| The runoff and sediment of the Yellow River have been decreasing in recent decades.How to evaluate the impacts of the climate change and human activity on the reduction ofrunoff and sediment is not only a hot topic currently, but also a research challenge confrontedby hydrologist and environment science. Therefore, the Yanhe Watershed (7591km2), atributary to the middle reaches of the Yellow River, was selected as research area in this study.Based on the observed data of major meteorological stations and hydrological data of the lastcontrol station (Ganguyi) from1952-2008, the Mann-Kendall test, Sen slope method andother statistical methods were employed to detect the long-term trends of hydro-climaticvariables in the Yanhe Watershed for the last past57years, and the impacts of climaticvariable change and human activity on runoff and sediment were evaluated. This paper alsoanalyzed the land use structure and its dynamic changes of Yanhe Watershed during1975-2000. Moreover, the Soil and Water Assessment Tool (SWAT) was applied to thedifferent areas in the Yanhe Watershed to investigate the effect of DEM resolution andwatershed subdivision on hydrological simulation, and explore the temporal and spatialresponse of runoff and sediment to the land use dynamic changes. The main conclusions wereas follows:(1) The annual mean temperature, annual mean maximum and minimum temperatureincreased significantly during1952-2008at the rate of0.038℃/a,0.017℃/a and0.038℃/a,respectively. The most evident warming occurred in winter, and its mean temperature, meanmaximum and minimum temperature exhibited significant increasing trends at the rate of0.053℃/a,0.029℃/a and0.070℃/a. The mean temperature, mean maximum and minimumtemperature in spring had significant increasing trends. Besides, the significant increasingtrends of mean temperature and minimum temperature were observed in spring and autumn,but the mean maximum temperature for annual, spring and autumn showed a slightlyincreasing trends with no statistical significance. The annual and seasonal mean temperatureand minimum temperature showed increasing trends with statistical significance (α=0.01).While only the significant increasing trend were observed in annual, spring and winter time series for maximum temperature, and the increasing magnitudes were lower thancorresponding mean temperature and minimum temperature.(2) The normal precipitation year accounted for80.7%during1952-2008, and mostyears were lower and low flow years in the57years. Annual precipitation exhibited adecreasing trend in the study period at the rate of-0.64mm/a, with no statistical significance;while the annual streamflow had a significant downward trend at the rate of-109.90×104m3/a.The abrupt changes of annual precipitation and streamflow occurred in1995and2005,respectively. The seasonal precipitation of spring, summer and autumn had weak decreasingtrends during1952-2008, but a significant increasing trend was found in winter season at therate of0.18mm/a. In four seasons, a declining trend for streamflow was identified in spring,but the change was not significant; the downward trends were significant in summer andautumn at the rate of83.83×104m3/a and26.16×104m3/a, respectively; while the significantincreasing trend was detected in winter with the magnitude of15.62×104m3/a. Comparedwith the baseline period of1952-1994, the reduction of precipitation and streamflow in theperiod1995-2008was11.1%and27.3%, respectively; precipitation change contributed46.2%to the reduction of streamflow, while human activities resulted in53.8%of streamflowreduction.(3) The precipitation, water discharge and sediment load in the flood season (from Juneto September) accounted for71.8%,64.1%and98.6%of the total annual amount,respectively. The precipitation exhibited a downward trend during the flood season, and theabrupt change occurred in1990, especially the precipitation frequency and events of25-50mm exhibited significant downward trends in past57years. The water discharge and sedimentload in the flood season exhibited significant decreasing trends at the rate of-99.2×104m3/aand-51.8×104t/a, respectively, and the abrupt change occurred around2005. Taking theperiod of1952-1994as the reference period, the reductions of the flood-season precipitation,water discharge and sediment load during1995-2008were11.4%,35.1%and50.2%,respectively. The decrease in precipitation was responsible for43.4%and35.9%of thedecrease in the water discharge and sediment load from1995-2008, respectively. While thecontribution of human activity to the decreasing trends in the water discharge and sedimentload were56.6%and64.1%, respectively.(4) The sensitive parameters of the Soil and Water Assessment Tool (SWAT) to simulaterunoff and sediment were calibrated, and the model performance of SWAT on simulating dailyrunoff and sediment in the Zhifanggou watershed was assessed. The result showed that theevaluation statistics for the runoff and sediment simulation showed that the model performedwell; the values of r2ranged from0.79to0.82, the values for NSE ranged from0.53to0.62, and the values for PBIAS ranged from-0.63%to6.50%for the calibration and validationperiods. However, the calibrated model tended to underpredict runoff and sediment duringhigh-flow season and overestimate the flow in the dry season. In addition, the SWAT couldsimulate annual runoff and sediment satisfactorily in the Xingzihe and Yanhe watershed.(5) The digital filter method was used to separate base flow from the measuredstreamflow data at the Zhao'an hydrological station from1958to1974, and the impacts ofDEM resolution and watershed subdivision on the hydrological simulation were investigated.The results showed that the DEMs with fine resolution (i.e.,20to150m) were required toachieve satisfactory accuracy (R2>0.93, NSE>0.51,-9.06%<PBIAS<24.20%) in simulatingannual streamflow, surface runoff, base flow and sediment yield. For DEM resolutions greaterthan150m, the accuracy of the simulation varied among the hydrologic elements. Thethreshold values of the watershed subdivision had little effect on streamflow, surface runoffand base flow. However, the threshold values had a large effect on sediment yield. Thereasonable watershed subdivision thresholds for sediment yield simulation ranged from12to100km2. Threshold values less than12km2or greater than100km2led to theunderestimation of sediment yield. These findings indicated that choices regarding the DEMresolution and threshold value of the watershed subdivision should be based on thehydrological element of interest so as to have higher simulation accuracy and runningefficiency.(6) The land use dynamic characteristics of Yanhe Watershed were investigated by usingthe land use data in1975,1990and2000. The result showed that the grassland wasdecreasing because some part of grassland converted into cultivated land and woodland. Inaddition, SWAT model was applied to the Yanhe Watershed to study the effect of land usedynamic changes on the runoff and sediment. Compared with land use data in1975, thesimulated streamflow in1990and2000were increased by1.84%and5.67%, and thesimulated sediment load were increased by3.36%and8.01%, respectively. The hydrologicalresponses of HRU to the different land use types were also analyzed. Due to the increase ofcultivated land and woodland, the total surface runoff, streamflow and sediment load wereincreased with the land use dynamic changes, especially, the cultivated land contributed thelargest amount of sediment among different land use types. Moreover, it was found that thesevere soil erosion which was classified by the Standard of Soil Erosion Intensity wasincreasing with the land use dynamic changes.
|
PDF Full Text Request