Evolution And Early Warning Analysis Of Ecosystem Pattern In Nansi Lake Wetland

Wetland is a transitional zone between the aquatic ecosystem and the terrestrial ecosystem,known as the"Kidney of the Earth".It has many irreplaceable unique functions,such as providing water for people,animals,industry and agriculture,regulating regional microclimate,and protecting biological diversity.In the past 30 years,the surrounding economy of Nansi Lake wetland had developed rapidly and the process of urbanization has accelerated.It was faced with many problems,such as the decrease of Lake area year by year,the deterioration of water quality and so on.At the same time,due to the unbalanced development of the ecosystem,the quality of the ecological environment also had a downward trend.Therefore,the analysis and early warning of wetland ecosystem pattern in Nansi Lake was of great practical significance for the balanced development of ecosystem in this area.This paper took Landsat remote sensing data,measured data and statistical data as the data support.With the help of RS and GIS technology,paper proposed a multi classifier integrated information extraction method for the four remote sensing data of Nansi wetland in1987,1997,2006 and 2017.Based on the results of information extraction,this paper analyzed the evolution process and characteristics of wetland ecosystem pattern in the past 30 years,and made warning analysis based on CA Markov prediction model.The article clarified the main problems and influencing factors of the Nansi Lake wetland ecosystem pattern,and put forward suggestions for optimization.The main conclusions are as follows:(1)For the four periods of remote sensing data selected in this study,the accuracy of single classifier information extraction results of support vector machine,random forest and artificial neural network were all lower than that of information extraction results after classifier integration.The overall accuracy of information extraction results of multi classifier integration was higher than 90%,and kappa coefficient was about 0.87.(2)The wetland ecosystem pattern of Nansi Lake had complex characteristics of temporal and spatial changes.From the perspective of time,in the past 30 years,the ecosystem area of Nansi Lake wetland had shown a trend of diversity,such as increasing first and then decreasing,decreasing first and then increasing,increasing and decreasing alternately.The trend of change reflected the complexity of ecosystem change in the study area.From the perspective of space,the conversion area of ecosystem types in the three periods was more than 1000 km~2.The junction of water and land was the most obvious and complex area of ecosystem type transformation.The results showed the fragility of ecosystem in the region.The complexity of ecosystem pattern change in the study area was affected by many natural factors such as precipitation and temperature,as well as population,economy and policy.(3)According to the four selected remote sensing data in this paper,the prediction result of CA Markov model based on Markov suitability atlas was more accurate than that based on MCE suitability atlas.From 2017 to 2027,the area of wetland ecosystem and dry land ecosystem in the study area changed a lot,while that of forest ecosystem and urban ecosystem changed a little.The early warning areas in the next 10 years included the junction of water and land,the South Bank of the lake area and the high altitude areas in the study area.(4)The change trend of wetland ecosystem types in Nansi Lake was diverse,and the change process was complex.In order to ensure the diversity of regional ecosystem pattern and the stability of ecological environment quality,this paper put forward three aspects of management and improvement from the ecological perspective:water ecosystem protection and management,improving the intensive utilization level of farmland ecosystem,and overall planning and arrangement of urban ecosystem pattern.The results of this study further proved that the accuracy of multi classifier integrated classification was higher than that of single classifier,and provided direction for wetland remote sensing image information extraction research.This paper provided data support for the wetland ecosystem protection of Nansi Lake,and also provides ideas for the implementation and legislation of wetland protection.