The 3D Modeling Methods And Uncertainties Analysis Of Soil Properties

Well soil ecological environment is the necessary basement of achieving sustainable agricultural development.As a farming field,soil nutrients are directly related to crop growth in the field.The study of the spatial variability of soil properties in the field is beneficial to the accurate description of soil fertility,the realization of scientific fertilization and crop production,and to provide reliable data for the realization of precision agriculture.The current research on soil properties of field scale is mostly limited to the surface soil or horizontal direction.The description of soil properties in 3D space requires a reliable three-dimensional interpolation method to improve the prediction accuracy of soil properties.This paper combined field sampling,physical and chemical analysis,classical statistical methods,and the method of geostatistical interpolation is extended to obtain soil samples with 10cm as sampling gradient through grid samples.The 3D model construction and spatial variability of soil organic matter,pH,heavy metals,calcium and zinc are constructed and analyzed.The main results are as follows:?1?Methods and modelsEvaluate the accuracy of interpolation results by calculating MAE and RMSE.The results show the Inverse Distance Weighting method can retain the true volatility of the data to a certain extent,the interpolation results obtained by the Ordinary Kriging interpolation method have smaller MAE and RMSE,which can reflect the real situation of soil properties in the field.The 3D Kriging interpolation method is more accurate than the 3D Inverse Distance Weighting method in predicting the coefficient R²,RMSE is smaller and the accuracy is higher.?2?The distribution of soil propertiesThe spatial distribution of soil properties is examined through 3D models visually and directly,calcium and zinc have their own variation result.The calcium showed obvious vertical variation,the content of the tillage layer was about 7000 mg·kg^-1,and the 31-60cm depth Ca content increased to 9500 mg·kg^-1,and gradually decreased deeper than the 61cm depth.The variation of zinc in the 31-100cm is almost the same,and there is a region higher than 95 mg·kg^-1 in the southwest corner of the study area,while the northeast direction is gradually reduced to 60 mg·kg^-1.The change pattern in 0-30cm depth is similar,but the range of content is smaller.?3?Uncertainty analysisBy the uncertainty analysis,the soil type and elevation have little influence on the residual error of the interpolation results,but the absolute value of the residual is larger in the 10m buffer zone formed by the road,and the variation range of the absolute value of the residual difference is close to the result of the 10m radius when the radius of the buffer increases 20m and 30m.Therefore,the disturbance nearby the road is one of the environmental factors which cause the uncertainties.Different ways of interpolation generate uncertainties to some extent,but an optional way and parameters can reduce the uncertainty.