Support vector machine applied to regularities of trace element content in soil of China

Xu Zhanfeng,Zhang Zhen,Zhang Liangmiao,Lu Wencong,Chen Nianyi

DOI: https://doi.org/10.16866/j.com.app.chem2006.04.003

2006-01-01

Abstract:Support vector classification and regression methods have been applied to investigate the data of trace element content in the soil of 29 provinces, autonomous regions and cities of China. It has been found that there exist obvious correlation relationships between the data of the contents of many trace elements and that of major elements or soil acidities. The elements Na,K, Mg, Ca, Sr and Ba, forming basic hydroxides , exhibit positive correlation with the pH of soil; while the elements Al, Fe, Ti, Zr, Ta forming sparsely soluble hydroxides, exhibit negative correlation with the pH of soil. These facts are the results of the leaching action by humid and hot weather environment in the course of the formation and weathering of soil. On the other hand, the data of the contents of Rb and Cs in Chinese soil seem to be the exceptions of the above-mentioned regularities. The red soil of Nanling region and South Tibet is the weathering products of re-melted granite with concentrated lithophile elements, and the Rb and Cs in these red soil are concentrated into muscovite, which is rather stable in weathering process. So that the Rb and Cs contents of these red soil are especially high. This fact indicates that the trace element content in soil is not only dependent on leaching process, but also influenced by the phase relationship of soil-forming rocks. In view of the fact that the contents of many trace elements in soil are correlated with the contents of some other elements or the pH of soil, it should be careful to differentiate the true causal relationship and the indirect correlation if some correlation between certain trace element content and certain environmental effect has been found by data processing.