Spatiotemporal Variation Of Soil Erosion And Its Influence Factors In The Middle Reaches Of The Yellow River

Understanding the spatiotemporal variations of soil erosion in the Loess Plateau and its influence factors can supply suggestions to control soil erosion and improve ecological environment. Therefore, choosing the middle reaches of the Yellow River as study area, the specific contents of this paper were:(1) to detect the change trends of vegetation cover using Normalized Difference Vegetation Index(NDVI) data;(2) to analyze the change trends of soil erosion using Mann-Kendall trend test;(3) to understand the characteristic of soil erosion in different plant zone, topographic zone and soil erosion type region;(4) to quantify the contribution of environment factors on sediment yield using partial least squares regression(PLSR). The main results and conclusions of the present study are as follows:(1) The NDVI data from 1982 to 2013 was used to investigate the chang trend of vegetation cover in study area. The results showed that the average year NDVI value for the study area is 0.313 from 1982 to 2013. Summer has the highest NDVI value(0.306), followed by autumn(0.275). And winter has the lowest NDVI value(0.123). Compared to the west part of the study area, the south and east part has higher NDVI values. Summer has the fastest NDVI growth rate(0.0055), followed by autumn(0.0053), spring(0.0040), and winter(0.0004). The results of Mann-Kendall test showed that the significant changes of spring and autumn NDVI both occur in 2002; significant change of winter NDVI occurs in 2008; significant changes of summer, growing season and year NDVI occur in 1998-1999 and 2002. From 1982 to 2013, precipitation and temperature in the study area both show an increased trend. The result of partial correlation analysis showed that NDVI has a positive relationship with precipitation. And except winter, NDVI also has a positive relationship with temperature in other period. Temperature change had a lagging effect on vegetation growth, but precipitation change did no.(2) In 1955-1969, the average sediment yield of middle reaches of the Yellow River is 9949.7t/km2?a; compared with sediment yield of 1959-1969, the sediment yield in 1970-2007 has a significant reduction, and is 4710.5t/km2?a. Sediment yield of 1970-1989 and 1990-2007 is 6316.7t/km2?a and 2925.8t/km2?a, respectively.Sediment yield data from 1960-1989 of 8 hydrological stations(Pianguan, Huangfu, Hanjiamao, Hengshan, Qingyangcha, Zichang, Ganguyi and Jixian) was selected for Mann-Kendall test. The results of Mann-Kendall test showed the sediment yield of 8 hydrological stations has a decrease trend. Overall, the significant changes occurring in the 1970 s and late 1980 s.(3) To analyse the spatial distribution of sediment yield, the study area was divided into different regions based on plant, topographic and soil erosion type. Sediment yield data from 1955-1989 and 2006-2011 of 44 hydrological stations was used. The results showed that sediment yield has a downward trend in all plant regions, topographic regions and soil erosion type regions. From the view of plant regions, temperate forest steppe zone has higher sediment yield in different period. From the view of elevation regions, the sediment yield of region with elevation>1400m and <1200m is the lowest and highest sediment yield in all elevation regions in different period(during 1955-1969, 1970-1979, 1980-1989 and 2006-2011), respectively. From the view of soil erosion type region, forest loess hilly and gully region has the lowest sediment yield in 7 soil erosion type regions in different period, while, loess gentle slope hilly and gully region, loess Mao hilly and gully region and loess Liang hilly and gully region has higher sediment yield in different period.(4) The effects of watershed characteristics on sediment yield were analysed in 7 soil erosion type region using PLSR. A total of 28 watershed descriptors, including rainfall, runoff, soil variables, topographic variables, land use and land use pattern, were considered to be potential factors affecting sediment yield. The results showed that the first and second PLSR components for different soil erosion type region sediment yield are all dominated by rainfall, runoff, plan curvature, soil clay and sand content, the areal percentages of grass and forest, largest patch index, contagion, Shannon’s diversity index and Simpson’s diversity index. It can be seen from the variable influence on projection(VIP) that rainfall, runoff, the areal percentages of forest, perimeter-area fractal dimension and aggregation index has larger influences on sediment yield in all soil erosion type regions, compared with other watershed characteristics. And sediment yield increases with the increase of rainfall and runoff. The sediment yield has a negative relationship with the areal percentages of forest in wind and sand loess hilly and gully region, loess gentle slope hilly and gully region, loess Mao hilly and gully region and forest loess hilly and gully region. In wind and sand loess hilly and gully region, loess Liang hilly and gully region, loess Mao hilly and gully region and forest loess hilly and gully region, sediment yield decreases with the increase of perimeter-area fractal dimension. In wind and sand loess hilly and gully region, loess Liang hilly and gully region and loess Mao hilly and gully region, sediment yield has a positive relationship with aggregation indext.