Abstract:
Abstract: The surface soil saturated hydraulic conductivity on farmland is one of the most important factors affecting water infiltration and distribution in soils and is also an important parameter in most soil water flow models. Previous studies have shown that saturated hydraulic conductivity is a highly spatial varied parameter under field conditions. Therefore, understanding and quantifying spatial variability at field scale is valuable to better simulate soil water movement dynamics through incorporating spatially-distributed saturated hydraulic conductivity into soil water flow models. This could help to evaluate impacts from different management practices and to develop precision irrigation management practices. The objectives of this study were to characterize spatial variability of the surface soil saturated hydraulic conductivity and explore its potential association with soil properties. The experiment was conducted on a 7.6 hm2 vineyard in an arid region of northwest China. Soil saturated hydraulic conductivity and other properties (clay, silt, sand, soil bulk density and organic matter) were measured for 0 - 10 cm soil of the geo-referenced points, which were located on a regular grid of 25 m × 25 m. At each sampling point, the soil saturated hydraulic conductivity was determined by the variable water level method. Spatial structure of spatial saturated hydraulic conductivity was described by a fitted variogram model based on a computed sample variogram, and possible spatial relationship between saturated hydraulic conductivity and other soil properties were evaluated through cross-correlograms. The regression kriging, based on step-wise linear regression of the saturated hydraulic conductivity with other soil properties, was used to predict spatial saturated hydraulic conductivity. I Its performance was compared to ordinary kriging and simple linear regression methods based on ME and RMSE computed from observed and predicted saturated hydraulic conductivity values. For this study, 70% of the measured data of the 135 sampled points were randomly selected to calibrate the models while the remaining 30% were used as a validation dataset, and the same calibration and validation datasets were used for the different methods. Main results from the study were: 1) according to descriptive statistics analysis, the soil saturated hydraulic conductivity showed strong spatial variability with mean of 1.64 cm/d and CV of 117% and the CV value was over 10 times larger than that of other soil properties; 2) the sample variogram was best fitted by an exponential variogram model, and the results showed that the correlation range was about 165 m which was comparable with results from other studies in fields with the similar size. The results also showed that about 60% of surface soil saturated hydraulic conductivity variability could be attributed to random variability from measurement error or sampling variability at distance shorter than our sampling distance; 3) the correlation analysis showed that soil saturated hydraulic conductivity was significantly correlated with silt, sand, clay and organic matter content and the correlation length was about 120 m while uncorrelated with soil bulk density; 4) among the four prediction methods, regression-kriging performed the best in the medium zone where saturated hydraulic conductivity was between the first and third quartiles of its values, and performed similarly with ordinary kriging at both lower and higher zones.