Investigation Of Health Assessment For Lake Ecosystem Based On Support Vector Machine (SVM)

In recent years, environmental pollution problems continued to emerge with therapid development of human society, and the physical conditions of many ecosystemswere deteriorating. In this case, studies of ecosystem health gradually became theresearch priorities and hotspot in the field of international environmental protectionwhich were driven by the concept of sustainable development. In many ecosystems,the health status of lake ecosystems received much concern due to the direct and closeconnexion of it and the human beings. Currently, lake ecosystems were withstandingincreasing external pressure year by year with the growth of population anddevelopment of industrial and agricultural production. Growing degradation of lakeecosystems have been a serious threat to the sustainable development of sustainabledevelopment and people’s health. Thus related researches on lake ecosystems’ healthare imminent. As investigation of lake ecosystems health assessment is the primaryfactor to realize the health status of lake ecosystems, it must be paid sufficientattention and the scientifically correct approach must be ascertained to conductunified management on the lake ecosystem health.A health assessment model for lake ecosystems was proposed based on theadvantages of support vector machine (SVM) on dealing with classification, smallsample size problem, generalization and promotion. Meanwhile, tracking andmonitoring of ecosystem and social investigations were conducted in the early stageof establishing Baiyun Lake of Guangzhou Baiyun District. Moreover, five indicatorsincluding BOD5, the synthesized trophic state index, biodiversity index, dissolvedoxygen and structural energy quality were selected to analyze Baiyun Lake’s currentsituation based on the obtained data. Five indicators were classified based on NationalInstitute of Standards and previous academic experience, and then a Baiyun Lakeecosystem health assessment index system involving five health levels of “veryhealthy”,“healthy”,“subhealthy”,“general pathological” and “pathological” wasconstructed. Finally, lake ecosystem health assessment model based on support vectormachine (SVM) was trained according to the established index system. The modelwas then applied to the Baiyun Lake ecosystem health assessment to obtain the resultsof Baiyun Lake ecosystem health evaluation. It was demonstrated by the assessmentresults that the ecosystem of Baiyun Lake was in pathological state, and didn’t have the expected function of purifying water. To improve the ecological system health ofBaiyun Lake, three methods were suggested including improving input water quality,intercepting pollution sources and enriching biomass.In addition, the results of Baiyun Lake ecosystem health evaluation based onsupport vector machine (SVM) and traditional comprehensive entropy weight healthindex method together with entropy weight fuzzy comprehensive evaluation methodwere compared in this paper. It demonstrated the scientificalness of lake ecosystemhealth assessment model based on support vector machine (SVM). Compared withtwo traditional evaluation methods (Entropy weight comprehensive health indexmethod and Entropy weight fuzzy comprehensive evaluation method), this model ismore objective and scientific on evaluating the health of lake ecosystem. It canprovide some basises for health management of lake ecosystem. Overall, this modelhas a promising application prospect.