Application Of Chemometrics In Analysis And Modeling For Chemical Data Of Precipitation

Acid rain pollution has been paid great attention by environment protectors as a global affair. It does make direct or indirect negative impact on hydrophytic ecosystem, grand ecosystem, buildings and humans' health as well. Acid rain is made from sedimentation of acidic materials. The pH value of ordinary precipitation can reach 5.6 since the carbon dioxide can dissolve to form saturated acidic water. Hence the precipitation with pH value less than 5.6 is called acid rain. Human activity can create millions tones of sulfur dioxide and nitrogen oxide every year, which is oxidized in air and mixed with water, making pH value of rain drop down the level of 5.6. The paper focuses on the precipitation data of Dongguan city from year 2000 to 2003.We've made some basic research on the fact of acid rain in Dongguan whichincludes the acidic condition of precipitation, the relationship of pH value of precipitation with its volume amount and happening time, the chemical substitute of precipitation etc. the results indicate that the pollution of acid rain has reached a severe level.The pH value of precipitation depends on the existence of various acidic and alkalescent irons, among which multiple pertinence exists greatly. According to the data of iron concentration in precipitation and corresponding pH value, chemometrics can help us to set up the model to predict the pH value of precipitation, screen out the irons contributing greatly to its solution, and extracthidden information among precipitation data. It is greatly important for theresearch of acid rain protection.In this article, various chemometrics methods, including stepwise regression (SR), multiple linear regression (MLR), principal component regression(PCR), project pursuit regression (PPR), and newly born support vector machine, are employed to parse the precipitation data of Dongguan and to set up the models of pH value of precipitation against iron concentrations.Since there are multiple pertinences among various factors, the pH value and iron concentrations of precipitation are hardly linearly interrelated. The results also support this because nonlinear methods overwhelm linear ones in prediction ability. Among all the methods, SVMR ranks first with its outstanding learning and generalization ability. The SVMR results of precipitation analysis uncover the fact that the pH value of precipitation in Dongguan has no relationship with the concentration of-S?43:: and NOi^yet effected greatly by alkalescent irons.