| Soil nutrients affect the yield and quality of crops.The acquisition of detailed soil nutrient information is the basic data necessary for environmental management,monitoring,modeling,and precision agriculture.The level of soil quality in my country is uneven.As time goes by and people use different ways,on the spatial scale,soil nutrients show obvious differences and have a high degree of spatial variability.The loess hilly and mountainous areas are complex and diverse,and land use patches are fragmented.To accurately obtain the spatial distribution of soil nutrients in the area,sufficient sampling points are required,which causes a large consumption of manpower and material resources.Therefore,it is of great significance to study the spatial interpolation method of soil nutrients in loess hilly and mountainous areas.This study combines geostatistics and soil landscape modeling theory,selects Pianguan County as the study area,and uses a digital elevation model(DEM)to generate three topographic factors including slope,aspect and topographic position index,based on the same topography Based on the principle of similarity of topographic features of the unit,clustering and partitioning are carried out to divide the topographic unit;the kriging interpolation method in geostatistics is used to interpolate the topographic unit hierarchically to predict the spatial distribution of soil nutrients;and the topographic unit is not divided.The interpolation results of global soil nutrient points are compared,and the prediction accuracy of sub-topographic unit interpolation and global interpolation is tested.The main conclusions of this study are as follows:(1)Based on the three terrain factors of slope,aspect and terrain position index,the cultivated land in the study area is divided into slope foot,flat land,flat top peak,valley,shade slope,sunny slope,semi-shady slope,semi-sun slope,and slope shoulder 11 types of terrain units,ridges,and steep slopes.Based on the principle of similarity of spatial features of the same terrain unit,the study area is divided into four terrain units: valley,ridge,shoulder and toe using systematic cluster analysis.(2)The order of the average content of organic matter in different terrain units is valley> ridge> foot of slope> shoulder;the order of average total nitrogen content is valley> ridge = foot of slope> shoulder;the order of average content of available phosphorus is valley> shoulder> ridge >Slope foot;the average content of available potassium is gully>ridge>slope shoulder>slope foot.(3)The linear function fitting values of the measured and predicted soil nutrient values in different terrain unit types are different from the global ordinary interpolation and are better than the global interpolation.(4)Based on the interpolated soil nutrient content of the terrain unit,the spatial prediction results are more detailed,and the degree of fine fragmentation is consistent with the distribution law of the terrain unit,which can more reveal the detailed information of the spatial distribution of soil nutrients.The prediction result has a smooth boundary and a gradual transition.The research results show that the interpolation after dividing the terrain unit can well eliminate the prediction error of the interpolation result of individual samples,and its prediction accuracy is higher than the global interpolation result,and it has the unbiased prediction.The prediction results of soil nutrients based on topographic units are all significantly correlated at the 0.01 level,and the significance is greater than the prediction results of global ordinary interpolation. |
