Informations et ressources scientifiques
sur le développement des zones arides et semi-arides

Accueil du site → Master → Chine → 2021 → Research on Rat Hole Recognition of Typical Desertification Grassland Based on UAV Hyperspectral Remote Sensing

Inner Mongolia Agricultural University (内蒙古农业大学) 2021

Research on Rat Hole Recognition of Typical Desertification Grassland Based on UAV Hyperspectral Remote Sensing

李鹤

Titre : Research on Rat Hole Recognition of Typical Desertification Grassland Based on UAV Hyperspectral Remote Sensing

Auteur : 李鹤

Grade : Master 2021

Université : Inner Mongolia Agricultural University (内蒙古农业大学)

Résumé
Rodents are an important part of China’s ecosystem and one of the main organisms that cause grassland degradation.The grazing behavior of prairie rats has a great influence on the grass growth on the grassland and the digging behavior of prairie rats also leads to grassland degradation.Currently China’s national standards are based on the proportion of bare soil and rat burrows to the total area of grasslands.In recent years in order to collect relevant data we have been using the traditional manual survey method which not only wastes manpower material resources and time but also fails to effectively monitor large-scale grassland degradation and rodent infestation.In this study UAVs and low spectral remote sensing technology were used to monitor the typical desert steppe.The experiment area of Gegentala Grassland in Inner Mongolia Autonomous Region was selected to collect low-level hyperspectral data by UAVs and then the first-order differential threshold was adopted.By searching the spectral information confidence interval and optimal threshold of rat hole the spectral method can effectively identify the typical rat hole in desert steppe.In this thesis the mouse hole in the typical desert steppe is taken as the research object.Firstly the most commonly used normalized vegetation index and soil adjusted vegetation index(NDVI and SAVI)were calculated to calculate the low altitude hyperspectral data and the spectral interval distribution of rat holes in the experimental area was statistically analyzed by the threshold method.The results showed that although the normalized NDVI and SAVI could effectively distinguish between vegetation and non-vegetation(bare soil and rat hole)the reflectance of rat hole and bare soil overlapped due to the low reflectance of the two plantations.Ground features therefore cannot be effectively distinguished.Therefore the first-order differential threshold method was used to distinguish rat hole from bare soil.Using vegetation as the background it overcomes the problem of invalid identification caused by the overlap of the reflectance of rat hole and bare soil.The results show that the first-order difference threshold method can be used for the identification of prairie rat holes.The overall accuracy was 91.058% and Kappa=0.821 which achieved the purpose of high precision identification.The results can provide a scientific basis for hyperspectral identification of rat holes in the desert grassland in the high altitude UAV area.

Mots clés : UAV ;Hyperspectral remote sensing ;Desert grassland ;Rat hole ;First order differential threshold method ;

Présentation (CNKI)

Page publiée le 4 mars 2022