Explicating Sediment Sources of the Catchment Upstream of the Miyun Reservoir in Beijing, China


 As the only water drinking resource in Beijing, the Miyun Reservoir is still suffered over ten thousand tons of sediment input from its upper catchment. Explicating sediment sources of the catchment upstream of the reservoir is urgently required to further implement soil conservation measures. In this paper, the Revised Universal Soil Loss Equation (RUSLE) and Sediment Delivery (SEDD) models were combined to explicate the major sediment source of the catchment through exploring the spatial distributions of soil erosion and sediment delivery as well as their relations with land use and topography, and sediment source areas were then identified. The catchment average soil erosion intensity (SEI) of 4.08 t ha− 1 yr− 1 was two times the soil loss tolerance (T = 2.00 t ha− 1 yr− 1) of the study region. The values of cell sediment delivery ratio (SDR) showed a network distribution pattern, ranging from zero to unit, with an average of 1.65%. Cell specific sediment yield (SSY) presented a similar spatial pattern to SDR, ranging from 0 to 902 t ha− 1 yr− 1, with an average of 0.04 t ha− 1 yr− 1. Bare land suffered the highest SEI of 39.01 t ha− 1 yr− 1, followed by shrub land and orchard field. Nearly 70% of the sediment came from grass land. Farmland was the second sediment contributor. Grass land and farmland are the two major sediment source areas. Soil conservation practices should be further implemented on these lands, especially on the 3–5°slopes with elevations less than 500 m a.s.l.

implemented since the 1980s in the catchment above the reservoir (Li and Li 2008). 47 However, in recent years, mean annual sediment input from the catchment upstream 48 to the reservoir still reaches ten thousand tons. Therefore, it is urgently required to 49 explicate the sediment source.  In the current study, the RUSLE and SEDD models were applied to explicate the

84
Study area description 85 The catchment upstream of the Miyun Reservoir is located in the northeast of Beijing 86 city. Chaohe River and Baihe River are the two major tributaries that drain into the 87 reservoir (Fig. 1) The main land use types are forest and grass, occupying 63.3% and 22.3% of the 98 total area respectively (Fig. 1). Forest, shrub, and grass are mainly distributed in 99 high-elevation and steep areas. Farmland occupies around 10%, and is mainly 100 distributed in gentle slope areas along the rivers (Fig. 2).  In October and December 2019, two field campaigns were carried out. In October, 146 runoff plots in Shixia subcatchment were investigated (Fig. 1b). The runoff plots were 147 10-m long with different slope gradients and land use types, some of which have been 148 implemented with soil conservation measures (Table 2). After each rainfall, sediment  In the current study, the RUSLE modeling approach was used to estimate water 158 erosion, which is calculated as the product of six factors: 159 E = R * K * LS * C * P (1) 160 where E is the soil erosion intensity (SEI; t ha -1 yr -1 ), R is the rainfall-runoff erosivity 161 factor (MJ mm ha -1 h -1 ), K is the soil erodibility factor (t ha h ha -1 MJ -1 mm -1 ), LS is a 162 combination of slope length L and slope gradient S factor (-), C is the crop/cover 163 management factor (-), and P is the soil conservation factor (-).

164
R factor 165 The method to estimate R factor has been significantly improved over that in the (2)

172
Where Ri is the half-month R-factor (MJ mm ha -1 h -1 yr -1 ), and Dj is the erosive 173 rainfall day j. Dj is equal to actual rainfall when it is greater than 12 mm. Otherwise it in the study catchment (Fig. 4a).  where SAN is the sand content (%), SIL is the silt content (%), CLA is the clay content 187 (%), SC is the soil organic carbon content (%) and SN = 1-SAN/100. A soil erodibility 188 map was thus obtained, with K-factor values ranging from 0 to 0.064 t ha h MJ -1 mm -1 189 ha -1 in the study catchment (Fig. 4b).
190 191 The L and S factors represent the effect of topography on water erosion. A 2D 192 approach was used to calculate L factor using the method proposed by Desmet and 193 Govers (1996 where Li,j is the slope length factor of the grid cell with coordinates i and j, Ai,j is the and lower values appeared along the valleys (Fig. 4c).

205
C factor 206 The C factor represents the impact of cropping and management practices on water 207 erosion. In the study area, C-factor values for the land use types were obtained from  (Table 3). The lowest C value of 0.001 was assigned to the 212 forest land (Fig. 4d).
213 Table 3 is about here 214 P factor 215 Because extensive terrace and fit-scale practices have been implemented on farmland 216 and in orchard field, an average value of P factor was set to 0.01 for the farmland, and 217 a value 0.69 was given to the orchard field. The P values of other land use types were 218 set to one (Table 3). The P map was illustrated in Fig. 4e.

SEDD description and calibration 220
The SEDD model was used to estimate cell SDRi: where β is a catchment-specific parameter, and ti is travel time of the eroded  also occurred at Baihebao hydrological station due to reservoir interception (Fig. 1).

246
Therefore, only the SYs from other six hydrological stations in Table 1 (Table 1).

252
The calibration procedure was similar to the methods by  (Table 5).
260 Table 5 is about here 261 The spatial distributed cell SDRi of the catchments was obtained using Eq.(9), 262 and then cell specific sediment yield (SSYi) was calculated based on the RUSLE 263 model: where Ei is the SEI , and the SY of a given catchment was obtained by multiplying the 266 mean SSYi of the catchment and its area (km 2 ). were extracted from the DEM data for the six subcatchments (Table 1). These indexes These indexes were then used to estimate a β value for the study catchment (Table   278   5).

Sediment trap efficiencies of the reservoirs 280
There are multiple methods to estimate a reservoir's sediment trap efficiency (STE).

281
In the study catchment above the Miyiun Reservoir, there are two large reservoirs 282 named Yunzhou and Baihebao which are located on the Baihe River (Fig. 1). In which are located at the outlets of the reservoirs, implying that their STEs were 100%.

285
There are also many smaller reservoirs or ponds in the catchment (Li, 2007), however, 286 their STEs were not obtained due to unavailable data in the current study.

288
Soil erosion 289 The mean annual SEI of the catchment upstream of the Miyun Reservoir ranged from 290 0 to 902 t ha -1 yr -1 , with an average of 4.08 t ha -1 yr -1 (Fig. 5). It was over two times 9.27%, and 6.3% respectively. The area percentage with SEI higher than 50 t ha -1 yr -1 295 was less than 1.0%.

296
Soil loss widely occurred in the catchment, and varied greatly. Higher SEI 297 mainly appeared in the areas along the channels (Fig. 5). Considering the drainage 298 area (15331 km 2 ) of the catchment, annual total soil loss reached 6,255,048 tons.  Reservoir was obtained using Eq. (9) with the estimated β value of 1.73 (Fig. 6a).

319
The values of cell SDRi ranged from 0 to 100%, with an average of 1.65%, and a 320 standard deviation of 12.69%. Around 98% cells had an SDRi less than unit. Spatially,

321
the SDRi showed a network distribution pattern. The farther away from the river 322 networks, the smaller the SDRi values were.

323
The mean SSYi of the catchment ranged from zero to 902 t ha -1 yr -1 , with an 324 average of 0.04 t ha -1 yr -1 (Fig. 6b). The distance to streams greatly influenced SSYi.

325
The spatial distribution pattern of the cell SSYi was quite similar to that of the SDRi,  (Table 7). In the current study, the estimated SEIs were 1.66

344
In respect of SEI on farmland, the estimated SEI of 6.15 t ha -1 yr -1 is much higher 345 than that (i.e., 0.10 t ha -1 yr -1 ) on runoff plot. Similarly, the estimated SEIs on other 346 types of lands are also higher than the counterparts on the runoff plots ( Xiabao, and Sanyimiao) were underestimated, and verse visa (Fig. 7). However, the    (Fig. 8ab). However, a threshold value of SSY occurred on the 3-5 degree slopes.

386
On the gentle slopes, sediment does not easily enter the steams. However, sediment 387 flow on steeper slopes usually requires a long distance to reach streams, resulting in a 388 less SSY. Their interaction can thus produce a threshold value of SSY with increasing 389 slope gradients.

397
Noticeably, forest, grass, and shrub lands were mainly distributed in in the steep 398 areas with higher elevations (Fig. 2), where the rates of soil loss were lower on the 399 forest and shrub lands, while higher SEI on the grass land (Table 6). The bare land 400 which had the largest SEI is mainly distributed in the areas above 1000 m a.s.l. Soil 401 loss control for these two types of lands also should be done.

Uncertainty analyses 403
The β parameter is a major source of uncertainty in the SEDD model (Batista et al.  Chaohe River outlets, respectively. The deviation could be explained by at least two 422 aspects. In the study area, there were many dams or reservoirs, including the two 423 largest ones named Yunzhou and Baihebao reservoirs on the Baihe River (Fig. 1).

424
Much of the eroded sediment from catchment upstream is trapped by the reservoirs.  The study catchment has an area of 15, 331 km 2 with zigzagged channels in the 432 downstream areas (Fig. 1). As a result, more sediment is deposited on the flood plains   Note: All the studies in "reference" column were conducted in the NEC.

627
628 629        The mean slope gradients and elevations of different land use types distributed in the study catchment. Note:the numbers in the x-axis represent land use types in Table 7   Figure 3 Pictures showing (a) runoff plot with bare soil, (b) forest, (c) terraced farmland, and (d) tree tray in the orchard eld Figure 4 Spatial distributions of the values of RUSLE-R, -K, -LS, -C, and -P factors in the study catchment Note: The designations employed and the presentation of the material on this map do not imply the expression of any opinion whatsoever on the part of Research Square concerning the legal status of any country, territory, city or area or of its authorities, or concerning the delimitation of its frontiers or boundaries. This map has been provided by the authors. Spatial distribution of the estimated soil erosion intensity (SEI) in the study catchment Note: The designations employed and the presentation of the material on this map do not imply the expression of any opinion whatsoever on the part of Research Square concerning the legal status of any country, territory, city or area or of its authorities, or concerning the delimitation of its frontiers or boundaries. This map has been provided by the authors. Map showing the spatial distributions of the estimated SDR (a) and SSY (b). Note: The designations employed and the presentation of the material on this map do not imply the expression of any opinion whatsoever on the part of Research Square concerning the legal status of any country, territory, city or area or of its authorities, or concerning the delimitation of its frontiers or boundaries. This map has been provided by the authors. Comparison between the measured SSY with the simulated ones for the six subcatchments in Table 1 Figure 8 Soil erosion intensity (SEI) and SSY (ac), and RUSLE-LS factor and ow length of the cells to the nearest channels with increasing slope gradients and elevations (bd)