원격탐사기법을 활용한 소나무재선충병 감염의심목 예찰 가능성 평가

초록

This study aimed to detect the pine wilt disease using remote sensing techniques in Yuseong-gu, Daejeon, Korea, where the disease first occurred in 2024. A vertical take-off and landing unmanned aerial vehicle was used to produce orthoimages with a resolution of 6.7cm, and tree species were identified through texture analysis. A random forest algorithm was applied to the orthoimages, identifying 46 suspected cases of pine wilt disease. Based on these results, a field survey was conducted to verify accuracy, confirming that 33 out of the 46 suspected trees were infected, resulting in an accuracy of 71.7%. The Normalized Difference Vegetation Index (NDVI) was compared to evaluate the feasibility of using medium-resolution satellite imagery from April to November 2024. A decrease in maximum and average NDVI values was observed in August compared to June, suggesting disease progression or tree mortality. However, there was no statistically significant difference detected in the summer images (June to October) (p>0.05), and the resolution (10×10m) was considered relatively low for determining the normalized vegetation index of an individual tree. Therefore, it is concluded that the detection of pine wilt disease using orthoimagery is feasible even in newly affected areas, while satellite imagery can be effectively utilized for large-scale damage assessment but may be insufficient for detecting individual trees. To enhance these results, increasing the number of training datasets for machine learning and acquiring high-resolution images using multispectral cameras will be necessary for more precise vegetation index analysis.

키워드