Water And Sediment Yield Processes To Land Use Change In Duhe Watershed In Water Source Area Of The Middle Route Project Under South-to-North Water Diversion

Soil and water loss are major global environmental problems, associated with the deterioration of water quality, impaired river ecosystems, and flooding. Water and sediment yield in watersheds are controlled by many factors, e.g., climate, geomorphology, soils and land use. Among these factors, land use is the most direct anthropogenic factor that influences water and sediment changes. Quantifying the responses of water and sediment yield processes to land use change is critical for watershed management in the current context. To examine emerging water and soil loss and flood hazards in water source areas of the Middle Route Project under the South-to-North Water Transfer Scheme(SNWTS), this study focused on a detailed analysis of the Upper Duhe Watershed and its sub-watersheds, which are the source of water for the SNWTS. Spatial and property databases were established based on hydroclimatic data, underlying surface data, geographic information technologies, and reconnaissance field surveys. Soil and Water Assessment Tools(SWAT) was used to simulate water and sediment yield in un-gauged sub-watersheds, to choose logarithmic normal(LN2) distribution for non-stationary frequency analysis of extreme-streamflow, and to address the relative importance of watershed characteristics to water and sediment yield. The main results are as follows:(1) Spatio-temporal analysis of land use composition and patterns at the sub-watershed scale was based on land use maps and landscape indexes extracted and calculated from their spatial distribution in 1987, 1999, and 2007. Results indicated that in cultivation period(1990s) the forest diminishments were linked with the physical accessibility to the forest stand, such as linear distances to the highway, road, and navigable river, farmland augmented associated with cultivation mainly occurred in land near the stream channels in the whole watershed. During the ecological restoration period(2000s), specific farmlands(normally with slopes >25°) were transformed to forests. Compared with the baseline period(1980s), land use in the sub-watersheds varied more significantly in the cultivation period(1990s), and the distribution of land use changed in the sub-watersheds and became more irregular during the ecological restoration period(2000s). Land use changes led to landscape pattern changes.(2) Baseflow was separated using digital filtering method for two gauge stations(Zhushan and Xinzhou) in the watershed with streamflow records. The average monthly baseflows of Zhushan/Xinzhou stations were 19.0/17.9, 14.7/13.3, and 16.6/15.7 mm for the 1980 s, the 1990 s, and the 2000 s, and changed significantly for both the Zhushan and Xinzhou stations. The SWAT model was used to simulate the baseflow for the 107 un-gauged sub-watersheds, with satisfactory efficiency. The baseflow in the sub-watersheds varied more significantly in the 1990 s than in the 1980 s, and the most significant variation in baseflow was noted in the 2000 s.(3) The dynamics of baseflow are closely associated with changes of land use, the dominant negative factors for baseflow are forest, farmland, and urban land, whereas grassland, shrubland, and barren do not exert a significant negative influence on baseflow. Baseflow regression of forests is stronger and sustains for longer time than that of farmland, thereby baseflow will increase with the replacement of forest by farmland. The increasement of non-contributing impervious areas mitigates the negative influences of urbanization on baseflow and thereby leads to the baseflow increase when forest changes to urban land. The relatively important of farmland and urban land are disparity in different seasons. The negative impact on baseflow of farmland is greater than that of urban land in summer, whereas lower in other seasons.(4) This study applied an integrated approach of the SWAT model and non-stationary frequency analysis(FA) to relate the quantile of extreme streamflow and the recurrence frequency of extreme streamflow events in the Upper Duhe Watershed; the frequency analysis of extreme streamflow events was performed using an LN2 distribution and maximum likelihood estimation. The results indicated that quantiles of streamflow for the different return periods varied significantly. Quantiles of streamflow decreased from the southwest to the northeast for the 2 years return period; however, higher quantiles were noted in the northwest and the southeast for the 20 years, 100 years, and 1000 years return periods. The extreme streamflow events of each decade exhibited similar dynamics for lower or higher return level ranges(RLR), and significantly different patterns existed between the lower and higher RLR. The maximum flood risk occurred in the 1980 s, whereas the minimum occurred in the 2000 s. Land use is the pivotal factor influencing streamflow risk for shorter return periods, and maximum streamflow dynamics for longer return periods can be attributed to the hydrological regulation of water reservoirs. The frequencies of extreme streamflow events are primarily determined by forest, urban, and farmland land uses. Farmland and urban uses increase streamflow risk, whereas forest decreases risk.(5) We investigated the combined effects of watershed complexity to examine how the land use and physiography are related to the specific sediment yield using hydrological modeling and partial least-squares(PLS) regression in the Upper Duhe watershed. The land use composition, land use pattern, morphometric variables, and soil properties of the watershed were considered as potential factors of influence. The results revealed that the specific sediment yields are closely associated with land cover patterns. The land use composition and land use pattern could account for as much as 65.2% of the variation in specific sediment yield. A set of physiographic indices was also considered to have a major effect on specific sediment yield and explained 17.7% of the variation in the specific sediment yield. The areal percentages of agriculture and forest, the patch density, the shannon‘s diversity index, the soil erodibility, contagion, the hypsometric integral, and the saturated soil hydraulic conductivity are identified as the primary metrics. The watershed size is highly correlated with the sediment delivery ratio(SDR). Shannon‘s diversity index, contagion, and the hypsometric integral can be considered important factors in the sediment delivery ratio.