Spatio-temporal characteristics of soil bulk density and saturated hydraulic conductivity at small watershed scale on Loess Plateau

Surface soil bulk density (BD) and saturated hydraulic conductivity are two of important parameters to assess rainfall infiltration, runoff formation and soil erosion. But little research has been done at a watershed scale. On the Loess Plateau of China, investigations on the spatio-temporal changes of BD and saturated hydraulic conductivity at a small watershed scale are essential for regional ecological restoration and may enrich our understanding on the related eco-hydrolgical processes. In this study, we measured BD and saturated hydraulic conductivity in August, September, and October of 2014 at 73 sites that were pre-located by using Global Positioning System in a 50 m × 50 m grid at the LaoYeManQu watershed on the wind-water erosion crisscross region of the Chinese Loess Plateau. Using classical statistical and geostatistical methods, we characterized and compared the spatial and temporal changes of BD and saturated hydraulic conductivity. The semivariograms and kriging method included in the geostatistical method were used to reflect the visual variation among different months from August to October. Results showed that: 1) Among the three months, BD had a normal distribution and exhibited a time stability while saturated hydraulic conductivity had a skewed distribution and varied greatly. BD variation was smaller than that of the saturated hydraulic conductivity at the watershed scale. 2) Semivariograms of BD and saturated hydraulic conductivity from August to October could be best fitted by an isotropic exponential model. Both BD and saturated hydraulic conductivity exhibited a moderate spatial dependence. Distribution maps produced by kriging indicated pronounced spatial patterns of BD and saturated hydraulic conductivity at the small watershed-at the slope top and the western part of the watershed, BD demonstrated an increasing trend while saturated hydraulic conductivity decreased gradually from August to October. 3) Pearson correlation analysis showed that correlations between BD and saturated hydraulic conductivity at each measured time from August to October were weak (r<0.4); while, for the same soil type (dry sandy entisol) in the watershed, BD was correlated with saturated hydraulic conductivity negatively at the significant level of 0.01 and for aeolian soils the correlation was not significant. Our results suggested that different soil types in the watershed may change the overall correlation between BD and saturated hydraulic conductivity of the three months. Understanding the spatial and temporal variation of BD and saturated hydraulic conductivity would increase the accuracy in modeling and predicting rainfall infiltration, soil water and solutes movement at the watershed scale.