| Direct determination of soil hydraulic properties and soil properties is often costly, time-consuming and laborious. Accurate prediction of soil hydraulic properties and soil properties and understanding their three-dimensional map are important for studying water movement and contaminant transport in soil. The measurement of soil hygroscopic water content is easy, but it was seldom regarded as predicor for soil hydraulic properties and soil properties. Due to the effect of profile depths on vertical distribution of some soil properties, the stationary assumption for them cannot be met in the vertical direction. Therefore three-dimensional ordinary kriging cannot be directly used to map these soil properties in three dimensions. Fistly, this study established the relationship of soil hygroscopic water content with the dry-end SWRC, the wet-end SWRC, clay content and texture, respectively. Secondly, we used easily measured hygroscopic water content data to infer clay content and soil texture, then mapped them in three dimensions using three-dimensional ordinary kriging and transition probability geostatistics methods. Finally, combining three-dimensional ordinary kriging and profile depth function, we developed a simple method for mapping SOC in three dimensions. The main results and conclusions of this study are as follows:The Oswin model yielded satisfactory fits to dry-end SWRCs for soils dominated by both2:1and1:1clay minerals. Compared to the Oswin model, the CS-K model (Campbell and Shiozawa model combined with the Kevin equation) produced better fits to dry-end SWRCs of2:1clay-dominated soils, but provided poor fits for1:1clay-dominated soils. The shape parameter a of the Oswin model was dependent on clay mineral types, and approximate values of0.29and0.57were obtained for2:1and1:1clay-dominated soils, respectively. Comparison of the Oswin-KRH50model (Oswin model combined with the Kelvin equation, with water potential estimated from ORH50S-A model (CS model combined with the Arthur equation) and CS-KRH50model (CS-K model, with water potential obtained from ORH50indicated that on soils dominated by2:1clay minerals, the predictive ability of Oswin-KRH50model was compatible to the CS-KRH50model when ORH5o was the input parameter, but performed better than the CS-A model where clay content was the input parameter. The Oswin-KRH50model also has the potential for predicting dry-end SWRCs of1:1clay-dominated soils.In this study, we provided a simple approach for deriving the wet-end section of SWRC using the information of the dry-end section of the same SWRC and soil bulk density. A total of21soils with different textures (clay contents vary between0%and47%) were used for evaluating the method. Results showed that the root mean square errors and the mean errors generated by the proposed method were in the ranges of0.009to0.113g g-1and-0.098to0.002g g-1, respectively. These results were comparable with those from literature, suggesting that the proposed method may be used to predict the wet-end section of SWRC. The prediction performances were obviously improved when residual water content was set to a nonzero small value such as a value at log10(-Ψ)=5and for sandy soils the edge points were increased from-104. cm to-103.8cm. In general, the measurement of the dry-end section of SWRC is helpful to the prediction of the wet-end section. The new linear model was established by integrating with the Campbell and Shiozawa model and the Kelvin equation. A total of114soil samples were collected from14soil profiles (0to2m) near Tai’an city of China. Soil clay content and0at three different RHs (21,45, and60%) were measured and used to validate the New model, separately. The0values varied from0.23to4.86%, and fcL ranged from2.84to37.3%. The results showed that the values predicted by the New model agreed well with measured data. The further comparison using measured and literature data indicated that the root mean squared errors of the New model, the Resurreccion model, and the Schneider-Goss model were in the ranges of2.3to6.6%,3.1to8.8%and4.4to9.2%, respectively, with the New model having the least prediction error. The New model will provide a useful method to estimate clay content of soils dominated by2:1minerals using0at an arbitrary RH. Nevertheless, the New model may produce significant errors on coarse-textured soils with high OC contents and for soils with clay minerals dominated mostly by1:1types.The patial correlation ranges of soil clay content are55.0m and1.16m in the horizontal and vertical directions, respectively. The subsoil clay content is low in the northeast part of the study area while the soil layers with high clay content are mainly distributed in the upper section of soil profiles in the southern part of the study area. This reflects that the fluvial deposits in the study area have a fining-upward tendency. Cross-validation showed that the3D kriging method could effectively capture the spatial distribution characteristics of soil clay content in alluvial soils in the North China Plain.Soil texture had siginficant effect on hygroscopic water content. The ranges of hygroscopic water content for sand, sandy loam, loam and clay were<0.88%,0.88-1.37%,1.37-2.51%and>2.51%, respectively. The main textural type in the study area was loam, and sandy loam was the least textural type. The orders of continuity for the four textural types in horizontal and vertical directions were consistent, and the continuity was in the order of loam>sand>clay>sandy loam. Soil texture in topsoil was homogeneous, and the main textural type was loam. The main profile configurations were loam-clay-loam in the southern part of the study area. When the texture was sand in subsoil, the texture was mainly loam in topsoil. These results were helpful for understanding the spatial distribution characteristics of soil texture in the alluvial plains.The exponential equation can well describe the vertical distribution of mean SOC. The coefficient of determination (R2), the root mean square error (RMSE) and the mean prediction error (ME) between the measured and the predicted SOC content were0.52,1.82g kg-1and-0.24g kg-1, which suggest that the KPDF method can be used to produce the three-dimensional map of SOC. The surface SOC content was high in the mid-west and south regions, and low value lies in the southeast corner. The SOC contents between five different depths have significant positive correlation and the correlations of SOC contents are larger in adjacent layers than non-adjacent layers. The SOC content for fine textural soil is higher than that of soil with coarse texture, and the SOC content of topsoil in vegetable land was higher than that in wheat land and cotton land. The effect of soil texture on SOC is higher than land use types. |
PDF Full Text Request