三峡库区低山丘陵区多尺度景观指数响应及适宜粒度

景观格局具有典型的空间异质性和尺度依赖性。在对景观格局进行分析研究时需跨越多个尺度，空间粒度大小在尺度聚合分析中至关重要。三峡库区低山丘陵区斑块破碎、景观格复杂性显著、景观格局特征及尺度变异规律仍待明确，本文以三峡库区秭归县为研究区，设定1—400m内23个粒度梯度水平，定量评估县域、乡镇以及小流域不同幅度上的景观格局指数以及拟合函数，探讨多空间幅度景观格局指数随粒度大小的变化特征。基于景观指数的粒度效应特征和拟合函数的曲线特征(最大曲率点、极值点),明确适宜不同幅度的山地景观格局指数研究的粒度阈值，以揭示库区低山丘陵区景观结构的复杂性和变异性。结果表明：不同景观指数对空间粒度变化和空间幅度变化的响应存在差异，不同景观指数的空间粒度响应主要呈现增加、降低、波动和无明显规律变化的趋势，其中斑块密度、最大斑块面积等指数对斑块形状和大小的变化敏感，而多样性指数的粒度变化敏感度较低；部分指数如斑块密度、边缘密度、周长面积分维数等对空间幅度的变化并不敏感，而最大斑块面积、景观形状指数、散布与并列指数、分离度等指数对空间幅度的变化敏感，适合进行不同幅度适宜粒度阈值的推定；边缘密度、平均斑块大小、景...

Suitable granularity and response of multi-scale landscape in low mountain and hilly area of the Three Gorges Reservoir

Landscape patterns have typically spatial heterogeneity and scale dependence; thus it needs to span multiple scales in landscape pattern analysis. Therefore, the spatial granularity is crucial in scale aggregation. In the hilly area of the Three Gorges Reservoir area, patches are fragmented, landscape complexity is significant, and landscape pattern characteristics and scale variation are still to be clarified. In this study, we focused on Zigui County in the Three Gorges Reservoir, set 23 grain size gradient levels within 1-400 m, quantitatively evaluated the landscape pattern index and fitting function of the county, towns, and small basins at different amplitudes to explore more margin space landscape pattern index characteristics associated with the change in particle size. Based on the grain size effect characteristics of the landscape index and the curve characteristics (maximum curvature point and extreme value point) of the fitting function, a suitable grain size threshold of the mountain landscape pattern index with different amplitudes was determined, so as to reveal the complexity and variability of mountain landscape structure in the hilly region of the Three Gorges Reservoir. The responses of different landscape indices to the changes in spatial granularity and spatial amplitude differed. The responses of different landscape indices to the changes in spatial granularity can be summarized as four trends:increase, decrease, fluctuation, and no obvious regular change. Patch Density and the Largest Shape Index are sensitive to changes in patch shape and size, whereas the sensitivity of Diversity Index to the change in grain size is inapparent. Patch Density, Edge Density, and Perimeter-Area Fractal Dimension are not sensitive to the change in spatial amplitude, whereas the Largest Patch Index, Landscape Shape Index, Interspersion and Juxtaposition Index, and Splitting Index are sensitive to the change in spatial amplitude and are suitable for prediction of the threshold of appropriate grain size for different amplitudes. Landscape indices such as Edge Density, Mean Patch Size, Landscape Shape Index, Contagion Index, and Percentage of Like Adjacency can be highly fitted to the curve function, and the curve characteristics of the fitting function can be used to identify the turning point of the change in the landscape index and reduce the difference in artificial judgment. Based on the changing trends and turning points of the landscape indices of 6072 raster data of 251 small basins, 12 townships, and Zigui County, the first-scale granularity threshold for landscape pattern analysis of small basins, towns, and Zigui County is 3 m, 4 m, and 7 m, while the second-scale particle size threshold is 50 m, 100 m, and 100 m, respectively, integrating the turning points of each landscape index. Thus, the smaller the spatial amplitude, the smaller the suitable grain size for landscape pattern analysis.