Study On The Estimation Of The Damage Degree Of Rats In Zoige Alpine Grassland Based On Low-altitude Remote Sensing

In recent years,the continuous occurrence of grassland rodent damage has led to the increasingly serious contradiction between grass and livestock,which has seriously restricted the economic development of local animal husbandry and the protection of ecological environment.Using UAV image to estimate the range and harm degree of rodent damage in Zoige alpine grassland,it can quickly,efficiently and accurately obtain the area and harm range of rodent damage,accurately monitor and predict rodent damage,and provide scientific reference for scientific rodent control and accurate rodent control,which is of great significance to the ecological protection of grassland and the sustainable development of animal husbandry.In this study,the typical rodent damage area of Zoige Wetland Nature Reserve was taken as the study area,and the high-resolution orthophoto image and digital elevation model image of the main rodent damage species in this area were obtained by aerial photography at low altitude.Four classification algorithms based on object-oriented classification were used to obtain the best extraction methods of the two kinds of rodent damage information by comparing the accuracy of the two kinds of rodent damage information extraction.Using remote sensing images with different spatial resolution to extract the mouse hole information,the influence of resolution on the mouse hole information extraction is analyzed from many angles.Based on the sample data of field survey,the effective hole opening coefficient of high altitude rat rabbit and high altitude rat was statistically analyzed,and the number of extracted rat holes was converted into relative effective hole density.Then,based on the GIS grid,the weight of each evaluation index was determined by entropy method based on the variables of population density,hole area,vegetation cover and area of rat wasteland.The main findings are as follows:(1)Using various object-oriented classification methods to extract rodent damage information from aerial images,and comparing and analyzing the extraction results,the optimal method model of rodent damage information extraction is obtained.According to the precision comparison analysis of four supervised classification methods,the object-oriented nearest neighbor classification method is the best method for extracting rodent damage information from highland rat rabbits,and the object-oriented Bayes classification method is the best method for extracting rodent damage information from highland rat.Using remote sensing images obtained at different flight altitudes to extract rodent damage information,the analysis shows that with the increase of flight altitude,the lower the spatial resolution of the image,the lower the time,classification accuracy and the number of patches required for its classification.(2)Based on the GIS grid,the damage degree model was constructed and the damage area was counted.The results showed that the regional distribution of rabbit hazards in the 1:1 plateau showed the trend of multi-nuclear agglomeration and zonal distribution.The main damage in aerial photography area was mild(grade I)and moderate(grade II).The area of mild and moderate hazards was 90703.8 m~2,which accounted for 65.5%of the total area,and the area of extreme hazards(grade IV)accounted for 16%of the total area.22096.8 m~2In the northern part of the sample area,mild and moderate hazard area was 42484.94 m~2,the area was 69.7%,the extreme hazard was concentrated in the central and eastern regions,a small part was concentrated in the southern and western regions,and the severe and moderate hazards were concentrated in the extreme hazard distribution area as the core.