Research review on methods of spatial prediction of soil nutrients

Soil nutrients are the necessary nutrients which the soil provides for plants growth.Some research showed that the soil nutriment has high spatial variability which affected by natural and human factors such as soil parent materials,regional climate,terrain,cropping system,fertilization and so on.The characteristic of soil spatial variability can be found not only in the regional scale but also in the field scale.In recent years,for aiming at minimizing fertilization and optimizing management of precision agricultural,many methods which improved continuously from fisher statistics to geostatistics,and then the new methods of neural network,geographical information technology and high accuracy surface model have been applied on measuring the spatial variability and distribution of the soil nutrients in different scales.Taking the geostatisics applied on the research on the spatial variability of soil nutrients as the starting point,the status of geostatistics-based research on the spatial variability of soil nutriments has been reviewed,which include using geostatistics to determine the reasonable soil sampling scheme,quantitative study and scale effect of the spatial variability of soil nutrients.On this basis,the recent research progress of methods and technologies of soil nutrients spatial variability were briefly described,which includes neural network technology,3S technology and high accuracy surface model.At the last,it comparatively analyzed shortages of all methods in researching the spatial variability of soil nutrients.Geostatistics method is concerned with detecting,estimating and mapping the spatial variability of soil nutrients,however it usually with intensive subjectivity and based on to much assumed conditions,moreover,the smoothing effect was existed in Kriging interpolation.The neural network technology and high accuracy surface model can improved the accuracy of interpolation and eliminate the smoothing effect,but the parameters of neural network can not be determined easily and the data operand and computer memory capacitance of high accuracy surface model are extremely heavy.The prediction of soil nutrients spatial variability by using geographical information technology has some uncertainty.At the same time some research directions of spatial variability of soil nutrients in the future are discussed.More scientific efforts should be focused on research in the spatial variability of soil nutrients in the different periods of crop growth,evolution mechanism of soil in four dimensional spaces and the uncertainty of environmental information.