The Study On The Regression Model Of Shrub And Herb Layers Leaf Area Index Based On Parameters Of Community

Urban green land plays an important role in improving city environment quanlity. The afforestation level of urban is regarded as the important attribute to urban sustainable development. The important target which is ecological benefit to appraise the urban sustainable development mainly lies on plant community’s structure and the three dimensional green quantity, meanwhile it is leaf area index which is the important target to reflect the three dimensional green quantity that provides the theory basis for the quantitative analysis city ecological benefit.We chose 30 low-stature bush and herb plants which were common and representative in artificial green space of Wu Han as study object, then calculated the actual leaf area index of these plants, meanwhile we established LAI regression equations. The main results are as follows:(1) We established regression models including unitary and multielement linear regression, as well as curvilinear regression for 18 plants, then got the optimal regression models which F and T test were all prominent (pr<0.01). Fitting effect best plant was Dichondra repens, R2=0.857, regression equation was Y=0.021e0.043x1, Fitting effect worst plant was Ophiopogon japonicus, R2=0.487, regression equation was Y=0.444e0.021x1.(2) On the level of community, the optimal regression models of actual LAI value were all extremely prominent level (pr<0.Ol). Meanwhile, the models of single-layered obviously preceded the models of compound-layered, and the curve returns surpassed linear regressions, in the curve return, the exponential function equation fitting effect was best. the Fitting effect of vine layer model was the best, R2=0.668, the optimum regression equation:Y=0.034x1+3.656x2-1.17, the Fitting effect of bush level was the worst, R2=0.375, the optimum regression equation:Y=0.145e0.033x1.(3) The results of optimal regression models precision test revealed:on the level of single plants, when a=0.05, only Parthenocissus tricuspidata、Ligustrum×vicaryi and Buxus sinica whose models forecast precision were outweigh 80%; when a=0.01, only Parthenocissus tricuspidata whose model forecast precision was outweigh 80%. On the level of community, when a=0.05, except independent bush level, other seven classes whose models forecast precision were outweigh 80%; when a=0.01, except independent bush level and bush level, other six classes whose models forecast precision were outweigh 80%.