Recognition Of Surface Features And Cyanobacteria Blooms And Study On Groundwater Level Evolution Based On Multi-source Remote Sensing

Remote sensing image classification is an important part of remote sensing research.How to select the appropriate classification method and remote sensing image to recognize the ground object and meet certain accuracy is of great significance to the research of remote sensing image.In recent years,remote sensing data has also been widely used in small-scale and large-scale ecological environment research.With the rapid economic development of Chaohu Lake Basin,a large amount of pollution enters Chaohu Lake,resulting in frequent occurrence of cyanobacteria blooms.Using remote sensing technology to identify the cyanobacteria blooms in Chaohu Lake and analyze the temporal and spatial dynamic changes of cyanobacteria blooms can provide the basis for monitoring and early warning of cyanobacteria blooms.On the other hand,shallow groundwater is an important water resource for industrial and agricultural production in Huaibei plain.Analyzing the evolution of shallow groundwater level and its influencing factors,and combining with remote sensing data to study the ecological response relationship between surface water,lakes,vegetation and groundwater is of great significance to regional economic development and environmental protection.In view of the above situation,this paper mainly carried out the following four aspects of work:(1)based on the incentive function and learning rate of BP neural network to improve,comparative analysis of its classification effect.Using this method,ZY-3 image and GF-1 image are classified,and the feasibility of replacing ZY-3 image with GF-1image is discussed.(2)NDVI was used to extract the information of cyanobacteria bloom in Chaohu Lake in 2018,and the spatial and temporal distribution characteristics and influencing factors of cyanobacteria bloom were analyzed.(3)The collaborative Kriging method is used to interpolate the groundwater level data of different generations in Huaibei plain,and the evolution process of groundwater level and spatial distribution of mining intensity are analyzed.(4)MNDWI and NDVI were used to extract the information of lakes and vegetation in Huaibei plain,and the responses of groundwater and vegetation were studied.The main results and conclusions are as follows:(1)compared with the maximum likelihood method and BP network classification method,the overall accuracy of the improved BP neural network method is increased by 15.30% and 23.81% respectively,and the kappa coefficient is increased by 0.18 and 0.28 respectively,which can effectively improve the classification speed and accuracy.According to ZY-3 image,the overall accuracy of GF-1 image classification is 88.02%,which can replace ZY-3 image in most cases.(2)In 2018,the cyanobacteria blooms in Chaohu Lake first appeared in July,and the outbreak was the most serious in September.The outbreak frequency of cyanobacteria bloom in West Chaohu Lake was the highest,followed by Zhongchao lake.In addition,temperature is the premise of cyanobacteria,and wind power is the key factor to control the accumulation of cyanobacteria.(3)The distribution of groundwater level in Huaibei Plain is V-shaped,and the evolution process of groundwater level is as follows: the groundwater level in the first generation(1978-1993)is in a natural state;in the second generation(1994-2003),the artificial exploitation activity is the most intense,the precipitation in the dry year is less,and the groundwater level is obviously reduced due to the influence of human exploitation;and the groundwater level is strong in the third generation(2004-2010)The groundwater level was restored.In addition,the distribution of mining intensity in the second generation is extremely unbalanced,and the mining intensity in Huaibei,Suzhou and Bozhou is the largest.(4)There is a good response relationship between groundwater depth and surface water area and vegetation in Huaibei Plain: the surface water area decreases with the increase of groundwater depth,which is most obvious when the groundwater depth is 2.5?3.5m;when the depth is less than 1.5m or more than 3.5m,the vegetation development is poor,and 1.5?3.5m is the suitable depth for vegetation growth.The research can provide reference for remote sensing image classification,monitoring and treatment of cyanobacteria bloom in Chaohu Lake,rational utilization of groundwater resources and groundwater related ecological environment protection.