| Soil moisture is an important research content of soil physics, and is also the fundament of sustainable land use, water resources planning&management and water-saving agricultural technology. Monitoring and simulation of soil water has become a hot cutting-edge research area internationally. Many organizations and countries have attached great importance to this area, especially in arid and semi-arid regions. Variation in soil moisture is of critical importance in many scientific and operational activities, such as groundwater recharge, climate studies, and numerical weather prediction. Understanding soil moisture variability is also essential for quantifying relationships among hydrology, ecology and physiography in a given region. Therefore, the study of the spatio-temporal variability of soil moisture has been a hot spot in the fields of hydrology and soil science. Spatial patterns and temporal variability of soil moisture are influenced by different environmental factors at different scales. Characterizing its relationship to environmental factors at different scale is of critical importance for soil moisture prediction, environmental quality evaluation and other operational activities.As the source of water for the Middle Route Project of South-to-North Water Diversion, the Danjiangkou reservoir is also one of the important strategic water resources areas in China. Considering the serious soil-erosion problem and the vegetation degradation, studying the relationship between spatio-temporal variability of soil moisture and environmental factors in the typical small catchment is of great importance for the ecological security and ecological capacity of the Danjiangkou reservoir area. The objectives of this study is:(1) to investigate the spatial variability of soil moisture and its relationship with environmental factors;(2) to partition the variance of the soil moisture data in different wet-dry periods to the contributions of two subsets of environmental variables; and (3) to analyze the relationship of these main factors with soil moisture in different wet-dry periods. The main results are as followed:(1) ANOVA was applied to analyze the differences of soil moisture during each measuremt process. For all the12measurement processes, mean soil moisture shown significant difference (p<0.01) by step of two days during one measurement process. The spatial variability of soil moisture and its relation with mean soil moisture were analyzed in the Wulongchi catchment. The spatial variability of soil moisture increased with decreasing soil moisture for all of the12measurement processes. In the latter part of the measurement process1,3,4and5, the spatial variability gradually stabilized. But for measurement process9and10, their soil moisture gradually declined, but its spatial variability fluctuated.(2) These12measurement processes were classified into three different wet-dry periods, such as humid, moderate and dry period. Redundancy analysis was applied to determine the first two main factors on each day of the12measurement processes. The effects of environmental factors on soil moisture vary along with changes in soil moisture. Although the two main factors may be same on some of the measurement days, they were different more or less at most cases. On the whole, slope position, the topographic wetness index, soil thickness and slope shape are the main factors affecting soil moisture variability during the humid period. For the moderate periods, soil thickness and slope position and slope shape are the main factors regulating soil moisture variability. The main factors control soil moisture variability during dry periods are slope position, the topographic wetness index and soil thickness. During dry period, the main factors of the6days in each measurement period were still different from each other, but for all that they were relatively stable than that of the humid and moderate period.(3) The spatial variability of the three different periods was analyzed. Spatial variability of the humid and moderate period increased with decreasing soil moisture. But in the dry period, the soil moisture decreased during this period, however the spatial variability first increased and then decreased. The redundant environmental factors of these three different wet-dry periods were excluded using forward selection and Monte Carlo test. And then we partitioned the variance of the soil moisture in different wet-dry periods into the contributions of two subsets of environmental variables. After removing the redundant variables, the main factors for the humid and moderate period were soil thickness and slope position, but for the dry period, soil thickness, slope position and topographic wetness index were the main factors. Dyring the three wet-dry periods, soil thickness was the most important factors respectively. The relative contribution without the shared portion of soil thickness for the variability of soil moisture in different wet-dry periods were39.89%,35.37%and33.44%, respectively. The relative contribution without the shared portion of slope position for the three different wet-dry periods were21.48%,25.00%and9.15%, respectively. The relative contribution without the shared portion of topographic wetness indexin the dry period was only3.49%。(4) Based on the three different wet-dry periods, the effect of main factors at different level on soil moisture were analyzed. The result indicated that the lag effect of main factors’ role were different due to the different initial soil moisture. The higher the initial soil moisture, the more obvious lag effect. During the humid period, lag effect of soil thickness was evident, but slope position did not have the lag effect. During the moderate period, the lag effect of soil thickness and slope position were both evident. And during dry period, all of soil thickness, slope position and topographic wetness index had no such effect.(5) Soil moisture of different wet-dry periods were predicted using exponential smoothing method. The results indicated that the prediction accuracy increase sharply from dry period, moderate period to humid period. For the humid period, the value of Nash-Sutcliffe efficiency for the prediction was0.90, but for the moderate it was0.26, and for the dry period, it was-0.61. |
