional impact assessment is presented, and examples of approaches to climate and land use change impact studies are described. SWIM is a continuous-time semi-distributed ecohydrological model, integrating hydrological processes, vegetation, nutrients (nitrogen and phosphorus) and sediment transport at the river basin scale. Its spatial disaggregation scheme has three levels: (1) basin, (2) sub-basins and (3) hydrotopes within sub-basins. The model was extensively tested and validated for hydrological processes, nitrogen dynamics, crop yield and erosion (mainly in mesoscale sub-basins of the German part of the Elbe River basin). After appropriate validation in representative sub-basins, the model can be applied at the regional scale for impact studies. Particular interest in the global change impact studies is given to effects of expected changes in climate and land use on hydrological processes and agro-ecosystems, including water balance components, water quality and crop yield. This paper (a) introduces the reader to the class of process-based ecohydrological catchment scale models, (b) introduces SWIM as one such model, and (c) presents two examples of impact studies performed with SWIM for the federal state of Brandenburg (Germany), which overlaps with the lowland part of the Elbe drainage area. The impact studies provide a better understanding of the complex interactions between climate, hydrological processes and vegetation, and improve our potential adaptation to the expected changes. Copyright 2005 John Wiley & Sons, Ltd.

DOI

[20]
Krysanova V, Wechsung F, Arnold J et al., 2011. Soil and Water Integrated Model User Manual. Beijing: China Meteorological Press. (in Chinese)

[21]
Li Jinfeng, Huang Chongfu, Zong Tian, 2005. Anti-accuracy phenomenon and formalization. Systems Engineering:Theory & Practice, 25(4): 128-132. (in Chinese)

[22]
Liu Zhuo, Liu Changming, 2006. The analysis about water resource utilization, ecological and environmental problems in Northeast China.Journal of Natural Resources, 21(5): 700-708. (in Chinese)Some ecological and environmental problems are caused due to water resources extensive utilization,which has affected water resources sustainable utilization and development.Northeast China,which includes three provinces Liaoning,Jilin,Heilongjiang and the east of Inner Mongolia is famous for its old industrial base.it is featured by better growing conditions and rich natural resources feature.The regional development is based on water resources exploitation.In the past 50 years,the exploitation and utilization of water resources have extended to a certain degree with industrial and agricultural development,causing a serious of water-related serious ecological and environmental problems:①Severe water pollution:the length of reaches satisfyingⅠ～Ⅲ water quality criteria was less than 40% of the total qualified reaches to be evaluated and the length of reaches below water quality criteria was over 60% in 2003;②land desertification: Land desertification has been aggravating in the western part of Wortheast China,river cut-off has happened in a number of watercourses;and ③groundwater overdraft and large-scale wetland shrinkage occurred in many places.To solve the above problems,several suggestions and countermeasures are put forward:firstly,ecological restoration and environmental protection must follow the law of nature;secondly,ecological and environmental water requirement should be satisfied;Finally,monitoring system should be strengthened to supply real-time data.Generally,ecological and environmental problems are correlated with water resources utilization of Northeast China.Water resources must be used with in the limit of the available amout and exploited according to the local conditions,and equal stress should be laid on economical and high-effecient utilization so as to realize sustainable development of Northeast China.

DOI

[23]
Maidment D R, 1996. GIS and hydrologic modeling: An assessment of progress. In: Proceedings of the Third International Conference on GIS and Environmental Modeling, Santa Fe, New Mexico, January 22-26, 1996.The First International Conference on GIS and Environmental Modeling was held in Boulder, Colorado, in September 1991. At that Conference I presented a survey of the state of GIS and hydrologic modeling as the subject then existed (Maidment, 1993). The intent of this

[24]
Mo Fei, 2008. The hydrological effects of forest/vegetation and simulation in the small watershed of Honggou, Liupan mountains [D]. Beijing: Chinese Academy of Forestry. (in Chinese)

[25]
Romanowicz A A, Vanclooster M, Rounsevell Met al., 2005. Sensitivity of the SWAT model to the soil and land use data parametrisation: A case study in the Thyle catchment in Belgium.Ecological Modelling, 187(1): 27-39.The sensitivity of the distributed hydrological SWAT model to the pre-processing of soil and land use data was tested for modelling rainfall-runoff processes in the Thyle catchment in Belgium. To analyse this sensitivity, 32 different soil and land use parameterisation scheme were generated and evaluated. The soil input data sources were a generalised soil association map at a scale of 1:500,000, a detailed soil map at a scale of 1:25,000 and the soil profile analytical database AARDEWERK. These soil data were combined with a detailed and a generalised land use map. The results suggest that the SWAT model is extremely sensitive to the quality of the soil and land use data and the adopted pre-processing procedures of the geographically distributed data. The resolution and fragmentation of the original map objects are significantly affected by the internal aggregation procedures of the SWAT model. The catchment size threshold value (CSTV) is thereby a key parameter controlling the internal aggregation procedure in the model. It is shown that a parabolic function characterises the relationship between the CSTV and the hydrological modelling performance of the uncalibrated model, suggesting that optimal uncalibrated modelling results are not obtained when the CSTV is minimised. The hydrological response of the SWAT model to the calculated soil properties is significant. Therefore preference should be given to the calculation of the derived hydrologic soil properties prior to averaging of the profile data. Finally some general guidelines are suggested for parameterising soil and land use in the SWAT model application.

DOI

[26]
Rougier J, Richmond A D, 2005. Sensitivity of the SWAT model to the soil and land use data parametrisation: A case study in the Thyle catchment, Belgium.Ecological Modelling, 187(1): 27-39.The sensitivity of the distributed hydrological SWAT model to the pre-processing of soil and land use data was tested for modelling rainfall-runoff processes in the Thyle catchment in Belgium. To analyse this sensitivity, 32 different soil and land use parameterisation scheme were generated and evaluated. The soil input data sources were a generalised soil association map at a scale of 1:500,000, a detailed soil map at a scale of 1:25,000 and the soil profile analytical database AARDEWERK. These soil data were combined with a detailed and a generalised land use map. The results suggest that the SWAT model is extremely sensitive to the quality of the soil and land use data and the adopted pre-processing procedures of the geographically distributed data. The resolution and fragmentation of the original map objects are significantly affected by the internal aggregation procedures of the SWAT model. The catchment size threshold value (CSTV) is thereby a key parameter controlling the internal aggregation procedure in the model. It is shown that a parabolic function characterises the relationship between the CSTV and the hydrological modelling performance of the uncalibrated model, suggesting that optimal uncalibrated modelling results are not obtained when the CSTV is minimised. The hydrological response of the SWAT model to the calculated soil properties is significant. Therefore preference should be given to the calculation of the derived hydrologic soil properties prior to averaging of the profile data. Finally some general guidelines are suggested for parameterising soil and land use in the SWAT model application.

DOI

[27]
Shu Chang, Liu Suxia, Mo Xingguoet al., 2008. Uncertainty analysis of Xinanjiang model parameter.Geographical Research, 27(2): 343-352. (in Chinese)The uncertainty problem in hydrological model is an important issue of scientific research at present,which covers three aspects of data,model structure and parameters.Parameter is one of the key roles in analyzing model uncertainty problem.The value of parameters depends on characteristics of a basin,but in fact it is difficult to obtain because there are few observation stations.In general,it needs to confirm parameters by several calibration methods including Genetic Algorithm,Simulated Annerling and Artificial Neural Network.So there exists parameter uncertainty problem.The generalized likelihood uncertainty estimation(GLUE) methodology is an effective approach to study uncertainty of parameters.In this paper,the uncertainty in Xinanjiang model is examined by employing GLUE.Based on the simulation results of daily data from Jiuzhou(1978~1987) and Lushi(1980~1988) basins,it is found that the phenomenon of "equifinality" exists among parameters groups for both of the basins.According to comparison result of scatter plots,parameters of Xinanjiang model can be classified into three groups: sensitivity parameters such as UM,EX;non-sensitivity parameters such as KC,CS and regional sensitivity parameters such as B,WM.The conclusion is favorable for understanding parameters of Xinanjiang model so as to provide valuable scientific information for simulating hydrological processes.Finally it puts forward the main contents on future uncertainties research in hydrological modeling.

DOI

[28]
Song Xiaomeng, Zhang Jianyun, Zhan Cheshenget al., 2015. Review parameter sensitivity analysis hydrologic modeling.Advances in Science and Technology of Water Resources, 23(6): 105-112. (in Chinese)

[29]
Song Yanhua, Ma Jinhui, 2007. Applicability of SWAT model in Longxi of the Loess Plateau.Arid Land Geography, 30(6): 933-938. (in Chinese)In recent years,SWAT(Soil and Water Assessment Tool)model is widely used in the world.However,most of the applications are focused on humid regions.This paper explored the possibility of applying the SWAT model to an arid and semi-arid region-Longxi of the Loess Plateau.Because of the special precipitation mode,the runoff simulation results of SWAT model are not receivable for all years of this region.However,the results are perfect when simulating the runoff of the periods with more slightly changing in precipitation.So in these periods,SWAT model can be used for various simulation analyses which are related to the runoff.

DOI

[30]
Stefanova A, Krysanova V, Hesse Cet al., 2015. Climate change impact assessment on water inflow to a coastal lagoon: The Ria de Aveiro watershed, Portugal.Hydrological Sciences Journal, 60(5): 1-20.

DOI

[31]
Tao Y, Wang X, Yu Zet al., 2014. Climate change and probabilistic scenario of streamflow extremes in an alpine region.Journal of Geophysical Research Atmospheres, 119(14): 8535-8551.Abstract Future projections of streamflow extremes are of paramount significance in assessing the climate impacts on social and natural systems, particularly for the Himalayan alpine region in the Tibetan Plateau known as the Asian water tower. This study strives to quantify the uncertainties from different sources in simulating future extreme flows and seeks to construct reliable scenarios of future extreme flows for the headwater catchment of the Yellow River Basin in the 21st century. The results can be formulated as follows: (1) The revised snow model based on a daily active temperature method is superior to the commonly used degree-day method in simulating snowmelt processes. (2) In general, hydrological models contribute more uncertainties than the downscaling methods in high flow and low flow over the cryospheric alpine regions characterized by the snow-rainfall-induced runoff processes under most scenarios. Meanwhile, impacts to uncertainty vary with time. (3) The ultimate probability of high flow exhibits a downward trend in future by using an unconditional method, whereas positive changes in the probability of low flow are projected. The method in the work includes a variety of influence from different contributing factors (e.g., downscaling models, hydrological models, model parameters, and their simulation skills) on streamflow projection, therefore can offer more information (i.e., different percentiles of flow and uncertainty ranges) for future water resource planning compared with the purely deterministic approaches. Hence, the results are beneficial to boost our current methodologies of climate impact research in the Himalayan alpine zone.

DOI

[32]
Tzyy-Woei Chu, 2003. Modeling Hydrologic and water quality response of a mixed land use watershed in piedmont physiographic [D]. Graduatie School of the University of Maryland.

[33]
Wang Genxu, Li Yuanshou, Wu Qingbaiet al., 2006. The relationship between frozen soil and vegetation in permafrost regions of the Qinghai-Tibet Plateau and its impact on the alpine ecosystem. Science in China Series D: Earth Sciences, 36(8): 743-754. (in Chinese)

[34]
Wortmann M, Krysanova V, Kundzewicz Z Wet al., 2014. Assessing the influence of the Merzbacher Lake outburst floods on discharge using the hydrological model SWIM in the Aksu headwaters, Kyrgyzstan/NW China.Hydrological Processes, 28(26): 6337-6350.Abstract Top of page Abstract INTRODUCTION STUDY SITE AND DATA METHODS RESULTS DISCUSSION CONCLUSIONS ACKNOWLEDGEMENTS REFERENCES Glacial lake outburst floods (GLOF) often have a significant impact on downstream users. Including their effects in hydrological models, identifying past occurrences and assessing their potential impacts are challenges for hydrologists working in mountainous catchments. The regularly outbursting Merzbacher Lake is located in the headwaters of the Aksu River, the most important source of water discharge to the Tarim River, northwest China. Modelling its water resources and the evaluation of potential climate change impacts on river discharge are indispensable for projecting future water availability for the intensively cultivated river oases downstream of the Merzbacher Lake and along the Tarim River. The semi-distributed hydrological model SWIM was calibrated to the outlet station Xiehela on the Kumarik River, by discharge the largest tributary to the Aksu River. The glacial lake outburst floods add to the difficulties of modelling this high-mountain, heavily glaciated catchment with poor data coverage and quality. The aims of the study are to investigate the glacier lake outburst floods using a modelling tool. Results include a two-step model calibration of the Kumarik catchment, an approach for the identification of the outburst floods using the measured gauge data and the modelling results and estimations of the outburst flood volumes. Results show that a catchment model can inform GLOF investigations by providing ormal (i.e. without the outburst floods) catchment discharge. The comparison of the simulated and observed discharge proves the occurrence of GLOFs and highlights the influences of the GLOFs on the downstream water balance. 2013 The Authors. Hydrological Processes Published by John Wiley & Sons Ltd.

DOI

[35]
Xu Yanhui, 2011. A study on influences of land use change to Huangqian Watershed based on SWIM model [D]. Tai’an: Shandong Agricultural University. (in Chinese)

[36]
Yang Dawen, Li Chong, Ni Guanghenget al., 2004. Application of a distributed hydrological model to the Yellow river basin.Acta Geographica Sinica, 59(1): 143-154. (in Chinese)For implementing water resources management in the Yellow River Basin, water resources assessment is very necessary and important. The accuracy of water resources assessment relies on predictability of the hydrological cycle. Different land uses, topographical features, geological and soil conditions, and artificial water uses (mainly agricultural irrigation) determine the complexity of hydrological characteristics in this basin. With the limited observation of the river discharge, it is difficult to develop a lumped model for simulating hydrology in different sub-basins based on parameter calibration. The physically-based hydrological model can be helpful in this case. The present research attempts to incorporate all available spatial information into the hydrological modeling by a distributed approach. A physical model is developed using the physical governing equations for description of the hydrological processes. It carries out a 10-year (1980-1989) simulation of the natural hydrological cycle. Based on the hydrological simulation, the paper discusses the spatial-temporal hydrological characteristics and the status of water resources in the Yellow River Basin.

DOI

[37]
Zhang Shulan, 2011. The assessment of impact of land use change and climate variability on hydrologic process in basin [D]. Beijing: Chinese Academy of Forestry. (in Chinese)

[38]
Zhang Shulan, Yu Pengtao, Wang Yanhuiet al., 2011. Estimation of actual evapotranspiration and its component in the upstream of Jinghe basin.Acta Geographica Sinica, 66(3): 385-395. (in Chinese)The ecohydrological model SWIM was used to simulate the hydrological process based on the data of vegetation,soil,climate,hydrology in the upstream of Jinghe River located at the Jingchuan station,and actual evapotranspiration and its components were estimated.The result showed that SWIM model was applicable to this area,and the average annual actual evapotranspiration (AET) was 443 mm from 1997 to 2003 in the watershed,with soil evaporation being 259 mm,vegetation transpiration 157 mm,and canopy interception 27 mm.In the watershed,the AET of the forest and non-forest area in rocky mountain was 484 mm in maximum and 418 mm inminimum respectively.And the AET of loess area was 447 mm lower than that of forest area and higher than that of non-forest area in rocky mountain.In addition,soil evaporation in forest area was significantly low,while the transpiration and canopy interception in the area were obviously high in the watershed.The inter-annual AET mainly occurred in the period from May to August,which was 60% of the total annual AET,with the great proportion 63% of canopy evapotanspiration.The AET of wet year increased by 78 mm compared with that of dry year,among which,soil evaporation increased greatest,followed by transpiration and canopy interception.

DOI

[39]
Zhang Shulan, Yu Pengtao, Zhang Haijunet al., 2015. A simulation study on the hydrological impacts of varying forest cover in the stony mountain area and loess area of the upper reaches of Jinghe basin.Acta Ecologica Sinica, 35(4): 1068-1078. (in Chinese)The upper reach of Jinghe Basin is one of the main water-head areas and key afforestation areas in the Loess Plateau of China. A relatively large scaled evaluation of the forest hydrological impacts in this region is important for guidingthe rational ecological afforestation,ensuring the regional water supply safety and sustainable development. In order to possibly remove the disturbances from topography,soil and climate,the upper reach of Jinghe Basin was divided into the stony mountain area and the loess area,and several scenarios were set up in each area. The distributed and eco-hydrological watershed model of SWIM was calibrated and validated by using the meteorological and hydrological data measured in1997 1999 and 2000 2003 respectively,and then was used to simulate the impacts of varying forest cover and its spatial distribution on the annual evapotranspiration,water yield,groundwater recharge,deep soil percolation,and the peak daily runoff. The simulated results in the stony mountain area showed that increasing forest cover will increase the basin evapotranspiration and reduce the water yield. For example,when compared the current forest / vegetation scenario( 13. 8%of the whole basin) with the scenario of changing all forests into grassland( 0% of the whole basin),the basin annual evapotanspiration is changed from 445. 4 mm to 427. 7 mm( a decrease of 17. 4 mm and 4%),the annual water yield is changed from 42. 4 mm to 53. 5 mm( an increase of 11. 1 mm and 26. 3%),the annual groundwater recharge is changed from 61. 6 mm to 76. 9 mm( an increase of 15. 3 mm and 24. 8%),the deep soil percolation is changed from 72. 9 mm to88. 3 mm( an increase of 17. 7 mm and 24. 3%). In average,an increase of forest cover of 10% will lead to an increase of basin annual evapotranspiration of 12. 8 mm,a reduction of annual water yield of 8. 0 mm,and a reduction of annual groundwater recharge of 11. 1 mm in the stony mountain area. In the relatively dryer loess region with deep soil,the forest cover increase will also increase the basin evapotranspiration and decrease the water yield,but in an obviously smaller variation range compared with the stony mountain area where the annual precipitation is higher and the soil layer is thinner.In average,the basin annual evapotranspiration is increased by 9. 0 mm,the annul water yield is decreased by 4. 5 mm,and the annual groundwater recharge is reduced by 8. 8 mm,when the basin forest cover is increased by 10%. In addition,the hydrological impacts caused by afforestation on more gentle slopes are stronger than those caused by afforestation on steeper slopes. When comparing the monthly distribution pattern of forest hydrological impacts,there is also a clear difference between the stony mountain area and the loess area. For example,the significant increase of evapotranspiration is found in the period from May to July,and the decrease of deep soil percolation is found in the period from May to October in the stony mountain area; while in the loess area it is in the periods of May-October and July-October respectively. In addition,the impact of varying forest cover on the peak daily runoff is not significant in the stony mountain area,probably because of the high infiltration ability of the soil with high stone fragment content; but significant( decrease) in the loess area,probably mainly because of the obviously increased rainwater infiltration into soil after afforestation and then the reduced surface runoff generation.

DOI

[40]
Zhang Shulan, Zhang Haijun, Wang Yanhuiet al., 2015. Influence of vegetation type on hydrological process at landscape scale in the upper reaches of Jinghe basin.Scientia Geographica Sinica, 35(2): 230-236. (in Chinese)Profoundly understanding the influence of vegetation types and their distribution pattern on the hydrological processes is of great significance for water resources management and vegetation reasonable recovery in a large basin scale. In this study, the upper reaches of Jinghe Basin with stony mountain area was selected as the research area, and the dynamic process-based eco-hydrological model(SWIM) was used to simulate hydrological effect of different vegetation types at landscape scale, the impact of the vegetation distribution on hydrological pattern was further analyzed by distinguishing elevation in view of stony mountain area and loess area. Results showed that hydrological effects including evapotranspiration and its components, water yield and deep soil percolation among forest, farmland and grassland had significant difference, and the same vegetation type in different regions(stony mountain area and loess area) had obviously different effects on hydrological process. Because of the area and elevation difference of vegetation landscape pattern, the hydrological process in different areas and elevation sections had difference. For example, in stony mountain area, the precipitation and evapotranspiration in elevation section of 2 250-2 922 m dominated by forest was largest to 641 mm and 484 mm respectively, while the precipitation in elevation section of 1 750-2 250 m with farmland, grassland and forest scattered around was larger to 590 mm, but the evapotranspiration was lowest to 434 mm; in arid loess area with precipitation of 514 mm, the evapotranspiration in two elevation sections of 1 026-1 350 m and 1 350-1 750 m with farmland and grassland were 458 mm and 440 mm respectively. In addition, the difference of the ratios between hydrological process elements and precipitation in both areas was more significant,but is not obvious for different elevation sections in the same area.

[41]
Zhang Yongfang, Deng Junli, Guan Dexinet al., 2011. Spatiotemporal changes of potential evapotranspiration in Songnen Plain of northeast China.Chinese Journal of Applied Ecology, 22(7): 1702-1710. (in Chinese)Based on the daily meteorological data from 72 weather stations from 1961-2003,a quantitative analysis was conducted on the spatiotemporal changes of the potential evapotranspiration in the Plain.The Penman-Monteith model was applied to calculate the potential evapotranspiration;the Mann-Kendall test,accumulative departure curve,and climatic change rate were adopted to analyze the change trend of the evapotranspiration;and the spatial analysis function of ArcGIS was used to detect the spatial distribution of the evapotranspiration.In 1961-2003,the mean annual potential evapotranspiration in the Plain was 330-860 mm,and presented an overall decreasing trend,with the high value appeared in southwest region,low value in surrounding areas of southwest region,and a ring-belt increasing southwestward.The climatic change rate of the annual potential evapotranspiration was-0.21 mm a-1.The annual potential evapotranspiration was the highest in 1982,the lowest in 1995,and increased thereafter.Seasonally,the climatic change rate of the potential evapotranspiration in spring,summer,autumn,and winter was-0.19,0.01,-0.05,and 0.03 mm a-1,respectively,suggesting that the potential evapotranspiration had a weak increase in winter and summer and a slight decrease in spring and autumn.

DOI PMID

[42]
Zhang Zhengtao, 2015, The influence of different spatial resolution and algorithm selection on the SWIM hydrological model [D]. Wuhu: Anhui Normal University. (in Chinese)

Options
Outlines

/

〈 〉