Material Sources Of Laterite In South Of China Based On BP Neural Networks

There are various types of Quaternary sediments in the subtropical regions of China, such as XiaShu Loess,red clay and old red sand,however,the material relation between them have not clearly,specially with red clay in south China.The paper analyses and chooses the type section of the possible material source of red clay in south China-XiaShu Loess, floodplain in the Yangtze River,old red sand and the local Granite weathering crust,then take samples from them. Under the direction of predecessors’ research findings about the material relation of Loess in the north and red clay in the south, choose the geochemical parameters which are relatively stable in supergene conditions and highly sensitive barometer of material source information to build the BP neural network,and make pattern classification and recognition on red clay in south China.Try to discuss the material relation between red clay and other deposit sediment.The paper chooses the geochemical parameters which are relatively stable in supergene conditions like Ti、Al、Sc、Nb、Cr、Co.,and utilization deposit sediment of floodplain in the Yangtze River(CJ), XiaShu Loess located in XinGang(XG),old red sand located in FuJian QinFeng(QF) and ZhouShan DongSha(DS), local Granite weathering crust located in XinChang RuAo(RA) to build BP neural network, then make pattern classification and recognition on red clay in south China,which conclude Fujiawan section (FJW),Fulu section(FL) in Anhui Xuancheng(XC),Yuetang section(YT),Luodian section(LD),Xianqiao section(XQ),Baishaxi section(BSX),Yiwu section(YW),Pujiang section(PJ) in Zhejiang Jinhua,Miaozihu section(MZH) in Zhejiang Zhoushan.The main conclusion are as follows: (1) The main material source of Red clay in south China (red clay in Zhejiang eg) is CJ,the dust come from Nothwest China have less effect than red clay,DS and RA just affect some red clay section,but don’t have universality.the effect of old red sand in QF almost can ignore when take the systematic error into consideration. Make pattern classification and recognition on red clay in south China with3BP neural network:CJ-XG-QF、J-XG-DS、CJ-XG-RA all show that CJ is the main material source, results are69.86%,73.06%,69.86%in3BP neural network;meanwhile,28.31%,20.55%,23.74%of red clay are recognized as XG; samples that are recongnized as QF are only1.83%, which almost can ignore when take the systematic error into consideration.at the same time, samples that are recongnized as DS and RA are identical percentage:6.39%. which is not high as a whole, in addition,they are mainly distribution in BSX and YW section,and don’t have universality.(2) The affect of the dust come from Nothwest China on Red clay in south China reduce gradually From the north to the south,meanwhile, the affect of CJ increase gradually From the north to the south.The percetage of red clay in XC which been classified into CJ are65.12%,72.09%,65.12%in3BP neural network,while75.56%,78.52%,72.59%for JH;samples which are classified XG are32.56%,25.58%,34.88%in XC,and22.96%,12.59%,17.04%in JH.The result of MZH is similar to XC,maybe because of the position.(3) The notogaea of Loess in North China probably reach the south of the Yangtze River, Anhui Xuancheng and Zhejiang Zhousan also have been affected,but,it’s effect is weak when reach Zhejiang Jinhua.The result of3BP neural network show that there are about30%samples are classified XG in XC,but,the proportion is cut in half almost in JH.what indicate that the effect of winter monsoon more widespreader than now in a way.(4) The origin of XiaShu Loess is accept as wind deposition,however,its material source is not clear yet, as a result of the effect of CJ have reach JH,the XiaShu Loess which distributes around XG should be affected by CJ,so,we can infer that the material source of XiaShu Loess should not just the Loess in north or CJ, it should be combined action by both.