In recent years, the natural Pinus sylvestris var. mongolica (Mongolian pine) forests on the Hulunbuir sandy lands in northeastern China have experienced substantial decline, with the affected area continuously expanding. Although this decline is often linked to combined effects of human disturbance and climate change, understanding its underlying mechanisms first requires defining the key decline characteristics, such as the minimum detectable area, severity grades, and spatial distribution patterns. However, previous studies have lacked clear quantitative criteria for determining when a stand can be considered in decline, leading to inconsistencies in assessment across different spatial scales. In this study, we used multispectral unmanned aerial vehicle (UAV) imagery and a random forest algorithm to evaluate vegetation indices for detecting forest decline. The species-area curve and upscaling methods were applied to determine the minimum detectable area and to map decline distribution. The results indicated that the Green Normalized Difference Vegetation Index (GNDVI) is a reliable indicator of Mongolian pine decline. The optimal survey area for accurate assessment was 3.24 ha, and the decline severity could be classified into three levels: severe (VGNDVI < 0.57), mild (0.57-0.63), and normal (> 0.63). Over the study region, approximately 1 % of the forest area showed signs of decline, with severe cases being sparsely distributed. Overall, this study identifies GNDVI as an effective indicator and clarifies the characteristic scales and spatial patterns of Mongolian pine decline, providing a scientific foundation for precision monitoring and restoration planning.