Study On The Hyperspectral Remote Sensing Monitoring Method Of Prairie Plant Diversity

Under the circumstance of rapid decrease in global biodiversity,it is of great significance to use remote sensing technology to realize the large-scale and real-time monitoring of biodiversity.Biodiversity plays an irreplaceable role in maintaining the stability of the ecosystem in the grassland area.However,compared with forests,deserts and other ecosystems,grassland areas have always been difficult for remote sensing monitoring of biodiversity due to the short and staggered growth of plants.Therefore,in order to solve the problem of how to use remote sensing technology to monitor plant diversity in grassland areas,this paper used the Baorixile grassland as the main research area,using hyperspectral remote sensing images obtained from an airborne platform and combining ground survey data to conduct the following study:(1)A spectral library of 32 typical vegetation in the study area was constructed,and a method for remote sensing mapping of biodiversity index based on species classification results was proposed.Comparison with the ground survey values,it is believed that the direct monitoring method of plant diversity based on species identification has certain applicability in grassland areas when there is a plant spectrum library,but it is easy to cause an underestimation of diversity.(2)The correlation between spectral diversity information,texture information,vegetation height heterogeneity information and plant diversity was studied,a remote sensing indirect monitoring method for grassland plant diversity that integrates multiple remote sensing heterogeneity information was proposed,and plant diversity mapping was realized.Comparison with the results obtained by the direct monitoring method,it is found that although the precision of the indirect monitoring method is low,it has great application potential because it does not need to know the species distribution in the study area in advance.(3)To solve the problem of little research on ? diversity,a spectral diversity index-ED based on hyperspectral remote sensing images was established after referring to the mathematical Euclidean distance.After comparing with the measured values,it was found that the ED index can better reflect the plant ? diversity in the study area.The correlation coefficient with the Bray-Curtis index is 0.73,and the correlation coefficient with the Jaccard index is 0.56.(4)The indirect remote sensing monitoring method of plant diversity was applied to low-spatial resolution images obtained by UAV platforms and aerial platforms.It was found that although the precision of the model has been reduced,precise estimates can still be obtained.It showd that when the size of the image pixel was reduced from close to the size of the monitored object to much larger than the size of the monitored object,the indirect remote sensing monitoring method of plant diversity proposed in this paper was still applicable.At the same time,a general multi-scale remote sensing model for biodiversity monitoring based on species-area relationship and spectral diversity-area relationship was proposed,which proved to be feasible to a certain extent.This paper taken the grassland vegetation diversity as the research object,respectively proposed remote sensing direct monitoring method and indirect monitoring method,and analyzed the respective advantages and disadvantages and application conditions of the two methods,and a method to construct a general multi-scale remote sensing monitoring model was proposed,which is of certain guidance significance for the future remote sensing monitoring of vegetation diversity in prairie.There are 64 figures,26 tables and 107 references in this thesis.