Research on air negative oxygen ions concentration prediction in Fujian Province based on PCA-RF combination model
The concentration of air negative oxygen ions(NOI)is an important indicator to evaluate the freshness and cleanliness of the air.In order to improve the monitoring capability of NOI concentration,the key influencing factors of NOI concentration were analyzed by considering meteorological elements and remote sensing factors,and a PCA-RF prediction model of NOI concentration was constructed by using Pearson correlation analysis,PCA analysis,and random forest machine learning method(RF)in Fujian Province.It was found that:⑤The distribution of NOI concentration was significantly correlated with wind speed(Wair),air temperature(Tair),atmospheric pressure(Pair),visibility(IvIS),aerosol optical thickness(hA0D),vegetation index(INDVI),humidity index 1(INDMI1),vegetation water supply index(IVSWI),and brightness index(INDSI)(all passed the 0.01 significance test),where Wair,IVIS,INDV1,and IVSWI were positively correlated with NOI,and Tair,Pair,hA0D,INDMI1 and INDSI were negatively correlated with NOI.②When the principal component score was 7,the cumulative contribution rate of variance reached 93.36％,which can represent most of the information of all factors.③The opti-mal ntree and mtry of the PCA-RF model were 400 and 7,respectively.The top three factors that have a significant impact on the NOI con-centration in the Fujian region are Pair,IVIS,and IVSWI,respectively.④The RMSE of the PCA-RF model on the validation set was 803.73 ions/cm3,R2 was 0.44,and MAE was 548.79 ions/cm3.
air negative oxygen ionsmeteorological factorsremote sensing factorsPCA-RFprediction model
万方数据
维普
