Analysis of Basin Morphologic Characteristics and Their Inﬂuence on the Water Yield of Mountain Watersheds Upstream of the Xiongan New Area, North China

: In this study, based on the DEM, we extracted the drainage networks and watersheds of the Daqing River with ArcGIS, investigated the basin characteristics and the differences in their spatial distributions and analyzed the relations of the drainage density with some surface conditions and how the drainage density inﬂuenced the water yield. The results suggested a power function between the mainstream length and drainage area, showing that with the increase in basin area, the basins became longer. The result of the power function between the relief and drainage area with negative exponent values means the relief changed more slowly with increasing basin area. The values of the circularity ratio and elongation ratio indicate that the basin shape of the mountain watersheds in the Daqing River was narrow and predisposed to ﬂooding during periods of heavy rainfall. The orders of the streams in the mountain watersheds ranged from ﬁve to seven. The average bifurcation ratio of those nine mountainous watersheds reveals the order of the u + 1 rivers in each basin of the Daqing River was on average 4 times larger than that of order u rivers. The drainage density ( D d ) was high in the north and low in the south of the Daqing River. Rainfall was negatively correlated with drainage density, but the correlation between them was not signiﬁcant at the 0.05 level. Drainages developed in places with poor vegetation cover. The drainages in the southwest, north and west developed considerably, while drainages in the east and southeast did not develop much. Yet, the available data showed the impact of the watershed area, elongation ratio and drainage density on the water yield was not signiﬁcant. In contrast, there was a signiﬁcant positive correlation between channel slope and the water yield modulus. The hypsometric integrals and the relation between drainage density and hypsometric integral suggest that the landform evolution of the mountain basins along the Daqing River were in the old stage with no further increase trend of drainage density in the future.


Introduction
Mountain rivers are important corridors that link upland and lowland environments and mediate the supply, transport, andstorage of organic and inorganic materials [1,2]. Mountain rivers play an important role in flood control and perform other functions, such as conserving water resources, regulating the microclimate, and maintaining water ecology and biodiversity. However, they often face problems such as short-term flood responses, water shortages in the dry season, and river channel artificialization [3].
Mountain rivers are often confined by immobiletopographic features, such as bedrock and large boulders with channelgradients commonly exceeding 1% [2,4,5]. This leads to these types of rivershaving steep hydraulic rating curves that initiate rapid transport of small sand and gravel fractions within the large structural matrix formed by large, century-scale floods nested within an even larger geological context [2,6]. Stream network scholars in this basin study and a reference for further study of the water cycle process and effective utilization of water resources in the Daqing River basin and a scientific basis for the formulation of flood control policies and water conservation in the Xiongan New Area.

Study Area
The Daqing River is a primary tributary of the Haihe River in northern China and is located at 113 • 34 3 -117 • 46 7 E, 38 • 4 42 -40 • 3 2 N. The area of the watershed is 43,060 km 2 , with a length of 275 km, an average width of 156 km and an average slope of 5.22%. The average annual runoff is 4.3 × 10 8 m 3 . According to the Water Resource Zoning Map of the Haihe River basin, provided by the Haihe River Water Conservancy Commission (HRWCC), the Daqing River basin mainly consists of two geomorphic units: the upper reaches of the west are mountainous areas thataccount for 43% of the basin area, and the middle and lower reaches of the east are plains areas (Figure 1).The river basin is fan-shaped and divided into northern and southern branches. The southern branches of the mountain basin are mainly composed of six tributaries, including the Ci River (CIR), Sha River (SR), Tang River (TR), Jie River (JR), Cao River (CR) and Bao River (BR). After entering the plain, the Ci River and Sha River converge into the Zhulong River. The northern branches are mainly composed of four tributaries, namely, the Zhongyi River (ZYR), Beiyi River (BYR), Juma River (JMR) and Dashi River (DSR). After entering the plain, the Juma River divides into two branches, namely, the Beijuma River and Nanjuma River. These ten tributaries flow into Baiyangdian Lake. The terrain of the basin is higher in the northwest and lower in the southeast, with an elevation difference of nearly 2800 m ( Figure 1). Water2021, 13, 2903 3 of 15 of the mountain watersheds of the Daqing River; the relations of drainage density with topography, vegetation coverage and surface material composition; and discuss the implications of indices of drainage network and watershed geometry for water yield and watershed evolution. Our results may provide some basic watershed characteristic data for other scholars in this basin study and a reference for further study of the water cycle process and effective utilization of water resources in the Daqing River basin and a scientific basis for the formulation of flood control policies and water conservation in the Xiongan New Area.

Study Area
The Daqing River is a primary tributary of the Haihe River in northern China and is located at 113°34′3″-117°46′7″ E, 38°4′42″-40°3′2″ N. The area of the watershed is 43,060 km 2 , with a length of 275 km, an average width of 156 km and an average slope of 5.22%. The average annual runoff is 4.3×10 8 m 3 . According to the Water Resource Zoning Map of the Haihe River basin, provided by the Haihe River Water Conservancy Commission (HRWCC), the Daqing River basin mainly consists of two geomorphic units: the upper reaches of the west are mountainous areas thataccount for 43% of the basin area, and the middle and lower reaches of the east are plains areas (Figure 1).The river basin is fan-shaped and divided into northern and southern branches. The southern branches of the mountain basin are mainly composed of six tributaries, including the Ci River (CIR), Sha River (SR), Tang River (TR), Jie River (JR), Cao River (CR) and Bao River (BR). After entering the plain, the Ci River and Sha River converge into the Zhulong River. The northern branches are mainly composed of four tributaries, namely, the Zhongyi River (ZYR), Beiyi River (BYR), Juma River (JMR) and Dashi River (DSR). After entering the plain, the Juma River divides into two branches, namely, the Beijuma River and Nanjuma River. These ten tributaries flow into Baiyangdian Lake. The terrain of the basin is higher in the northwest and lower in the southeast, with an elevation difference of nearly 2800 m ( Figure 1). The Daqing River has a warm temperate monsoon climate with distinct seasons and an uneven distribution of precipitation during the year that is mainly concentrated from July-September. The average annual rainfall ranges from 500-700 mm and often comes The Daqing River has a warm temperate monsoon climate with distinct seasons and an uneven distribution of precipitation during the year that is mainly concentrated from July-September. The average annual rainfall ranges from 500-700 mm and often comes in the form of heavy rain in July and August. The exposed lithology on the surface includes granite gneiss, limestone and loose quaternary deposits. The landform is dominated by mountains and basins. Coarse bone soilis the main soil in the mountains and loess covers the hill platform around the basin [29].

Data
The topographic maps of the study area projected to produce a raster DEM with a resolution of 30 m × 30 m were downloaded from the Geospatial Data Cloud (http://www. gscloud.cn/, access date: 8 October 2021). The normalized differential vegetation index (NDVI) products for the year 2018 with a 1 km spatial resolution were also downloaded from the Geospatial Data Cloud (http://www.gscloud.cn/). Soil data with a resolution of 1 km were extracted from the 1:1,000,000 Soil Database of China downloaded from the Nanjing Institute of Soil Science, Chinese Academy of Sciences (http://www.issas.cas. cn/). The rainfall and runoff data of nine hydrology gauges, including the Hengshanling reservoir station in the Ci River, Wangkuai reservoir station in the Sha River, Xidayang reservoir station in the Tang River, Longmen reservoir station in the Cao River, Huanglongsi station in the Jie River, Angezhuang reservoir station in the Zhongyi River, Wanglong reservoir station in the Beiyi River, Zijingguan station in the Juma River and Manshuihe station in the Dashi River (Figure 1), were from the hydrological yearbook of the Haihe River basin of China, collected by HRWCC.

The Extraction of River Networks
Based on the DEM, the hydrological analysis tool set in the spatial analysis module of ArcGIS10.5 software was used to extract the river network through a series of processing steps, including filling depression, calculating flow direction and flow confluences and accumulations. The accumulation area threshold was 0.1 km 2 , which is the minimum accumulation area fora bedrock channel with hydraulic erosion in the main channel [30]. Additionally, the threshold is the lower limit of the gullies whose length is larger than 400 m and width is more than 100 m in the loess hilly region [31].

The Variables for Expressing Basin Form
The morphological feature parameters included watershed geometry (area, main stream length, perimeter, circularity ratio and elongation ratio) and drainage morphology (relief, total stream length and drainage density). The detailed calculation methods for those parameters are shown in Table 1 [8].  [9] According to the satellite images and field investigations, only the land surface of the upper mountain watersheds is rigged with dense drainages. Therefore, the drainage density in the upper mountain watersheds is meaningful for characterizing the broken degree of the land surface and is investigated in this study. For utilization of the information of the soil data, which has a spatial resolution of 1 km, the drainage density was calculated through ArcGIS 10.5 with a 1 km × 1 km grid unit. The coarse grid distribution of drainage density was converted to its continuous spatial distribution through the Ordinary Kriging method for calculating the areas of different levels of drainage density.

Topographical Attributes
Topographical attributes include the slope gradient and aspect. The catchment slope (slope, degrees) and aspects were calculated using ArcGIS and the algorithm proposed by Burrough and McDonnell [34].

Hypsometric Integral
The hypsometric integral curve shows the basin topography on the coordinate with the ratio of horizontal area between the contour and the upper perimeter (a i , km 2 ) to the total drainage basin area (A, km 2 ) as the abscissa values (x i = a i /A) and the ratio of height of the contour above basin outlet (h i , m) to the total height of the basin (H t, m) as the ordinate values (y i = h i /H t ), expressed by the function y = f (x). The integral of this function is named a hypsometric integral and has a value between 0 and 1 [35].The hypsometric integral (HI) was calculated as follows, where x is the ratio of the horizontal area between the contour and the upper perimeter (km 2 ) to the total drainage basin area (km 2 ).

Analysis of Drainage Density
As a unique property of the landscape, D d is related the underlying geomorphic processes acting in a catchment based on its topography [36]. Usually expressed as the ratio of total channel length to total catchment area [9], D d is controlled by the local lithology [37], topography [38], vegetation [39,40] and regional climatic patterns [41,42]. Therefore, we analyzed the relationships between drainage density and precipitation, surface material composition, vegetation coverage and topography. The software of Statistical Product and Service Solutions (SPSS) and Microsoft Excel were used to analyze the correlations and trends.

Results and Discussion
Because of the small mountainous area of the Bao River, we extracted the river network and watershed boundaries of the other nine tributaries in mountainous areas from a DEM, including the Ci River, Sha River, Tang River, Jie River, Cao River, Zhongyi River, Beiyi River, Juma River and Dashi River ( Figure 2).

Watershed Geometry
Many watershed geometry features are related to the drainage area [43].We analyzed the correlation between the main stream length (L, km) and area (A, km 2 ) ( Figure  3), relief (H, m/km) and A ( Figure 4). Both relationships were power exponents, and the variation of trends were significant ( .82-0.99, p < 0.01); namely, with the increase of area, the main stream length increases and relief decreases significantly, respectively. If L was proportional to A 0.5 , the geometric morphology upstream was similar to that downstream [44]. As shown

Watershed Geometry
Many watershed geometry features are related to the drainage area [43].We analyzed the correlation between the main stream length (L, km) and area (A, km 2 ) ( Figure 3), relief (H, m/km) and A ( Figure 4). Both relationships were power exponents, and the variation of trends were significant (L = aA b , b = 0.64-1.87, R 2 = 0.95-0.99, p < 0.01; H = cA d , d = −1.36-−0.33, R 2 = 0.82-0.99, p < 0.01); namely, with the increase of area, the main stream length increases and relief decreases significantly, respectively. If L was proportional to A 0.5 , the geometric morphology upstream was similar to that downstream [44]. As shown in Figure 3, L and A were positively correlated, but the minimum b value was 0.64, which was greater than 0.5. Therefore, we believe that the basin forms of the mountain watersheds in the Daqing River change along the channel. With the increase in basin area, the basins became longer. H and A were negatively correlated, and the relief changed more slowly with increasing basin area. From Figures 3 and 4, except for the Dashi River, the b value and d value of the other eight rivers were close, which means that their basins had similar forms.  There are many indices expressing the basin shape [44], but only two parameters There are many indices expressing the basin shape [44], but only two parameters are widely used. They are the circularity ratio and elongation ratio ( Table 1).The circularity ratio (R c ) and elongation ratio (R e ) were additional parameters that were used to describe the watershed geometry morphology. If the drainage basin shape is round, the value of R c and R e should be equal to 1 and 1.275, respectively; if the shape is square, R c and R e should be 0.785 and 1.128. With the increase in stream length, R c continues to decease and R e approaches zero, and the shape becomes narrower and longer [32,33]. The R c of mountain watersheds in the Daqing River ranged from 0.14 to 0.36, and the values of R e were 0.36-0.94. We can conclude that the basin shape of the mountain watersheds in the Daqing River was narrow. According to Liu [45], the flood confluence time is short for a narrow basin. Therefore, the small basin circularity ratio of the Daqing River is favorable to the formation of high flood peaks, which easily delivered a large amount of runoff from the basins during the rainy day to the Xiongan New Area.  There are many indices expressing the basin shape [44], but only two parameters are widely used. They are the circularity ratio and elongation ratio ( Table 1).The circularity ratio (Rc) and elongation ratio (Re) were additional parameters that were used to describe the watershed geometry morphology. If the drainage basin shape is round, the value of Rc and Re should be equal to 1 and 1.275, respectively; if the shape is square, Rc and Re should be 0.785 and 1.128. With the increase in stream length, Rc continues to decease and Re approaches zero, and the shape becomes narrower and longer [32,33]. The Rc of mountain watersheds in the Daqing River ranged from 0.14 to 0.36, and the values of

River Network Morphology
As mentioned above, the upper and lower reaches of the Daiqing River have distinctive landforms. In the lower reaches of the Daiqing River, there are no tributaries on both banks. Thus, the river network morphological characteristics of the upper reaches of the Daiqing River, including river bifurcation ratio and drainage density were analyzed here.

Stream Order
The ordering scheme proposed by Strahler [43] was used here. The bifurcation ratio R b = N u /N u+1 , where N u and N u+1 denote the number of stream segments of order u and u+1, respectively. Watersheds with different natural geographical conditions tend to have different drainage bifurcation ratios [46]. For the watersheds in flat or hilly and gully regions, the R b is close to 2; for the watersheds in mountainous regions, it is 3-5. The R b in the mountain area is larger than that in the plain area. The R b grows with the development of a drainage system [44].
Strahler [43] pointed that, except for the areas with hard geological substrata, the differences of R b between regions were small. Morisawa [47] proved that the R b in fifteen small watersheds on the Appalachian Plateau was close to a constant. Therefore, the average drainage bifurcation ratio is the most reasonable parameter to reflect the drainage structure.
We ordered the extracted drainage network using the Strahler criterion [43]; namely, the primary finger-tip stream is the order I, and the new stream formed by the convergence of the two order I streams is the order II, and in this way, the streams in the whole basin will be ordered until the end. The channel that flows through the whole basin with the amount of water and sediment is called the highest order stream. Therefore, the stream orders of the mountain watersheds in the Daqing River ranged from five to seven.Specifically, the upper reaches of the Jie River and Beiyishui River were 5th-order streams, Ci River, Cao River and Dashi River were 6th-order streams, and the other four rivers were 7th-order streams. The average bifurcation ratio of nine rivers was calculated according to the stream order results. For mountain watersheds, the bifurcation ratio ranged from 3-4. The bifurcation ratio of areas with loose rocks, sparse vegetation, and heavy rainfall was high [46]. The average bifurcation ratio of those nine mountain watersheds was 3.8-4.8. In other words, the order of the u + 1 rivers in each basin of the Daqing River was on average 4 times larger than that of order u rivers.

Drainage Density
The drainage density of mountain watersheds in the Daqing River ranged from 2-2.3 km/km 2 , and the spatial difference was small. According to the method mentioned above, we obtained a spatially continuous distribution of drainage density ( Figure 5). D d can be divided into four areas, including 0-1 km/km 2 , 1-2 km/km 2 , 2-3 km/km 2 and above 3 km/km 2 . For mountain watersheds in the Daqing River, D d was mainly distributed from 1-2 km/km 2 and 2-3 km/km 2 ; specifically, the distribution areas of the southern watersheds fluctuated little, and the average distribution area of 1-2 km/km 2 and 2-3 km/km 2 was 71%. The 1-2 km/km 2 distribution area of the Juma River and Dashi River in northern watersheds increased significantly, and the average distribution area from 2-3 km/km 2 was 68%.

Relationship between Dd and Rainfall and Vegetation Cover
Collins and Bras [40] summarized the feedback of vegetation and runoff under varying mean annual precipitation levels in a schematic representation showing an initial increase in drainage density in arid areas, followed by a decrease in semiarid regions and an increase in humid environments.
We drew rainfall and Dd with SPSS as follows: Dd = −0.001xrainfall + 2.501 (R = 0.542, N = 9, p = 0.132). Rainfall is negatively correlated with drainage density, which agrees with the results of Collins and Bras [40] in semiarid regions, but the correlation between them did not show no significance at the 0.05 level. Although rainfall is the driving force behind gully development, for the small areas of mountain watersheds in the Daqing River, the average spatial variation in rainfall is not obvious.
Based on the change in NDVI in the upper reaches of each watershed in 2018, the correlation between the NVDI and drainage density was analyzed, and the correlation equation was obtained by SPSS as Dd = −0.198xNDVI + 1.227 (R = 0.738, N =9, p = 0.02). This indicated that gullies developed in places with poor vegetation cover but not in places with good vegetation cover. Of course, it is difficult to determine the causal relationship between these factors because there may be a positive feedback mechanism; that is, the development of gullies in places with poor vegetation is conducive to the development of gullies, while the development of gullies, strong erosion and soil erosion inhibit the growth of vegetation [8].

Relationship between D d and Rainfall and Vegetation Cover
Collins and Bras [40] summarized the feedback of vegetation and runoff under varying mean annual precipitation levels in a schematic representation showing an initial increase in drainage density in arid areas, followed by a decrease in semiarid regions and an increase in humid environments.
We drew rainfall and D d with SPSS as follows: D d = −0.001x rainfall + 2.501 (R = 0.542, N = 9, p = 0.132). Rainfall is negatively correlated with drainage density, which agrees with the results of Collins and Bras [40] in semiarid regions, but the correlation between them did not show no significance at the 0.05 level. Although rainfall is the driving force behind gully development, for the small areas of mountain watersheds in the Daqing River, the average spatial variation in rainfall is not obvious.
Based on the change in NDVI in the upper reaches of each watershed in 2018, the correlation between the NVDI and drainage density was analyzed, and the correlation equation was obtained by SPSS as D d = −0.198x NDVI + 1.227 (R = 0.738, N =9, p = 0.02). This indicated that gullies developed in places with poor vegetation cover but not in places with good vegetation cover. Of course, it is difficult to determine the causal relationship between these factors because there may be a positive feedback mechanism; that is, the development of gullies in places with poor vegetation is conducive to the development of gullies, while the development of gullies, strong erosion and soil erosion inhibit the growth of vegetation [8].

Relationship between D d and Soil Type
Basin lithology affects the extent of landscape dissection. Some variation in D d by lithology type was also observed; the average D d for shale and schist was well above the observed mean D d , whereas the values for limestone and acidic volcanic rocks were well below the mean D d [48]. The national soil database was used to obtain the classification of surface substances in the Daqing River (according to the classification standard of FAO90). The extracted river networks in the study area were superimposed on the soil type, and the distribution of gullies on each soil type was analyzed. The drainage density of each soil type was calculated ( Figure 6). There are seven types of surface materials in mountain watersheds of the Daqing River, namely, cambisols (CM), luvisols (LV), regosols (RG), fluvisols (FL), leptosols (LP), anthrosols (Atc) and gleysols (Glm). The proportional distribution area of each soil type and the percentages of sand, silt and clay are shown in Figure 6. The main soil types in the mountain watersheds of the Daqing River were CM, LV and RG, with a distribution area of 91%. These three soil types all had a high content of sand gradation, approximately 50%, followed by silt. Clay made up the lowest proportion of the soil content, at approximately 20%.  Figure 6. The main soil types in the mountain watersheds of the Daqing River were CM, LV and RG, with a distribution area of 91%. These three soil types all had a high content of sand gradation, approximately 50%, followed by silt. Clay made up the lowest proportion of the soil content, at approximately 20%. Dd was the lowest in Glm (1.15 km/km 2 ) and the highest in FL (2.54 km/km 2 ). For the other soil types, Dd was between 2 and 2.5 km/km 2 . If soil particles are coarse and the permeability of the soil is high, the soil corrosion factor K value is low [49]. Melton [50] also found that Dd decreased with soil infiltration capacity. Figure 6 shows that the coarse gradation content of FL was the highest, at 85.5%. Thus, we can explain why the Dd of FL was the largest. The ability to resist soil erosion influences Dd. In addition, although the sand content of Atc was low, the Dd as relatively high, at 2.44 km/km 2 . It can be concluded that soil erosion is affected not only by soil properties but also by anthropic factors.

The Drainage Density on Different Slopes
The drainage density on different slopes is of particular interest as it helps to understand the relationship between erosion rates and patterns of channelization, which are critical for testing eco-geomorphic landscape evolution models [38,51].
Except for the maximum slope of the Juma River, which was 47°, the slopes of all other river basins were below 42°. We divided the slope into seven grades: 0-6°, 6-12°, 12-18°, 18-24°, 24-30°, 30-36° and over 36°. We calculated the distributions of gullies on different slope grades (Figure 7). The Dd in each study area decreased with increasing slope; there were no gullies with slopes higher than 36° in the Ci River, Jie River, Cao River, Zhongyi River or Beiyi River. In addition, the slopes of the Juma River and Dashi River mainly ranged from 6-18°, and the slopes in the other research areas were mainly distributed in the range from 0-12°, accounting for 62-81% of the whole mountain area. All gullies were concentrated between 0° and 12°, and the density value was above 2 km/km 2 . When the slope was greater than 24°, the value of Dd decreased to 0.8 km/km 2 and below. The maximum Dd of all rivers was between 0 and 6°, but the value of maximum Dd was different. Dd was highest in the Dashi River (4.1 km/km 2 ), followed by the D d was the lowest in Glm (1.15 km/km 2 ) and the highest in FL (2.54 km/km 2 ). For the other soil types, D d was between 2 and 2.5 km/km 2 . If soil particles are coarse and the permeability of the soil is high, the soil corrosion factor K value is low [49]. Melton [50] also found that D d decreased with soil infiltration capacity. Figure 6 shows that the coarse gradation content of FL was the highest, at 85.5%. Thus, we can explain why the D d of FL was the largest. The ability to resist soil erosion influences D d . In addition, although the sand content of Atc was low, the D d as relatively high, at 2.44 km/km 2 . It can be concluded that soil erosion is affected not only by soil properties but also by anthropic factors. The drainage density on different slopes is of particular interest as it helps to understand the relationship between erosion rates and patterns of channelization, which are critical for testing eco-geomorphic landscape evolution models [38,51].
Except for the maximum slope of the Juma River, which was 47 • , the slopes of all other river basins were below 42 • . We divided the slope into seven grades: 0-6 • , 6-12 • , 12-18 • , 18-24 • , 24-30 • , 30-36 • and over 36 • . We calculated the distributions of gullies on different slope grades (Figure 7). The D d in each study area decreased with increasing slope; there were no gullies with slopes higher than 36 • in the Ci River, Jie River, Cao River, Zhongyi River or Beiyi River. In addition, the slopes of the Juma River and Dashi River mainly ranged from 6-18 • , and the slopes in the other research areas were mainly distributed in the range from 0-12 • , accounting for 62-81% of the whole mountain area. All gullies were concentrated between 0 • and 12 • , and the density value was above 2 km/km 2 . When the slope was greater than 24 • , the value of D d decreased to 0.8 km/km 2 and below. The maximum D d of all rivers was between 0 and 6 • , but the value of maximum D d was different. D d was highest in the Dashi River (4.1 km/km 2 ), followed by the Juma River (3.6 km/km 2 ), and the rest of the rivers were all approximately 3.0 km/km 2 . The D d values were high in the north and low in the south.

The Drainage Density in Different Aspect of Slopes
Aspect is an important terrain factor. Some studies have shown that the differences of rainstorms, rainfall erosivity, soil moisture and vegetation growth conditions make the soil erosion pattern and intensity in different aspects have obvious asymmetry [52]. For revealing the impacts of aspects on drainage development in the Daqing River, we divided the slope aspect into north (N), northeast (NE), east (E), southeast (SE), south (S), southwest (SW), west (W) and northwest (NW). There were differences in Dd on each aspect. The maximum Dd was 2.45 km/km 2 , and the minimum was 1.63 km/km 2 ( Figure  8). SPSS was used to compare and test the average Dd of each aspect, and they had values of 2.16 km/km 2 (N), 2.08 km/km 2 (NE), 2.03 km/km 2 (E), 2.02 km/km 2 (SE), 2.10 km/km 2 (S), 2.20 km/km 2 (SW), 2.16 km/km 2 (W) and 2.07 km/km 2 (NW). In the formation process of the gullies, there was a directivity of upstream erosion, subdivision of gullies and development of branches. Relatively speaking, the gullies in the southwest, north and west developed considerably, while gullies in the east and southeast did not develop much. The measurement coefficient of the linear correlation between drainage density and aspect was 0.44, p = 0.042 < 0.05, which passed the significance test. This indicated that aspect influenced drainage density, but the two did not have a simply linear relationship. This is because the difference in aspect results in differences in other influencing factors, such as rainfall and solar radiation, and the change in vegetation growth caused by different solar radiation leads to different development and evolution of erosion gullies [52].

The Drainage Density in Different Aspect of Slopes
Aspect is an important terrain factor. Some studies have shown that the differences of rainstorms, rainfall erosivity, soil moisture and vegetation growth conditions make the soil erosion pattern and intensity in different aspects have obvious asymmetry [52]. For revealing the impacts of aspects on drainage development in the Daqing River, we divided the slope aspect into north (N), northeast (NE), east (E), southeast (SE), south (S), southwest (SW), west (W) and northwest (NW). There were differences in D d on each aspect. The maximum D d was 2.45 km/km 2 , and the minimum was 1.63 km/km 2 (Figure 8). SPSS was used to compare and test the average D d of each aspect, and they had values of 2.16 km/km 2 (N), 2.08 km/km 2 (NE), 2.03 km/km 2 (E), 2.02 km/km 2 (SE), 2.10 km/km 2 (S), 2.20 km/km 2 (SW), 2.16 km/km 2 (W) and 2.07 km/km 2 (NW). In the formation process of the gullies, there was a directivity of upstream erosion, subdivision of gullies and development of branches. Relatively speaking, the gullies in the southwest, north and west developed considerably, while gullies in the east and southeast did not develop much. The measurement coefficient of the linear correlation between drainage density and aspect was 0.44, p = 0.042 < 0.05, which passed the significance test. This indicated that aspect influenced drainage density, but the two did not have a simply linear relationship. This is because the difference in aspect results in differences in other influencing factors, such as rainfall and solar radiation, and the change in vegetation growth caused by different solar radiation leads to different development and evolution of erosion gullies [52].

Indicative Significance of Dd
As shown in Figure 9, the HI values in the mountain basins of the Daqing River were all below 0.4. The correlation between HI and Dd was significant (R = 0.87, p = 0.015), but these factors were not simply linearly related. With the increase in HI, Dd first decreased and then increased. With 0.3 as the boundary, the closer HI was to 0.3, the smaller Dd was, and conversely, the larger Dd is. It was concluded that the landform of the mountain basin of the Daqing River is in an advanced stage of erosion development and that Dd will no longer increase [35].

Effect of Channel Morphology on Water Yield
The measured runoff data of the Ci River, Sha River, Tang River, Cao River, Zhongyi River, Juma River and Dashi River were analyzed. Based on SPSS, the relationship between A and the water yield modulus (Sr, 10 4 t/km 2 •a) was Sr = 5962.3/A + 7.3 (R = 0.421, N = 7, p = 0.346). The relationship between Re and Sr was Sr = −528.6Re 3 + 457.5Re − 140.7 (R = 0.714, N = 7, p = 0.202). According to the above two equations, although the correlation coefficient between Sr and A and Re was not low, neither of them passed the significance level of 0.05.  Figure 8. Distribution of drainage densities in different aspects of mountain watersheds in the Daqing River (referring to Figure 2 for the abbreviated names of rivers).

Indicative Significance of D d
As shown in Figure 9, the HI values in the mountain basins of the Daqing River were all below 0.4. The correlation between HI and D d was significant (R = 0.87, p = 0.015), but these factors were not simply linearly related. With the increase in HI, D d first decreased and then increased. With 0.3 as the boundary, the closer HI was to 0.3, the smaller D d was, and conversely, the larger D d is. It was concluded that the landform of the mountain basin of the Daqing River is in an advanced stage of erosion development and that D d will no longer increase [35].

Indicative Significance of Dd
As shown in Figure 9, the HI values in the mountain basins of the Daqing River were all below 0.4. The correlation between HI and Dd was significant (R = 0.87, p = 0.015), but these factors were not simply linearly related. With the increase in HI, Dd first decreased and then increased. With 0.3 as the boundary, the closer HI was to 0.3, the smaller Dd was, and conversely, the larger Dd is. It was concluded that the landform of the mountain basin of the Daqing River is in an advanced stage of erosion development and that Dd will no longer increase [35].

Effect of Channel Morphology on Water Yield
The measured runoff data of the Ci River, Sha River, Tang River, Cao River, Zhongyi River, Juma River and Dashi River were analyzed. Based on SPSS, the relationship between A and the water yield modulus (Sr, 10 4 t/km 2 •a) was Sr = 5962.3/A + 7.3 (R = 0.421, N = 7, p = 0.346). The relationship between Re and Sr was Sr = −528.6Re 3 + 457.5Re − 140.7 (R = 0.714, N = 7, p = 0.202). According to the above two equations, although the correlation coefficient between Sr and A and Re was not low, neither of them passed the significance level of 0.05.

Effect of Channel Morphology on Water Yield
The measured runoff data of the Ci River, Sha River, Tang River, Cao River, Zhongyi River, Juma River and Dashi River were analyzed. Based on SPSS, the relationship between A and the water yield modulus (S r , 10 4 t/km 2 •a) was S r = 5962.3/A + 7.3 (R = 0.421, N = 7, p = 0.346). The relationship between R e and S r was S r = −528.6R e 3 + 457.5R e − 140.7 (R = 0.714, N = 7, p = 0.202). According to the above two equations, although the correlation coefficient between S r and A and R e was not low, neither of them passed the significance level of 0.05.
The relationship between the slope of the channel (S, %) and S r was S r = e 3.6−24.13/S (R = 0.78, N = 7, p = 0.039). There was a significant positive correlation between these factors, that is, the greater the slope was, the greater the water yield.
Drainage density is not only an important index of watershed erosion but also an important factor affecting catchment confluence and sediment transport in gullies. The relationship between D d and S r was S r = 21.5 D d − 33.27 (R = 0.214, N = 7, p = 0.645). The water yield modulus was proportional to the drainage density; that is, the water yield modulus increased as the drainage density increased. Melton [50] also observed that D d increased with increasing percentage of bare ground and runoff, but in this region, the two were not significantly correlated.
In addition, the impacts of other factors, such as rainfall, relief, slope gradient, etc., may produce large deviations in the constructed relations between specific water yield and indices of basin geometry, so the small number of available data samples do not prove the influence of basin geometry on water yield.

Conclusions
Based on the DEM, which had a resolution of 30 m, we extracted the river network of nine mountain watersheds in the Daqing River. The shape of the mountain watersheds in the Daqing River was narrow. The stream orders ranged from five to seven. The average bifurcation ratio was 3.8-4.8. The drainage density was in the range from 2-2.3 km/km 2 , and there was low spatial variation.
Although rainfall is the driving force of gully development, for the small areas of mountain watersheds in the Daqing River, the average spatial variation of rainfall is not obvious. Rainfall was negatively correlated with drainage density, but the correlation between them was not significant at the 0.05 level. The correlation between NVDI and D d indicated that gullies developed in places with poor vegetation cover but not in places with good vegetation cover. D d decreased with increasing slope; it was highest in the Dashi River (4.1 km/km 2 ), followed by the Juma River (3.6 km/km 2 ), and the rest of the rivers had values of approximately 3.0 km/km 2 . D d values were high in the north and low in the south. The gullies in the southwest, north and west developed considerably, while gullies in the east and southeast did not develop much.
With 0.3 as the boundary, the closer HI was to 0.3, the smaller drainage density was, and conversely, the larger drainage density was. The landform of mountain basins in the Daqing River was in an advanced stage of erosion development, and the drainage density is no longer increasing. In addition, there was a significant positive correlation between the channel slope and water yield modulus, and the other watershed parameters were not significantly correlated with the yield modulus.