简介：Dustemissionsfromsmelters,asamajorcontributortoheavymetalcontaminationinsoils,couldseverelyinfluencesoilquality.Downwindsurfacesoilswithin1.5kmofazincsmelter,whichwasactivefor10yearsbutceasedin2000,inMaguTown,GuizhouProvince,ChinawereselectedtoexaminePb,Zn,andCdconcentrationsandtheirfractionationalongadistancegradientfromazincsmelter,andtostudythepossiblee?ectsofPb,Zn,andCdaccumulationonsoilmicroorganismsbycomparingwithareferencesoillocatedatadownwinddistanceof10kmfromthezincsmelter.Soilswithin1.5kmofthezincsmelteraccumulatedhighlevelsofheavymetalsZn(508mgkg-1),Pb(95.6mgkg-1),andCd(5.98mgkg-1)withlowratiosofZn/Cd(59.1-115)andPb/Cd(12.4-23.4).Compositepollutionindices(CPIs)ofsurfacesoils(2.52-15.2)were3to13timeshigherthanthereferencesoils.Inmetalaccumulatedsoils,exchangeablepluscarbonate-boundfractionsaccountedformorethan10%ofthetotalZn,Pb,andCd.Thesaturationdegreeofmetals(SDM)insoilswithin1.5kmofthesmelter(averaging1.25)wassixtimeshigherthanthatofthereferencesoils(0.209).Asmallersoilmicrobialbiomasswasfoundmorefrequentlyinmetalaccumulatedsoils(85.1-438μgCg-1)thaninreferencesoils(497μgCg-1),andanegativecorrelation(P<0.01)ofsoilmicrobialbiomasscarbontoorganiccarbonratio(Cmic/Corg)withSDMwasobserved.Microbialconsumptionofcarbonsourceswasmorerapidincontaminatedsoilsthaninreferencesoils,andashiftinthesubstrateutilizationpatternwasapparentandwasnegativelycorrelatedwithSDM(R=-0.773,P<0.01).Consequently,dustdepositedPb,Zn,andCdinsoilsfromzincsmeltingwerereadilymobilized,andweredetrimentaltosoilqualitymainlyinrespectofmicrobialbiomass.

简介：土壤侵蚀过程复杂，很难直接应用土壤侵蚀预报方程进行定量计算。作为一种新的机器学习算法，支持向量机在样本有限的情况下，采用结构风险最小化准则，把学习问题转化为一个二次规划问题，从而得到唯一的全局最优解。首次尝试将最小二乘支持向量机技术用于土壤侵蚀预测，并与BP神经网络的方法进行了对比，取得了较好的预测精度。