Impacts of Land-Use Conversions on the Water Cycle in a Typical Watershed in the Southern Chinese Loess Plateau

Journal of Hydrology 593 (2021) 125741

Contents lists available at ScienceDirect

Journal of Hydrology

journal homepage: www.elsevier.com/locate/jhydrol

Research papers Impacts of land-use conversions on the water cycle in a typical watershed in the southern Chinese Loess Plateau

Jingyi Hu a, Yiping Wu a,*, Lijing Wang a, Pengcheng Sun a, Fubo Zhao a, Zhangdong Jin a,b,c, Yunqiang Wang a,b, Linjing Qiu a, Yanqing Lian b a Department of Earth & Environmental Science, Xi’an Jiaotong University, Xi’an, Shaanxi Province 710049, China b SKLLQG, Institute of Earth Environment, Chinese Academy of Sciences, Xi’an 710075, China c CAS Center for Excellence in Quaternary Science and Global Change, Xi’an 710061, China

ARTICLE INFO ABSTRACT

This manuscript was handled by Marco Borga, Land use change is one of the dominant driving factors of hydrological change at the watershed scale. Thus, Editor-in-Chief understanding the hydrological responses to land use changes can facilitate development of sustainable water resource management. The land use of the Wei River (the largest tributary of the Yellow River) Basin (WRB) has Keywords: changed greatly due to the large-scale ecological restoration program in the Chinese Loess Plateau (e.g., grain- ET for-green program), causing dramatic impacts on the water cycle. This study was to simulate the impacts of land Land use change use and land cover change (LUCC) on the key hydrological components, using the Soil and Water Assessment Soil water content SWAT Tool (SWAT). We investigated the spatiotemporal changes of LUCC from 1980 to 2010 in terms of four sub- The Wei River Basin regions (i.e., three subbasins and one mid-downstream area) and three landforms (i.e., mountain, hill, and plain), respectively. Then, we quantified the spatial heterogeneity of hydrological responses to land use change scenarios. Our LUCC analysis showed that cropland declined by about 0.8%, from 58,776 km2 in 1980 to 57,519 km2 in 2010, whereas the forest and grassland areas correspondingly increased 552 km2 and 16 km2, respec tively. The urban area changed more dramatically in the middle and lower reaches of the Wei River owing to the faster socioeconomic development in this region than the rest areas. In mountain area, the main types of land use were forest and grassland, and the main transformation was from cropland to forest land and grassland. The area of cropland in plain could be over 50%, and it was mainly converted to urban land. The hydrological simulation indicated that LUCC from 1980 to 2010 caused 5.3% of decrease in the water yield and 6.2% increase in soil water content, but there was nearly no change in evapotranspiration (ET). Scenarios about slopping land use conversion (SLC) program showed that the conversion of cropland to grassland or forest (i.e., grain-for-green) resulted in negative effects on soil water content and water yield, with a greater effect by the reforestation. In ◦ addition, we found that the change of ET was clear in areas where cropland with slope > 15 was converted to grassland or forest, suggesting slope is also an important factor for hydrological responses to LUCC. This study can provide valuable decision support for land use planning and water resources protection in the WRB.

1. Introduction underlying surface conditions, such as soil types, terrain slope, and plants (Deng et al., 2019). Land use activities, including the conversion Water scarcity is one of the most pervasive problems afflictingpeople of natural landscapes for human use and the changes of human domi throughout the world (Jiang, 2009; Mekonnen and Hoekstra, 2016). nated land management, have greatly transformed the global underlying Problems associated with water are expected to grow worse in the surface (Fan and Shibata, 2015). These activities can influencedifferent coming decades, especially the water shortage in the arid and semi-arid hydrological processes such as infiltration, groundwater recharge, ET, regions, with wide-ranging consequences for human societies and eco soil water content, and water yield (Foley et al., 2005; Liu et al., 2010). systems (Conrad et al., 2019; Shannon et al., 2010; Wang et al., 2019). Many studies have proved that severe human activities could change the Water availability in river basins are mainly determined by the land use patterns, thereby affecting water cycle and water resources

* Corresponding author. E-mail address: [email protected] (Y. Wu). https://doi.org/10.1016/j.jhydrol.2020.125741

(Gentine et al., 2012; Liu et al., 2013; Yao et al., 2014). Thus, land use different LUCC and hydrological responses spatially. Table A1 showed changes and its impacts on hydrology are becoming one of the hottest that the middle and lower reaches of the Wei River (MLWR) experienced topics in the global environment change field (Moran-Tejeda´ et al., faster urban development than other sub-regions. Although Li et al. 2010; Yang et al., 2015; Yao et al., 2012). (2016a) analyzed the spatial runoff responses to land use change at both Nowadays, the study about the influenceof land use and land cover yearly and seasonal scales and indicated that the winter runoff change (LUCC) on watershed hydrology is attracting an increasing decreased in the Jing River Basin but increased in other basins in the attention, and three major methods (paired watershed method, hydro scenario of returning farmland to forest, the spatial characteristics of logical characteristics analysis, and hydrological modeling) were hydrological effects caused by LUCC within the WRB is poorly under adopted to study hydrological impacts of LUCC at different scales such stood. Therefore, it is of importance to assess hydrological responses to as nation, region, urban, and river basin (Gwenzi and Nyamadzawo, heterogeneous LUCC for more accurate watershed management (Peng 2014; Teutschbein and Seibert, 2012). For hydrological modeling, the et al., 2013; Qiu et al., 2011). Soil and Water Assessment Tool (SWAT) had been proved to be one of This study employed an acceptable calibrated and validated the most suitable and efficient models to simulate the hydrological re physically-based SWAT hydrological model over the WRB, aiming to sponses to LUCC (Devia et al., 2015; Wang et al., 2014). Using an identify responses of hydrological processes to land use scenarios with improved SWAT model, Zhang et al. (2020) found that afforestation in three periods of LUCC over the past thirty years. We used four hypo North Johnstone River catchment could decrease surface runoff and thetical land conversion scenarios based on the soil and water conser increase ET. In addition, hydrological model is an effective tool for vation practices, respectively, in terms of four sub-regions (the upper quantitative assessment of hydrological responses to environmental reach of the Wei River (UWR), the Jing River subbasin (JR), the North changes, but many previous studies took the watershed as a whole, Luo River subbasin (NLR), and the MLWR) and three landforms regardless of the spatial heterogeneity of LUCC and their impacts as well (mountain, hill, and plain). The specificobjectives of this study were to: (Wang et al., 2013). In fact, the natural environment conditions and 1) analyze the spatial characteristics of LUCC patterns and compare the socioeconomic development strategies, such as landforms, slope, and changes in terms of sub-regions and different landforms, respectively, 2) urbanization level, could result in the heterogeneity of LUCC and the investigate the spatiotemporal changes of the key hydrological compo hydrological consequences (Li and Zhou, 2014, 2015). Huang et al. nents as a result of LUCC, and 3) predict the potential effects of the (2013) showed that water storage initially rose but then fell sharply with sloping land conversion (SLC) program on the water cycle. The results the slope increased from 8.8% to 36.4% using the simulated rainfall can support policy makers and planners in proposing suitable manage experiments. Definitely, a better understanding of hydrological re ment practices to reasonably allocate water resources and put forward sponses to the LUCC and their spatial heterogeneity would be greatly effective measures in achieving coordinated development of society, helpful for more accurate water resources management and optimiza economy, and ecological environment (Li et al., 2015; Sun et al., 2020). tion (Ghaffari et al., 2010). The northwest China is one of the most water scarcity region in the 2. Materials and methods world. According to statistics, the annual average amount of water re sources in northwest China only accounts for 5.84% of the national 2.1. Study area amount (Chen et al., 2014; Cheng, 2002). With the population growth and fast-growing cities and industries, northwest China has experienced The Wei River is the largest tributary of the Yellow River. The WRB is intensive land-use changes (Cao et al., 2009; Chen et al., 2018). The located in the southeastern part of the CLP with a drainage area of LUCC impacts on the water cycle, together with the increasing con 134,800 km2 (Fig. 1). According to the classification system of basic sumption of water resources due to social development caused the se geomorphic types (Li et al., 2008), the WRB was divided into mountains, vere water shortage in this area (Cao et al., 2009; Chen et al., 2018; hills and plains, accounting for 28.1%, 30.8%, and 41.1% in area, Wang et al., 2012a, 2012b; Yin et al., 2017; Zhai and Tao, 2017). At the respectively (Fig. A1). Also, we divided the WRB into four sub-regions end of the 20th century, the government has implemented various soil (UWR, JR, NLR, and MLWR) (Fig. A1). Among them, the JR and the and water conservation practices to mitigate the insufficient water re BR are the two large tributaries of the Wei River, accounting for 34% and sources and sever soil erosion in the Chinese Loess Plateau (CLP) (Feng 20% of the total area of the WRB. Based on the past 50-year et al., 2016; Li et al., 2016b; Zhou et al., 2012). As a result, this region (1965–2014) hydro-climatic data, the annual mean temperature (T) is ◦ has experienced significantlychanges of land use and hydrological cycle 9.7 C, and the annual mean precipitation and annual runoff depth are because of human activities (Wu et al., 2014; Zhang, 2005). It was re 559 mm and 57 mm, respectively. Besides, about 78% of precipitation ported that the vegetation coverage rate of the CLP increased from 6.5% occurs in summer and autumn. In the WRB, there are three main land in the 1970s to 51% in 2010 because of large-scale ecological restora use types – cropland, grassland, and forest, accounting for 43.6%, 37%, tion, resulting in a significant decrease in streamflow in the middle and 16.2%, respectively. Yellow River (Feng et al., 2016; Leng et al., 2016; Wang and Qin, 2017). The Wei River Basin (WRB), a typical catchment located in a semi- 2.2. Data humid and semi-arid transition zone in northwest China, has experi enced dramatic land use and hydrological changes in the past few de The 90 m Digital Elevation Model (DEM) was obtained from USGS/ cades (Zhao et al., 2015). Dong et al. (2015) found that agricultural land NASA Shuttle Radar Topography Mission (SRTM) Digital Elevation and water body had the greatest effect on direct runoff generation. Li Database (http://srtm.csi.cgiar.org/). Land use maps (1980, 1995, and et al. (2018a) presented an alternative approach of time-varying pa 2010) with 1 km resolution were collected from Data Center for Re rameters in SWAT and found that LUCC had significant impacts on the sources and Environmental Sciences Data Center (RESDC), Chinese watershed streamflow, especially on the streamflow during the dry Academy of Sciences (CAS) (http://www.resdc.cn/). The topographic season. The changed hydrological processes by the intensive human data were collected from the Institute of Geographic Sciences and Nat activities in the WRB have resulted in ecological problems (Xu et al., ural Resources Research, Chinese Academy of Sciences (http://www. 2017; Zhang et al., 2019). Although there are studies on the impacts of igsnrr.ac.cn/). Soil data at 1 km resolution were from the Cold and land use change on hydrological processes, they mostly focused on the Arid Regions Sciences Data Center at Lanzhou (http://westdc.westgis. whole basin or a single subbasin without investigating the differences ac.cn). Historical climate data from 21 national meteorological sta between sub-regions and landforms, which are very important for hy tions during 1951–2016 were provided by the China Meteorological drological characteristics. In fact, owing to policies, economic devel Administration (CMA), including daily precipitation, temperature, wind opment, and topographical conditions, the sub-regions of the WRB have speed, relative humidity and solar duration. Streamflow data of three

2 J. Hu et al. Journal of Hydrology 593 (2021) 125741

Fig. 1. Location of the Wei River Basin (WRB), monitoring stations, and four sub-regions (the upper reach of the Wei River (UWR), the Jing River subbasin (JR), the North Luo River subbasin (NLR), and the middle and lower reaches of the Wei River (MLWR)). hydrological stations (Linjiacun, Huaxian, Zhuangtou) in the WRB 2.5. Land use scenarios covered 1971–1990 were obtained from the Hydrology Bureau of the Yellow River Conservancy Commission. The ET data were obtained from Using three periods of land use data (1980, 1995, and 2010) and land the MODIS MOD 16 A3 data products (http://www.ntsg.umt. use transition matrix method (Appendix 2), this study reclassified land edu/project/modis/mod16.php). use into six types (cropland, forest, grassland, water, urban land, and barren) to analyze spatial and temporal heterogeneity of LUCC in the 2.3. Model description and setup WRB by these three aspects: (1) total land use change, (2) land use change in different regions, (3) land use change based on landforms. To The SWAT model has been widely used to simulate the impacts of compare the changes in different periods, we considered 1980–1995 as climate change and land management practices on river discharge, the first period and 1995–2010 as the second period. Besides, we water quality, and sediment yield in basins around the world (Gan and assessed the hydrological impacts of changing land use in the past de Zuo, 2016; Tan et al., 2017; Zhao et al., 2020). The hydrological com cades and hypothetical sloping land use conversion (SLC) scenarios in ponents in the SWAT is based on the water balance equation (Arnold space and time. et al., 1998): 2.5.1. Hydrological impacts of changing land use in the past decades ∑t We conducted the spatiotemporal analysis of land use change using SWt = SW + (Ri Qi ETi Pi QRi) (1) i=1 three land use conditions (land use maps from 1980, 1995, and 2010, corresponding to S1980, S1995, and S2010, respectively) to understand where SW is the soil water content, i is the time t (days) for the the historical land use change of the WRB. The land use of 1980 rep simulation period, R (mm), Q (mm), ET (mm), P (mm), and QR (mm) are resented lower human activity, the land use of 1995 indicated primary the daily amounts of precipitation, runoff, actual evapotranspiration, economic developing of rural area, and 2010 land use represented result percolation, and return flow, respectively. The main outputs of SWAT due to urban construction and ecological reconstruction. Firstly, the are surface runoff, lateral flow, baseflow, ET, soil water, water yield, changes of specificland use types were identifiedby overlaying the land sediment load, and nutrient loads. Further details about SWAT are use maps for different years. To isolate the impact of LUCC and climate available in its theoretical documentation (Neitsch et al., 2011). The change, we fixed the climate data from 1971 to 2015 when we ran the well-established SWAT model has been increasingly used to address SWAT model. Then, we assessed the main hydrological components (ET, numerous watershed issues especially under climate shifts and land use soil water, and water yield) under different land use conditions at the changes. Based on the DEM and digital stream network information, the HRU level. WRB was divided into 196 subbasins and 4,629 HRUs using a threshold of 4%, 5%, and 5% for land use, soil type, and slope, respectively. 2.5.2. Hydrological impacts of hypothetical slopping land use conversion (SLC) program 2.4. Model calibration and validation For a long time, forest reclamation and improper cultivation caused serious soil erosion. One of the counter actions taken by the government The Sequential Uncertainty Fitting version 2 (SUFI-2) algorithm was to halt the environmental degradation in the CLP was the introduction of adopted for the parameter optimization. The monthly streamflow data the Slope Land Conversation Program/Crop conversion Program in – were available for 20 years (1971 1990), from which the first ten-year 1999, stopping agricultural activities in slope areas, recovering forest – (1971 1980) record of monthly streamflow was used to calibrate the based on principle of suitable land and tree (Ostwald and Chen, 2006). – model, and the other ten-year (1981 1990) record was used for vali The program required that on the basis of stopping cultivation of sloping dation. We used a series of numeric criteria to evaluate the model per cropland which was easy to cause soil erosion and desertification, formance, including the Nash-Sutcliffe efficiency (NSE), coefficient of 2 appropriate tree species should be selected according to local conditions determination (R ), and percentage bias (PB). More details about model for afforestation and vegetation restoration to achieve the effect of soil ◦ calibration and validation were shown in Appendix 1. According to and water conservation. And reclamation on steep slope above 25 was Moriasi et al. (2007), model performance could be evaluated as ‘satis prohibited (Chen et al., 2015). Based on these policies, we divided the factory’ if NSE > 0.5, R2 > 0.5, and PB <±25% for monthly simulations.

3 J. Hu et al. Journal of Hydrology 593 (2021) 125741

◦ ◦ ◦ ◦ ◦ ◦ ◦ ◦ slope into the following fivelevels: 0 –6 , 6 –15 , 15 –25 , 25 –35 , and > 0.5) (Moriasi et al., 2007), the model performance could be rated as ◦ >35 (Fig. A2). Then we set up four hypothetical land conversion sce ‘very good’ in both calibration and validation periods. The simulation narios to depict the potential land use pattern. Scenarios S1 and S2 results of Zhuangtou (Fig. 2b) and Huaxian (Fig. 2c) were also within an ◦ involved the conversion of cropland on slopes >25 to forest and acceptable range according to the above criteria. The simulated ET in grassland, and Scenarios S3 and S4 involved the conversion of cropland different land use types against the MODIS ET were shown in Fig. 3. We ◦ on slopes >15 to forest and grassland. found that the R2 for cropland, forest, and grassland were 0.50, 0.74, and 0.76, respectively, and the |PB| were no >10%. Therefore, both 3. Results visual comparison and numerical evaluation demonstrated that the SWAT model performed well for simulating the hydrological processes 3.1. Model performance evaluation in the WRB.

A visual and numerical comparison of the simulated monthly 3.2. LUCC in the Wei River basin from 1980 to 2010 streamflow against the observed data during the calibration (1971–1980) and validation (1981–1990) periods was shown in Fig. 2. 3.2.1. Total land use change The evaluation indicators NSE, R2, and PB of Linjiacun flow gaging Fig. A1 showed that the cropland is mainly located in the south station for the UWR were 0.78, 0.79, and 3.9% for the calibration, and eastern part of the basin, the grassland is relatively evenly distributed, 0.82, 0.84, and 7.3% for validation, respectively (Fig. 2a). According to and the forest is mainly distributed in the Qinling Mountain region and evaluation criteria for a monthly step (NSE > 0.75, PB <±10%, and R2 the southern contiguous area of the Jing River and the North Luo River

Fig. 2. Monthly streamflow simulation at (a) Linjiacun, (b) Zhuangtou, and (c) Huaxian gaging stations during the ten-year (1971–1980) calibration and ten-year (1981–1990) validation periods.

4 J. Hu et al. Journal of Hydrology 593 (2021) 125741

Fig. 3. Comparison of the simulated and remotely sensed evapotranspiration (ET) in the three main land use types ((a) cropland, (b) forest, (c) grassland) dur ing 2000–2014. subbasins. Farmland, grassland, and forest were the main land use types cropland and grassland. The distribution direction and gravity center of (Table S1), accounting for >96% of the WRB. Table 2 listed the land use different land use types can reflect the spatial evolution process of the transition matrix in two periods (1980–1995 and 1995–2010). We found land use change and the transformation characteristics of various land that in the firstperiod (1980–1995), cropland was mainly converted into types clearly. From the spatial variation map of the WRB (Fig. S1), grassland, with the conversion area of 16,278 km2, followed by forest cropland mainly transferred to the eastern part of the basin during the and urban land. The forest was mainly converted into grassland and period of 1980–2010, and its discretization pattern changed slightly. cropland, with the conversion area of 4874 km2 and 2706 km2, Forest mainly transferred to the northern part of the basin with a respectively. Grassland was mainly converted into cropland and forest, discrete distribution especially in the northern of the WRB, and the with conversion area of 16,057 km2 and 4795 km2, respectively. The center of gravity of grassland mainly moved to the northeastern part of transfer direction between different land use types in the second period the basin. As for urban land, it mainly shifted to the southeastern part of (1995–2010) was the same as the first period. In addition, urban land the WRB and represented a centralized trend. mainly came from cropland, and water and barren mainly turned into

5 J. Hu et al. Journal of Hydrology 593 (2021) 125741

Table 1 Slope characteristics of each land use type at baseline and under different slope land conversion scenarios (unit: km2).

Scenarios Landform Slope Cropland Forest Grassland Water Urban land Barren ◦ BS (2010) Plain <15 25,964 3069 9921 606 2989 73 ◦ ◦ 15 –25 3783 1784 5298 40 102 2 ◦ >25 424 377 771 7 8 1 ◦ Hill <15 12,904 2230 13,084 105 317 46 ◦ ◦ 15 –25 4787 1167 5785 25 80 23 ◦ >25 316 95 477 5 3 5 ◦ Mountain <15 6546 5456 7375 114 297 9 ◦ ◦ 15 –25 2243 4988 4894 12 32 3 ◦ >25 476 3162 2188 13 8 7

◦ ◦ S1 (>25 , C-F) Plain <15 – – – – – – ◦ ◦ 15 –25 – – – – – – ◦ >25 0 801 – – – – ◦ Hill <15 – – – – – – ◦ ◦ 15 –25 – – – – – – ◦ >25 0 411 – – – – ◦ Mountain <15 – – – – – – ◦ ◦ 15 –25 – – – – – – ◦ >25 0 3638 – – – –

◦ ◦ S2 (>25 , C-G) Plain <15 – – – – – – ◦ ◦ 15 –25 – – – – – – ◦ >25 0 – 1195 – – – ◦ Hill <15 – – – – – – ◦ ◦ 15 –25 – – – – – – ◦ >25 0 – 793 – – – ◦ Mountain <15 – – – – – – ◦ ◦ 15 –25 – – – – – – ◦ >25 0 – 2664 – – –

◦ ◦ S3 (>15 , C-F) Plain <15 – – – – – – ◦ ◦ 15 –25 0 5567 – – – – ◦ >25 0 801 – – – – ◦ Hill <15 – – – – – – ◦ ◦ 15 –25 0 5954 – – – – ◦ >25 0 411 – – – – ◦ Mountain <15 – – – – – – ◦ ◦ 15 –25 0 7231 – – – – ◦ >25 0 3638 – – – –

◦ ◦ S4 (>15 , C-G) Plain <15 – – – – – – ◦ ◦ 15 –25 0 – 9081 – – – ◦ >25 0 – 1195 – – – ◦ Hill <15 – – – – – – ◦ ◦ 15 –25 0 – 10,572 – – – ◦ >25 0 – 793 – – – ◦ Mountain <15 – – – – – – ◦ ◦ 15 –25 0 – 7137 – – – ◦ >25 0 – 2664 – – –

Note: “–” used in columns 4–9 means no change compared to the baseline (BS with 2010 land use), C-F means the conversion of cropland to forest, and C-G means the conversion of cropland to grassland.

Table 2 Transition matrix of land use types in the WRB from 1980 to 2010 (unit: km2).

Period Land use Cropland Forest Grassland Water Urban land Barren

1980–1995 Cropland 37,506 2342 16,278 437 2149 55 Forest 2706 14,089 4874 48 105 12 Grassland 16,057 4795 28,493 160 310 92 Water 389 59 197 237 36 8 Urban land 1932 88 365 28 666 1 Barren 56 40 37 7 6 80

1995–2010 Cropland 36,753 2992 16,024 401 2422 54 Forest 2256 14,088 4899 50 111 9 Grassland 16,041 5135 28,441 184 411 32 Water 426 49 157 249 34 2 Urban land 1974 106 308 33 848 3 Barren 60 16 94 8 1 69

3.2.2. Land use change in different sub-regions cropland, forest, and grassland (Fig. 4). From Fig. A1, we found that As described previously (see Section 2.1), we divided the WRB into forest and grassland were mainly distributed in the marginal area of four sub-regions to study the spatial heterogeneity of land use changes each region, cropland was mainly distributed in the low altitude area, (Fig. 1). The main land use types of these four sub-regions were and urban land was mainly located near the river channel. The area of

6 J. Hu et al. Journal of Hydrology 593 (2021) 125741

Fig. 4. Area percentage and relative changes of different land use types in the upper reach of the Wei River (UWR, (a) and (b)), the Jing River subbasin (JR, (c) and (d)), the North Luo River subbasin (NLR, (e) and (f)), and the middle and lower reaches of the Wei River (MLWR, (g) and (h)). The gap in the bar chart indicates the breakpoint in the Y-axis. cropland in the UWR was greater than that of grassland and forest, and the rest. Besides, urban land increased in all sub-regions, especially in the cropland increased in the firstperiod (during 1980–1995) and then the MLRB, which was mainly transferred from cropland (Table S2). decreased (during 1995–2010) (Fig. 4a and b). In the Jing River, the About 1,299 km2 of cropland was converted to urban land during area of cropland and grassland was basically equal, and the variation of 1995–2010, which was>1,114 km2 during 1980–1995. cropland was basically the same with that in the UWR (Fig. 4c and d). Cropland was mainly distributed around the urban land and on the 3.2.3. Land use change based on landforms banks of the Jing River and the Wei River (Fig. A1). The area of grass As described previously (see Section 2.1), we divided the WRB into land in NLR was obviously greater than that of cropland and forest three landforms (i.e., mountain, hill, and plain) to study the spatial (Fig. 4e). As for MLWR, the urban land was obviously greater than that heterogeneity of land use changes (Fig. A1). Fig. 5 represented the in other three sub-regions, and the area continued to increase in past 31 composition of land use area in each of the three landform zones. The years (1980–2010) (Fig. 4g and h). In general, the area of cropland in all main types of land use in mountain area were forest and grassland, ac sub-regions represented a decreasing trend during 1980–2010, espe counting for >70%. The cropland and grassland were major land use cially in the MLWR. The forest both in UWR and MLWR increased first types both in hill and plain area, and the cropland area in plain can be and then decreased. The change of grassland area in MLWR was larger >50%. During the 31-year study period, the area of forest and urban than other regions, with an increase in the firstperiod and a decrease in land among these three parts showed an increasing trend. However, the

7 J. Hu et al. Journal of Hydrology 593 (2021) 125741

3.3. Hydrological impacts of LUCC

3.3.1. Hydrological impacts of the past land use changes We fixedthe input DEM, soil data, and climate data (1951–2016) for the SWAT model, and got the hydrological results from 1956 to 2016 (warm-period was 1951–1955) based on land use data for three periods (1980, 1995, and 2010). Then we processed the annual mean ET, soil water, and water yield on HRU level and showed the results in Fig. 7. Based on S1980, S1995, and S2010, we found that the soil water content increased slightly by 6.2% and water yield decreased by 5.3% with the land use change from 1980 to 2010, while ET had nearly no change. We found that in comparison with the northwestern part, the southeastern part of basin had a higher ET value, which is consistent with the spatial distribution of precipitation. Compared with 1980, ET increased mainly in the north part of WRB (in NLR and Jing River) under the 2010 land use condition. Higher soil water content was found in the eastern part with an increasing gradient from the west to the east, and it decreased in most part of the southern WRB under 2010 land use condition. The spatial distribution of water yield indicated that water resources were more abundant in the southern part, while less water would be in the NLR and the Jing River. The comparison between 1980 and 2010 showed that water yield tended to decline in the areas where ET increased sharply. From Table 3, we found that ET increased slightly in the NLR and the Jing River, and soil water content showed an obvious increasing tread in the same regions. However, the result of water yield was opposite, showing a strong decreasing trend. This phenomenon was greatly related to the large-scale changes of forest. Because the water consumption of forest was greater than other land use types, the affor estation/reforestation would reduce streamflow.

3.3.2. The sloping land conversion (SLC) program According to the Slope Land Conversation Program/Crop conversion ◦ Program, we set up four scenarios that transformed cropland of > 25 ◦ and > 15 into forest and grassland in mountain, hill, and plain (Table 1). We calculated the annual mean hydrological processes in the baseline condition (S2010) and four scenarios, then we compared the results based on three landforms (Fig. 8). Fig. 8 showed that the hy drological effects of the conversion of cropland into forest were greater than that of the conversion into grassland. The expanding area of the forest and grassland would increase the ET because ET in both scenarios was greater than that in the baseline period (2010), and ET under the conversion of forest scenario was slightly larger than that converted to grassland. The consumption of water resources in forest is obviously larger than that in grassland, so soil water content under the two sce Fig. 5. Proportion of land use in different landforms (plain, hill, and mountain) narios of conversion to forest and grassland was significantlylower than of the WRB from 1980 to 2010. The gap in the bar chart indicates the break that in the baseline period, and soil water content under forest scenario point in the Y-axis. was the lowest. Forest had a greater effect on reducing water yield than grassland because forest canopy and litter layer have good ability to area of cropland had an opposite trend, especially in the plain area intercept and conserve water and forest could consume more water than where cropland was mainly converted to forest (the major conversion grass. Consequently, forest would significantlyreduce streamflow.Slope outcome), grassland, and urban land. Grassland showed an increasing also played an important role for hydrological responses to LUCC, that ◦ trend in mountain and plain area because of the conversion of cropland, is, the hydrological effects when cropland with slopes > 15 was con while the grassland of hill showed a decreasing trend because of the verted into grassland or forest was more significant than with slopes > ◦ conversion of grassland area to forest area. Table S3 showed the area 25 was converted. From Fig. 8a, we found that the increase of ET of ◦ ◦ percentage of three landform conditions in each sub-region of the WRB, slope > 15 was more dramatic that of slope > 25 . Fig. 8b showed that and we found that the topographical features of the UWR was mainly soil water content declined by a greater amount in the scenario of slope ◦ hill, the MLWR were mainly mountain and plain area, and hill and plain > 15 . area were the main parts of the Jing River and the NLR. From Fig. 6, we > found that the area of cropland in the hill of the UWR could be 50%, 4. Discussion and its area increased first(1980 –2010) and then decreased from 1995 to 2010. The proportion of cropland and grassland area in hill and plain 4.1. Driving forces of land use changes in different sub-regions and of the Jing River was larger than other land use types, the sum of which landforms was >80%. The area of cropland continued to decrease, while area of grassland decreased in the firstperiod (1980–1995) and then increased Land use change is a direct manifestation of the combination of na from 1995 to 2010. The area of urban land constantly increased, and the ture, society and ecosystems in Earth’s Critical Zone, reflecting the in growth rate was getting faster during the study period. tensity of human activities and changes in natural factors over time

8 J. Hu et al. Journal of Hydrology 593 (2021) 125741

Fig. 6. Area percentage of different land-use types under three landforms (plain, hill, and mountain) in four sub-regions: (a) the upper reach of the Wei River (UWR), (b) the Jing River subbasin (JR), (c) the North Luo River subbasin (NLR), and (d) the middle and lower reaches of the Wei River (MLWR).

(Banwart et al., 2013; Liu et al., 2014). The driving forces of LUCC regions. The forest and grass areas in the NLR, located in the eastern part mainly include natural driving factors (e.g., landform and natural of the WRB, was larger than that in the UWR. Besides, some extreme disaster), and socioeconomic driving forces (e.g., population growth, meteorological events can also lead to the changes of land use types. industry development, and government behavior) (Dang and Kawasaki, Several serious droughts in the UWR during 1980 ~ 2010 caused many 2017; Liu et al., 2010). The natural driving factors are relatively more trees to die in the mountainous areas, and a certain range of forest was stable and have cumulative effects, and the socioeconomic driving fac mainly converted into grassland, resulting in a decrease in forest and an tors are more active, with direct impacts on land use change (Ni, 2005; increase in grassland (Chang et al., 2016; Huang et al., 2014). Serra et al., 2008; Xie et al., 2014). In the WRB, precipitation reduces In addition, the socio-economic driving factors played an important from southeast to northwest, and previous studies have shown that role in the spatial heterogeneity of land use change. The UWR and Jing vegetation coverage is positively correlated with precipitation (Huang River are mainly located in Gansu Province, and the NLR and MLWR are et al., 2015). We found that cropland was mainly distributed in the mainly located in Shaanxi. During 1980–1995, China implemented the MLWR, because this region had more precipitation than other three sub- family contract responsibility system in rural areas, leading to the

9 J. Hu et al. Journal of Hydrology 593 (2021) 125741

Fig. 7. Simulated ET, soil water, and water yield under the different land use scenarios (S1980, S1995, and S2010 in the left panel) and the difference of each variable between 1980 and 2010 (the right panel).

Table 3 Simulated ET, soil water content, and water yield in the four sub-regions under different land use scenarios.

Region ET Soil water content Water yield

S1980/mm S2010/mm Change rate /% S1980/mm S2010/mm Change rate /% S1980/mm S2010/mm Change rate /%

NLR 496.54 505.34 1.8 342.08 378.93 10.8 46.6 37.78 18.9 JR 444.38 449.48 1.2 299.56 335.96 12.2 54.16 48.52 10.4 UWR 421.62 422.47 0.2 111.48 113.4 1.7 69.27 68.64 0.9 MLWR 503.24 502.33 0.2 346.09 339.29 2.0 105.1 105.73 0.6

Note: UWR, JR, NLR, and MLWR represent the upper reach of the Wei River, the Jing River subbasin, the North Luo River subbasin, and the middle and lower reaches of the Wei River, respectively.

the reform of the rural economic system, the market demand for agri cultural products tended to be diversified,and the production and life of farmers began to shift from the relatively simple production of culti Table A1 Comparison of social and economic development in the upper reach of the Wei vated land to the diversification of agriculture, forestry and animal River (UWR), the Jing River subbasin (JR), the North Luo River subbasin (NLR), husbandry (Li et al., 2018b). The population growth, acceleration of and the middle and lower reaches of the Wei River (MLWR). urbanization, and implementation of the national policy of the grain-for- green program after 1998 led to a rapid decline in cropland, rapid Sub-region Population density GDP Number of ordinary high schools growth in urban land, and different trends in forest and grassland JR Low Middle Low (Zhang et al., 2013). In the NLR, the grain-for-green has been vigorously NLR Low Middle Middle UWR Middle Low Low promoted, resulting in the rapid growth of forest area in the mountain MLWR High High High area, hills, and plains (Wu et al., 2014). The rapid growth of forest and the construction of a large number of check dams in the late 1990s have greatly reduced the grassland area (He et al., 2016; Zhao et al., 2013), and thus caused the reduction of grassland area in the entire WRB. conversion of most grassland and natural forest into cropland (Cheng Land use has both natural and social attributes, and the changes in its et al., 2009; Zhen et al., 2005). Therefore, the cropland area in the UWR, use patterns will also have a certain impact on the natural ecological Jing River, and NLR showed an increasing trend. The MLWR, as the core environment and human activities (Ananda and Herath, 2008; Antipova – area of Shaanxi’s economy, is mainly located in the Guanzhong Plain. et al., 2011). Generally speaking, during the periods of 1980 1995 and – During this period, the government encouraged urban development, 1995 2010, the transfer direction of land use was mainly the mutual causing great progress in urban construction and city size. The large- transformation of cropland, forest and grassland. Compared with other scale conversion of cropland into urban land led to an increase in the landforms, the conversion of cropland mainly occurred in the plains. The area of urban land, especially in the plain area. From 1995 to 2010, with area of forest and grassland showed a trend of rising in mountain area and hills, where the ecological environment was more vulnerable.

10 J. Hu et al. Journal of Hydrology 593 (2021) 125741

Further, the vegetation coverage increased significantly, which was beneficialto prevent soil erosion in the vulnerable areas. It can be seen that the land use pattern of the WRB has been increasingly optimal over the past 30 years, which is conducive to the harmonious development of human-land relations.

4.2. Hydrological effects of land use change

Fig. 7 showed that the ET under 2010 land use condition was significantly higher than that in 1980, while the water yield was reduced. At the same time, the changes were particularly obvious in the NLR and JR (Table 3). This might be because of the increase in area of forest and grassland after the implementation of some soil and water conservation practices. Under SLC scenarios, increases in grassland and ◦ ◦ forest converted from cropland with slop > 15 or > 25 had positive effects on ET, and negative effects on soil water and water yield. Wang et al. (2013) found that grass cover types cannot protect the soil surface from solar radiation, leading to greater water loss via direct evaporation. Forest could capture more rainfall and uptake more water than other land use types (e.g., cropland), resulting in a lower water yield (Crockford and Richardson, 2000; Li et al., 2017). This result was similar with Yang and Lu (2018), who found that the Grain-for-Green Pro gramme reduced the runoff by 13.8% and the sediment load by 50.7% in Yanhe River basin. However, Li et al. (2018a)showed that the shrinkage of woodland would cause decrease of soil water, thus contributing to a small increase in streamflow during the dry season in the WRB. The hydrological effects of conversion of cropland into forest were greater than those of grassland. Compared with grassland, the large canopy, big leaf area, and deep root of forest would result in greater ET (Qubaja et al., 2020; Wang, 2000). Besides, canopy interception and root water absorption could lead to more reduction of soil water in the forest land (Marimon-Junior et al., 2020; Nepstad et al., 1994). Jia et al. (2017) collected soil moisture content data within the 5-m soil profileand found that artificial forests intensified soil moisture decline on the Loess Plateau, inducing dry soil layer formation. Such type of land use shift would reduce water yield (streamflow) in forest area. Therefore, forest might lose more water through ET compared with other land use types (Wang et al., 2011). This pattern was evident from the change in soil water content, with less amount in forest than in other areas. Four SLC scenarios were proposed to demonstrate the hydrological impacts of land use change. Under the land conversion from farmland ◦ ◦ with slope > 15 or > 25 to grassland/forest, there was a negative impact on soil water and water yield and a positive impact on ET. Additionally, after the GGP was implemented, the area with steep slope has less obvious changes of ET than areas with gentle slope. Thus, be sides the variability of climate and soil, topographic factors also played important roles in the spatial heterogeneity of the water balance (Luo et al., 2020; Reggiani et al., 2000). Combined SWAT and water-energy coupling framework to detect the hydrology heterogeneity, Yang et al. (2019) found that the contribution of land surface change on hydrology shifts in the arid alpine region of Heihe River Basin was gradually increasing because of the strengthening effect of land surface change. Previous studies have found that revegetation can cause soil water shortages in both the near-surface soil and deep soil layers, which might result in soil desiccation (Feng et al., 2016; Jia and Shao, 2013; Jian et al., 2015). Vegetation restoration plays an important role in pro longing concentration time and reducing water yield on the hillslopes (Jia et al., 2020). From an artificialrainfall experiment, Gu et al. (2020) Fig. 8. Changes of ET (a), soil water (b), and water yield (c) under different found that vegetation cover, the decrease in soil bulk density, and the land-use conversion scenarios when compared to the baseline condition (S2010, increase in belowground root biomass were the primary causes of runoff black line) (see Table 1 for definitions of scenarios). and sediment yield reduction on the slopes with vegetation restoration. Consequently, watershed management should consider all hydrological components, and optimize vegetative structure and management mea sures in order to improve hydrological functions and promote sustain able development in terms of socioeconomics and ecology.

11 J. Hu et al. Journal of Hydrology 593 (2021) 125741

4.3. Limitations and uncertainties of LUCC over the past thirty years and four hypothetical land conversion scenarios based on the soil and water conservation practices. Our This study is subjected to some limitations that need to be resolved in analysis showed that the soil water content increased slightly by 6.2% future work. On the one hand, the grid resolution of land use data was 1 and the water yield decreased by 5.3% for 2010 when compared to km, which might be coarse for identifying land use conversion and 1980. The water yield in NLR and Jing River declined obviously, by detecting hydrological responses. Therefore, the problem of land use 18.9% and 10.4%, respectively, though the ET had nearly no change. In data may bring some uncertainties in assessing LUCC and their hydro terms of the spatial analysis, ET increased significantlyin the north part logical consequences. Mukundan et al. (2010) found that different data of WRB (mainly in the NLR and the Jing River), and the decline of water resolutions would affect the delineation of HRUs and sediment and yield was more severe in the same area. The scenarios about SLC pro nutrient predictions using SWAT. gram showed that the conversion of cropland to grassland or forest On the other hand, the internal structure of the model determines exerted negative effects on soil water and water yield, and the negative that some simulations are biased. The simulated streamflow showed effects were more evident in the scenario of afforestation. The conver some deviation in the flood season, which may be related to the “mea sion of land use types with different slopes also had different hydro surement quality” and “model behavior” reasons. For example, SWAT logical effects. We found that the function of increasing ET was more ◦ uses total daily precipitation and does not consider rainfall intensity effective when cropland with slopes > 15 was converted into grassland ◦ within a day. Thus, it may underestimate streamflow for some precipi or forest than with slopes > 25 was converted. In summary, this study tation events (Qiu et al., 2017). Also, the use of runoff curve number to provided useful information for land use planning and soil and water simulate the surface runoff behavior dose not account for saturation conservation on the Loess Platea. The quantitative assessment of LUCC excess runoff or contributions from variable source areas (Easton et al., and their impacts on hydrology can provide valuable information for 2008). In brief, the limitations mentioned above might compromise the optimization of land use planning, soil and water conservation, and the results of our study; however, the methodology and data used in our eventual sustainable watershed management on the Loess Plateau. study can definitely be accepted and support our conclusions. None theless, addressing these limitations by comparing different datasets and CRediT authorship contribution statement different models can be a good subject of future studies. Jingyi Hu: Conceptualization, Validation, Formal analysis, Investi 5. Conclusions gation, Writing - original draft, Writing - review & editing. Yiping Wu: Supervision, Resources, Writing - review & editing. Lijing Wang: In this study, we investigated the spatial heterogeneity of LUCC from Conceptualization, Methodology, Software, Visualization, Validation, 1980 to 2010 and its impacts on key hydrological components in the Investigation. Pengcheng Sun: Investigation, Writing - review & edit WRB. The SWAT model performed satisfactorily with acceptable NSE, ing. Fubo Zhao: Visualization, Validation, Formal analysis. Zhangdong 2 R , and PBIAS values. Jin: Validation, Writing - review & editing. Yunqiang Wang: Valida The LUCC analysis results showed that cropland, forest, grassland tion, Resources. Linjing Qiu: Methodology. Yanqing Lian: Resources. were the dominant land use types in study area. The cropland exhibited a decreasing trend from 58,776 km2 in 1980 to 57,519 km2 in 2010, leading to an increase in forest, grassland, and urban land. Among four Declaration of Competing Interest sub-regions, the cropland reduced most obviously in UWR, with a decrease of 498 km2. The grassland area decreased in the UWR and the The authors declare that they have no known competing financial NLR, but increased in the Jing River and the MLWR. Different landform interests or personal relationships that could have appeared to influence conditions also affected the distribution and transformation of land use. the work reported in this paper. The main types of land use in mountain area were forest and grassland, and the main transformation was from cropland to forest land and Acknowledgements grassland. The area of cropland in plain could be over 50%, and it was mainly converted to urban land. Based on the analysis of the charac This study was funded by the Strategic Priority Research Program of teristics and driving factors of land use dynamic change in 1980–1995 Chinese Academy of Sciences (XDB40020205), the National Science and 1995–2010, we found that the LUCC in the two stages showed Foundation of China (31961143011), the National Key Research and different characteristics because of different driving forces, and the Development Program of China (2019YFC0507403), Shaanxi Key change degree in the latter stage was greater than that in the previous Research and Development Program of China (2018ZDXM-GY-030), stage. National Thousand Youth Talent Program of China. We also thank the The impacts of LUCC on the water balance components were HPCC Platform in Xi’an Jiaotong University for computing equipment assessed quantitatively using SWAT setting scenarios with three periods and computer maintenance.

Appendix 1. . Model performance assessment

To measure the model performance, the Nash-Sutcliffe Efficiency (NSE), the coefficient of determination (R2), and the percentage bias (PBIAS) (Mandeville et al., 1970) were used in this study. These criteria were calculated as follows: ∑n 2 i=1(Qm.i Qs,i) NSE = 1 ∑n 2 (1) i=1(Qm.i Qm,avg)

∑n 2 2 [ i=1(Qm.i Qm,avg)(Qs.i Qs,avg)] R = ∑n 2∑n 2 (2) i=1(Qm.i Qm,avg) i=1(Qs.i Qs,avg)

∑n i=1(Qs.i Qm,i) PBIAS = ∑n × 100% (3) i=1Qm.i

12 J. Hu et al. Journal of Hydrology 593 (2021) 125741

where Qm.i and Qs,i are measured and simulated streamflowat each time step i; Qm,avg and Qs,avg are the mean measured and simulated streamflow; and n is the number of time steps. The NSE describes the explained variance for the observed values over time that is accounted for by the model (Green and van Griensven, 2008), which ranges from ∞ to 1. The R2, ranging from 0 to 1, is used to assess how accuracy the model tracks the changes of the observation. And the PBIAS measures the average difference between observation and simulation. The closer NSE and R2 are to 1, and PBIAS to 0, the better the SWAT model performs.

Appendix 2. . Land use transition matrix

Land use transition matrix is a fundamental tool in analyses of land use change. It includes not only the static information of land use types in a certain period, but also the dynamic conversion information of various land use types at the beginning and end of the period, which can better show the transition of land use types (Mas et al., 2004). The mathematical expression of the transition matrix is as follows: ⃒ ⃒ ⃒ S S ⋯ S ⃒ ⃒ 11 12 1n ⃒ ⃒ S S ⋯ S ⃒ S = ⃒ 21 22 2n ⃒ ij ⃒ ⋯ ⋯ ⋯ ⋯ ⃒ (4) ⃒ ⃒ Sn1 Sn2 ⋯ Snn

where S refers to the area or percentage of a certain land use type, Sii refers to the area or percentage of the land use type that remains unchanged in a period of time, Sij refers to the area or percentage of the land use type transformed from i land use type to j land use type in a period of time, and n refers to the type of land use type.

Appendix A. Supplementary data

Supplementary data to this article can be found online at https://doi.org/10.1016/j.jhydrol.2020.125741.

Fig. A1. Three landforms, four sub-regions (the upper reach of the Wei River (UWR), the Jing River subbasin (JR), the North Luo River subbasin (NLR), and the middle and lower reaches of the Wei River (MLWR)) and three-period (1980, 1995, and 2010) land use maps of the Wei River Basin.

13 J. Hu et al. Journal of Hydrology 593 (2021) 125741

Fig. A2. Changes of land use area ratio with elevation and slope in the WRB.

References Deng, L., Fei, K., Sun, T., Zhang, L., Fan, X., Ni, L., 2019. Characteristics of runoff processes and nitrogen loss via surface flow and interflow from weathered granite slopes of Southeast China. J. Mountain Sci. 16 (5), 1048–1064. Ananda, J., Herath, G., 2008. Multi-attribute preference modelling and regional land-use Devia, G.K., Ganasri, B.P., Dwarakish, G.S., 2015. A Review on hydrological models. planning. Ecol. Econ. 65 (2), 325–335. Aquat. Procedia 4, 1001–1007. Antipova, A., Wang, F., Wilmot, C., 2011. Urban land uses, socio-demographic attributes Dong, L., Xiong, L., Lall, U., Wang, J., 2015. The effects of land use change and and commuting: A multilevel modeling approach. Appl. Geogr. 31 (3), 1010–1018. precipitation change on direct runoff in Wei River watershed, China. Water Sci. Arnold, J.G., Srinivasan, R., Muttiah, R.S., Williams, J.R., 1998. Large area hydrologic Technol. 71 (2), 289–295. modeling and assessment part I: model development Jawra. J. Am. Water Resour. Easton, Z.M., Fuka, D.R., Walter, M.T., Cowan, D.M., Schneiderman, E.M., Steenhuis, T. Assoc. 34 (1), 73–89. S., 2008. Re-conceptualizing the soil and water assessment tool (SWAT) model to Banwart, S., Chorover, J., Gaillardet, J., Sparks, D., White, T., Anderson, S., et al., 2013. predict runoff from variable source areas. J. Hydrol. 348 (3–4), 279–291. Sustaining Earth’s critical zone basic science and interdisciplinary solutions for Fan, M., Shibata, H., 2015. Simulation of watershed hydrology and stream water quality global challenges. University of Sheffield, Sheffield, p. 48. under land use and climate change scenarios in Teshio River watershed, northern Cao, S., Wang, X., Wang, G., 2009. Lessons learned from China’s fall into the poverty Japan. Ecol. Ind. 50, 79–89. trap. J. Policy Model. 31 (2), 298–307. Feng, X., Fu, B., Piao, S., Wang, S., Ciais, P., Zeng, Z., et al., 2016. Revegetation in Chang, J., Li, Y., Wang, Y., Yuan, M., 2016. Copula-based drought risk assessment China’s Loess Plateau is approaching sustainable water resource limits. Nat. Clim. combined with an integrated index in the Wei River Basin, China. J. Hydrol. 540, Change 6 (11), 1019. 824–834. Foley, J.A., DeFries, R., Asner, G.P., Barford, C., Bonan, G., Carpenter, S.R., et al., 2005. Chen, J., Wu, H., Qian, H., Li, X., 2018. Challenges and prospects of sustainable Global consequences of land use. Science 309 (5734), 570–574. groundwater management in an agricultural plain along the Silk Road Economic Gan, R., Zuo, Q., 2016. Assessing the digital filter method for base flow estimation in Belt, north-west China. Int. J. Water Resour. Dev. 34 (3), 354–368. glacier melt dominated basins. Hydrol. Process. 30 (9), 1367–1375. Chen, Y., Li, Z., Fan, Y., Wang, H., Fang, G., 2014. Research progress on the impact of Gentine, P., Troy, T.J., Lintner, B.R., Findell, K.L., 2012. Scaling in surface hydrology: climate change on water resources in the arid region of Northwest China. Acta Progress and challenges. J. Contemporary Water Res. Educ. 147 (1), 28–40. Geographica Sinica 69 (9), 1295–1304. Ghaffari, G., Keesstra, S., Ghoddousi, J., Ahmadi, H., 2010. SWAT-simulated Chen, Y., Wang, K., Lin, Y., Shi, W., Song, Y., He, X., 2015. Balancing green and grain hydrological impact of land-use change in the zanjanrood basin, Northwest Iran. trade. Nat. Geosci. 8 (10), 739. Hydrol. Process. 24, 892–903. Cheng, G., 2002. Evolution of the concept of carrying capacity and the analysis Green, C.H., van Griensven, A., 2008. Autocalibration in hydrologic modeling: Using framework of water resources carrying capacity in northwest of China. J. Glaciol. SWAT2005 in small-scale watersheds. Environ. Modell. Software 23 (4), 422–434. Geocryol. 24 (4), 361–367. Gu, C., Mu, X., Gao, P., Zhao, G., Sun, W., Tan, X., 2020. Distinguishing the effects of L. Cheng Z.-x. Xu R. Luo Y.-j. Mi Spatial-temporal characteristics of LUCC and driving vegetation restoration on runoff and sediment generation on simulated rainfall on factor analysis for the wei river basin from 1980 to 2000 Research of Soil Water the hillslopes of the loess plateau of China. Plant Soil 447 (1), 393–412. Conservation 5 2009 1–6+30+287. Gwenzi, W., Nyamadzawo, G., 2014. Hydrological impacts of urbanization and urban Conrad, A., Marz,¨ M., Chakrabarti, R., 2019. Adapted subsurface water solutions across roof water harvesting in water-limited catchments: A review. Environ. Process. 1 (4), continents. J. Water Security 5, 1–12. 573–593. Crockford, R., Richardson, D., 2000. Partitioning of rainfall into throughfall, stemflow He, Y., Wang, F., Tian, P., Mu, X.-M., Gao, P., Zhao, G.-J., et al., 2016. Impact assessment and interception: effect of forest type, ground cover and climate. Hydrol. Process. 14 of human activities on runoff and sediment of Beiluo River in the Yellow River based (16–17), 2903–2920. on paired years of similar climate. Polish J. Environ. Stud. 25 (1), 121–135. Dang, A.N., Kawasaki, A., 2017. Integrating biophysical and socio-economic factors for Huang, J., Wu, P., Zhao, X., 2013. Effects of rainfall intensity, underlying surface and land-use and land-cover change projection in agricultural economic regions. Ecol. slope gradient on soil infiltration under simulated rainfall experiments. Catena 104, Model. 344, 29–37. 93–102.

14 J. Hu et al. Journal of Hydrology 593 (2021) 125741

Huang, S., Chang, J., Huang, Q., Chen, Y., 2014. Spatio-temporal changes and frequency Peng, L., Zhou, Z., Wang, P., Wu, C., Xiao, W., 2013. Land use/cover changes and analysis of drought in the Wei River Basin, China. Water Resour. Manage. 28 (10), landscape fragmentation analysis in the Three Gorges Reservoir Area. J. Food Agric. 3095–3110. Environ. 11, 2059–2064. Huang, S., Hou, B., Chang, J., Huang, Q., Chen, Y., 2015. Spatial-temporal change in Qiu, G., Yin, J., Tian, F., Geng, S., 2011. Effects of the “Conversion of Cropland to Forest precipitation patterns based on the cloud model across the Wei River Basin, China. and Grassland Program” on the water budget of the Jinghe River catchment in Theoretical Appl. Climatol. 120 (1–2), 391–401. China. J. Environ. Qual. 40 (6), 1745–1755. Jia, C., Sun, B., Yu, X., Yang, X., 2020. Analysis of runoff and sediment losses from a Qiu, L., Wu, Y., Wang, L., Lei, X., Liao, W., Hui, Y., et al., 2017. Spatiotemporal response sloped roadbed under variable rainfall intensities and vegetation conditions. of the water cycle to land use conversions in a typical hilly–gully basin on the Loess Sustainability 12 (5), 2077. Plateau, China. Hydrol. Earth Syst. Sci. (12), 6485. Jia, X., Ma, S., Zhu, Y., Luo, Y., 2017. Soil moisture decline due to afforestation across the Qubaja, R., Amer, M., Tatarinov, F., Rotenberg, E., Preisler, Y., Sprintsin, M., et al., 2020. Loess Plateau, China. J. Hydrol. 546, 113–122. Partitioning evapotranspiration and its long-term evolution in a dry pine forest using Jia, Y., Shao, M., 2013. Temporal stability of soil water storage under four types of measurement-based estimates of soil evaporation. Agric. For. Meteorol. 281, revegetation on the northern Loess Plateau of China. Agric. Water Manag. 117, 107831. 33–42. Reggiani, P., Sivapalan, M., Hassanizadeh, S.M., 2000. Conservation equations governing Jian, S., Zhao, C., Fang, S., Yu, K., 2015. Effects of different vegetation restoration on soil hillslope responses: Exploring the physical basis of water balance. Water Resour. Res. water storage and water balance in the Chinese Loess Plateau. Agric. Forest 36 (7), 1845–1863. Meteorol. 206, 85–96. Serra, P., Pons, X., Saurí, D., 2008. Land-cover and land-use change in a Mediterranean Jiang, Y., 2009. China’s water scarcity. J. Environ. Manage. 90 (11), 3185–3196. landscape: A spatial analysis of driving forces integrating biophysical and human Leng, G., Tang, Q., Huang, S., Zhang, X., Cao, J., 2016. Assessments of joint hydrological factors. Appl. Geogr. 28 (3), 189–209. extreme risks in a warming climate in China. Int. J. Climatol. 36 (4), 1632–1642. Shannon, M.A., Bohn, P.W., Elimelech, M., Georgiadis, J.G., Marinas, B.J., Mayes, A.M., Li, B., Pan, B., Han, J., 2008. Basic terrestrial geomorphological types in China and their 2010. Science and technology for water purification in the coming decades. In: circumscriptions. Quaternary Sci. 28 (4), 535–543. Nanoscience and technology: a collection of reviews from nature Journals. World Li, J., Zhou, Z., 2014. Landscape pattern and hydrological processes in Yanhe River basin Scientific, pp. 337–346. of China. Acta Geographica Sinica 69, 933–944. Sun, P., Wu, Y., Wei, X., Sivakumar, B., Qiu, L., Mu, X., et al., 2020. Quantifying the Li, J., Zhou, Z., 2015. Coupled analysis on landscape pattern and hydrological processes contributions of climate variation, land use change, and engineering measures for in Yanhe watershed of China. Sci. Total Environ. 505, 927–938. dramatic reduction in streamflow and sediment in a typical loess watershed, China. Li, L., Wang, Y., Beringer, J., Shi, H., Cleverly, J., Cheng, L., et al., 2017. Responses of LAI Ecol. Eng. 142, 105611. to rainfall explain contrasting sensitivities to carbon uptake between forest and non- Tan, M.L., Yusop, Z., Chua, V.P., Chan, N.W., 2017. Climate change impacts under forest ecosystems in Australia. Sci. Rep. 7 (1), 1–9. CMIP5 RCP scenarios on water resources of the Kelantan River Basin, Malaysia. Li, Y., Chang, J., Luo, L., Wang, Y., Guo, A., Ma, F., et al., 2018a. Spatiotemporal impacts Atmos. Res. 189, 1–10. of land use land cover changes on hydrology from the mechanism perspective using Teutschbein, C., Seibert, J., 2012. Bias correction of regional climate model simulations SWAT model with time-varying parameters. Hydrol. Res. 50 (1), 244–261. for hydrological climate-change impact studies: Review and evaluation of different Li, Y., Chang, J., Wang, Y., Jin, W., Bai, X., 2016a. Spatiotemporal responses of runoff to methods. J. Hydrol. 456, 12–29. land use change in Wei River Basin. Trans. Chin. Soc. Agric. Eng. 32, 232–238. Wang, B., Oguchi, T., Zhang, L., 2019. Simulating impact of over-grazing on grassland Li, Y., Li, X., Tan, M., Wang, X., Xin, L., 2018b. The impact of cultivated land spatial shift degradation using GIS: A case study in the Heihe River Basin, China. Abstracts ICA 1, on food crop production in China, 1990–2010. Land Degradation Develop. 29 (6), 1–2. 1652–1659. Wang, G., Yang, H., Wang, L., Xu, Z., Xue, B.L., 2014. Using the SWAT model to assess Li, Y., Liu, C., Liu, X., Liang, K., Bai, P., Feng, Y., 2016b. Impact of the grain for green impacts of land use changes on runoff generation in headwaters. Hydrol. Process. 28, project on the land use/cover change in the middle Yellow River. Nat. Resour. J. 31, 1032–1042. 2005–2020. Wang, J., Chen, Y., Shao, X., Zhang, Y., Cao, Y., 2012a. Land-use changes and policy Li, Z., Deng, X., Wu, F., Hasan, S.S., 2015. Scenario analysis for water resources in dimension driving forces in China: Present, trend and future. Land Use Policy 29 (4), response to land use change in the middle and upper reaches of the Heihe River 737–749. Basin. Sustainability 7 (3), 3086–3108. Wang, S., Fu, B., Gao, G., Liu, Y., Zhou, J., 2013. Responses of soil moisture in different Liu, J., Kuang, W., Zhang, Z., Xu, X., Qin, Y., Ning, J., et al., 2014. Spatiotemporal land cover types to rainfall events in a re-vegetation catchment area of the Loess characteristics, patterns, and causes of land-use changes in China since the late Plateau, China. Catena 101, 122–128. 1980s. J. Geog. Sci. 24 (2), 195–210. Wang, X., Zhang, J., Shahid, S., ElMahdi, A., He, R., Bao, Z., et al., 2012b. Water Liu, J., Zhang, Z., Xu, X., Kuang, W., Zhou, W., Zhang, S., et al., 2010. Spatial patterns resources management strategy for adaptation to droughts in China. Mitig. Adapt. and driving forces of land use change in China during the early 21st century. J. Geog. Strat. Glob. Change 17 (8), 923–937. Sci. 20 (4), 483–494. Wang, Y., Qin, D., 2017. Influence of climate change and human activity on water Liu, M., Tian, H., Yang, Q., Yang, J., Song, X., Lohrenz, S.E., et al., 2013. Long-term resources in arid region of Northwest China: An overview. Adv. Clim. Change Res. 8 trends in evapotranspiration and runoff over the drainage basins of the Gulf of (4), 268–278. Mexico during 1901–2008. Water Resour. Res. 49 (4), 1988–2012. Wang, Y., Yu, P., Feger, K.H., Wei, X., Sun, G., Bonell, M., et al., 2011. Annual runoff and Luo, J., Zheng, Z., Li, T., He, S., 2020. Temporal variations in runoff and sediment yield evapotranspiration of forestlands and non-forestlands in selected basins of the Loess associated with soil surface roughness under different rainfall patterns. Plateau of China. Ecohydrology 4 (2), 277–287. Geomorphology 349, 106915. Wang, Y.P., 2000. Large-scale vegetation feedbacks on a doubled CO2 climate. J. Clim. Mandeville, A.N., O’Connell, P.E., Sutcliffe, J.V., Nash, J.E., 1970. River flowforecasting 13 (7), 1313–1325. through conceptual models part III - The Ray catchment at Grendon Underwood. Wu, L., Yang, S., Liu, X., Luo, Y., Zhou, X., Zhao, H., 2014. Response analysis of land use J. Hydrol. 11 (2), 109–128. change to the degree of human activities in Beiluo River basin since 1976. Acta Marimon-Junior, B.H., Hay, J.D.V., Oliveras, I., Jancoski, H., Umetsu, R.K., Geographica Sinica 69 (1), 54–63. Feldpausch, T.R., et al., 2020. Soil water-holding capacity and monodominance in Xie, H., Liu, Q., Liu, G., 2014. Evolution characteristics and driving forces of wetland Southern Amazon tropical forests. Plant Soil 450 (1), 65–79. changes in the Poyang Lake eco-economic zone of China. ScientificRes. Essays 9 (2), Mas, J.F., Velazquez,´ A., Díaz Gallegos, J.R., Mayorga Saucedo, R., Alcantara,´ C., 24–34. Bocco, G., et al., 2004. Assessing land use/cover changes: a nationwide multidate Xu, W., Dong, Z., Hao, Z., Li, D., Ren, L., 2017. River health evaluation based on the fuzzy spatial database for Mexico. Int. J. Appl. Earth Obs. Geoinf. 5 (4), 249–261. matter-element extension assessment model. Polish J. Environ. Stud. 26, (3). Mekonnen, M.M., Hoekstra, A.Y., 2016. Four billion people facing severe water scarcity. Yang, K., Lu, C., 2018. Evaluation of land-use change effects on runoff and soil erosion of Sci. Adv. 2 (2), e1500323. a hilly basin — the Yanhe River in the Chinese Loess Plateau. Land Degrad. Dev. 29 Moran-Tejeda,´ E., Ceballos-Barbancho, A., Llorente-Pinto, J.M., 2010. Hydrological (4), 1211–1221. response of Mediterranean headwaters to climate oscillations and land-cover Yang, L., Feng, Q., Yin, Z., Deo, R.C., Wen, X., Si, J., et al., 2019. Regional hydrology changes: The mountains of Duero River basin (Central Spain). Global Planet. Change heterogeneity and the response to climate and land surface changes in arid alpine 72 (1–2), 39–49. basin, northwest China. Catena 104345. Moriasi, D.N., Arnold, J.G., Van Liew, M.W., Bingner, R.L., Harmel, R.D., Veith, T.L., Yang, Q., Tian, H., Friedrichs, M.A.M., Liu, M., Li, X., Yang, J., 2015. Hydrological 2007. Model evaluation guidelines for systematic quantification of accuracy in responses to climate and land-use changes along the North American East Coast: A watershed simulations. Trans. ASABE 50 (3), 885–900. 110-year historical reconstruction. J. Am. Water Resour. Assoc. 51 (1), 47–67. Mukundan, R., Radcliffe, D.E., Risse, L.M., 2010. Spatial resolution of soil data and Yao, C., Chang, L., Ding, J., Li, Z., An, D., Zhang, Y., 2014. Evaluation of the effects of channel erosion effects on SWAT model predictions of flow and sediment. J. Soil underlying surface change on catchment hydrological response using the HEC-HMS Water Conserv. 65 (2), 92–104. model. Proc. Int. Assoc. Hydrol. Sci. 364, 145–150. Neitsch, S.L., Arnold, J.G., Kiniry, J.R., Williams, J.R., 2011. Soil and water assessment Yao, T., Thompson, L.G., Mosbrugger, V., Zhang, F., Ma, Y., Luo, T., et al., 2012. Third tool theoretical documentation version 2009. Texas Water Resources Institute. pole environment (TPE). Environ. Develop. 3, 52–64. Nepstad, D.C., de Carvalho, C.R., Davidson, E.A., Jipp, P.H., Lefebvre, P.A., Negreiros, G. Yin, J., He, F., Xiong, Y.J., Qiu, G.Y., 2017. Effects of land use/land cover and climate H., et al., 1994. The role of deep roots in the hydrological and carbon cycles of changes on surface runoff in a semi-humid and semi-arid transition zone in Amazonian forests and pastures. Nature 372 (6507), 666–669. northwest China. Hydrol. Earth Syst. Sci. 21 (1), 183. Ni, S., 2005. Some issues in the study on land use/cover change. J. Natural Resourc. 6 Zhai, R., Tao, F., 2017. Contributions of climate change and human activities to runoff (20), 932–937. change in seven typical catchments across China. Sci. Total Environ. 605, 219–229. Ostwald, M., Chen, D., 2006. Land-use change: Impacts of climate variations and policies Zhang, H., Wang, B., Liu, D.L., Zhang, M., Leslie, L.M., Yu, Q., 2020. Using an improved among small-scale farmers in the Loess Plateau, China. Land Use Policy 23 (4), SWAT model to simulate hydrological responses to land use change: A case study of a 361–371. catchment in tropical Australia. J. Hydrol. 585, 124822.

15 J. Hu et al. Journal of Hydrology 593 (2021) 125741

Zhang, K., Zhang, Y., Tian, H., Cheng, X., Dang, H., Zhang, Q., 2013. Sustainability of Zhao, G., Mu, X., Tian, P., Wang, F., Gao, P., 2013. Climate changes and their impacts on social–ecological systems under conservation projects: Lessons from a biodiversity water resources in semiarid regions: A case study of the Wei River basin, China. hotspot in western China. Biol. Conserv. 158, 205–213. Hydrol. Process. 27 (26), 3852–3863. Zhang, S., Wu, Y., Sivakumar, B., Mu, X., Zhao, F., Sun, P., et al., 2019. Climate change- Zhao, J., Huang, Q., Chang, J., Liu, D., Huang, S., Shi, X., 2015. Analysis of temporal and induced drought evolution over the past 50 years in the southern Chinese Loess spatial trends of hydro-climatic variables in the Wei River Basin. Environ. Res. 139, Plateau. Environ. Modell. Software 122, 104519. 55–64. Zhang, X., 2005. Land-use/cover change process and driving force analysis in jinghe Zhen, L., Xie, G., Yang, L., Cheng, S., Guo, G., 2005. Land-use change dynamics, driving river channels. J. Soil Water Conserv. 19 (6), 137–140. forces and policy implications in Jinghe watershed of western China. Resourc. Sci. 4, Zhao, F., Wu, Y., Yao, Y., Sun, K., Zhang, X., Winowiecki, L., et al., 2020. Predicting the 33–37. climate change impacts on water-carbon coupling cycles for a loess hilly-gully Zhou, D., Zhao, S., Zhu, C., 2012. The Grain for Green Project induced land cover change watershed. J. Hydrol. 581, 124388. in the Loess Plateau: A case study with Ansai County, Shanxi Province, China. Ecol. Ind. 23, 88–94.