The Spatial Variation Characteristics Of Soil Nutrients Are Responsive To Human Interference And Its Spectral Inversion

As an important part of the soil system,the degree of abundance and presence of soil nutrients directly determine the soil fertility and affect the growth of vegetation.Soil nitrogen nutrient can promote the growth of plant roots,stems and leaves;phosphorus plays an important role in photosynthesis,respiration and biochemical regulation of plants;potassium can regulate the water potential of plant cells,promote the photosynthesis of plants and increase enzyme activity.With the accelerating rate of population growth,due to its fixed position and limited number of land,the supply of land is obviously insufficient.The way,intensity and time of human activities will have an important impact on the soil properties and an important reason for the change of soil physical and chemical properties.The study on the spatial variability of soil nutrients helps to understand the process of soil formation,structure and function,and it is of great significance to improve the utilization efficiency of soil nutrients and make rational use of land.The rapid acquisition of soil nutrient content information is the key to the development of precision agriculture.Compared with traditional soil nutrient content determination methods,spectrum analysis technology is a rapid,non-destructive and simple measurement method,which provides a possibility for rapid determination of soil nutrients.There are fewer papers comparing the spatial heterogeneity of soil nutrients with human interference and no human interference,and the spectral modeling of different differential transformations.So,this paper selects 55 soil samples from undisturbed and artificially disturbed areas.Descriptive statistical analysis,correlation analysis and principal component analysis method are used to explore the characteristics of nutrient content in the two regions.Geostatistics was used to study the spatial heterogeneity of total nitrogen,total phosphorus,and total potassium nutrients in 0–10 cm soil layers.The spatial variability and spatial distribution characteristics of soil nutrients in the study area were obtained,and artificial soil reclamation and utilization were analyzed;using soil spectral analysis techniques,through the correlation analysis of soil spectral reflectance data of six differential transformations with the content of total nitrogen,total phosphorus,and total potassium,the characteristic bands that passed the significance test were selected and independent variables were used to establish the regression model,and then select the best inversion model,in order to provide a reference for the rapid monitoring of nutrients in the two regions.The results show:(1)Compared with the undisturbed area,most of the land in the human-interference area is flattened,and the original vegetation,litter,and root systems of the soil surface are removed.In the process of land plowing,soil permeability increases and the microbial activity increases.The conversion rate of soil nutrients is accelerated,which promotes mineralization or release of nutrients in the soil.It further improves the correlation between organic matter,total nutrients and available nutrients,and increases the content of available nutrients.In addition,artificially cultivated land and nursery land consume a certain amount of soil nutrients,and at the same time due to lack of artificial fertilizer supplementation,the content of organic matter and total nitrogen total phosphorus in human-disturbed areas is relatively reduced.The results of principal component analysis showed that total nitrogen,total potassium,total phosphorus and available nitrogen were the characteristic elements in the undisturbed area.And the available elements were available potassium,total potassium and available nitrogen in human-disturbed area.(2)The theoretical model of the best semivariogram function of soil total phosphorus in the undisturbed area is a spherical model.The Gaussian model is the optimal theoretical model for total nitrogen,total potassium in the undisturbed area and total nitrogen,total phosphorus and total potassium in human-disturbed area.Although R~2 of the whole model in the two regions was between 0.66 and 0.90,the RSS was less than 0.52,and the model fitting accuracy was high,but the R~2 in the undisturbed area model was higher and the RSS was smaller.The spatial variability of total nitrogen,phosphorus and potassium in the two regions was mainly caused by structural factors.Combined with the vegetation condition at the sampling site,the spatial distribution of soil nutrients was analyzed.It was found that the high-value areas of total nitrogen,phosphorus and potassium in the undisturbed area were all within 25%of the vegetation coverage.The human-disturbed areas,especially the forest areas where human activities have a greater impact on the soil surface,have less total nitrogen and total phosphorus.This is mainly due to the fact that the mineralization of organic matter accelerates the decomposition of total nitrogen and total phosphorus,while the artificial forestland consumes a certain amount of soil nutrients and lacks the return of organic matter or the addition of artificial fertilizer.Therefore,in order to meet the demand for nutrients in forestry and nursery land,it is necessary to add a certain amount of nitrogen fertilizer and phosphate fertilizer in a timely manner in the area.(3)In the stepwise regression model of total soil nitrogen,phosphorus and potassium established by the spectral reflectance data of the six differential transformations in the two regions,The model set R~2 of the total nitrogen and potassium prediction model established by second-order differential in the undisturbed area and the first-order differential of total phosphorus by logarithm were all over 0.9,and the test set RPD was greater than 2.0.The total nitrogen and total phosphorus prediction model established by the first-order differential of human disturbance area,not only had an RPD of>2.0 but also had an R2>0.9,and reached an excellent level.Among the sensitive bands of the RPD>2prediction model,the undisturbed areas are located in the visible near-infrared region,while the artificial disturbance regions are in the near-infrared region.