青海省（含三江源国家公园）藏羚羊适宜生境1km空间分布图(2019年)

青海省境内内且包括三江源国家公园范围内藏羚羊适宜生境空间分布栅格图。基于野生动物本底调查或者查阅文献获得的物种分布位点；收集4大类77小类环境变量，包括气候变量（19个生物气候因子、月降水量、月平均最低温度、月平均最高温度、平均温度）；地形变量（海拔、坡度、地形曲率）；植被变量（植被归一化指数、植被类型、植被覆盖度等）；人为干扰变量（人口分布距离、道路分布距离等）。删减在1 km内出现的多个重复分布点，剔除相关性较高（| r |≥0.8）的环境变量，只选择高相关性中一个具有更多生态学含义的环境变量用于模型的构建。在Maxent建模中，选取特征分类参数中5种参数的组合模式及1~5每间隔0.5设置不同的β值（正规化系数）选取响应曲线较光滑且AICc、BIC得分最低的参数组合。分别以75%和25%位点数据为训练集和测试集，选择刀切法研判环境变量贡献率，设置10次重复运算次数，采用AUC值（0.5-1）检验结果，以0.85为临界值，高于0.85即可后续分析。使用ArcGIS的重分类功能将分布概率（P）划分为4个等级：P∈[0.6，1]为物高适宜区，P∈[0.4，0.6）为中适宜区；P∈[0.2，0.4）为低适宜区；P∈[0，0.2）为不适宜区。

Spatial distribution raster map of suitable habitats for Tibetan antelopes within Qinghai Province, including the scope of Sanjiangyuan National Park. The species distribution sites were obtained through wildlife baseline surveys or literature reviews. A total of 77 subcategories under 4 major categories of environmental variables were collected, including: 1) Climatic variables: 19 bioclimatic factors, monthly precipitation, monthly mean minimum temperature, monthly mean maximum temperature, and mean temperature; 2) Topographic variables: elevation, slope, and terrain curvature; 3) Vegetation variables: Normalized Difference Vegetation Index (NDVI), vegetation type, vegetation coverage, etc.; 4) Anthropogenic disturbance variables: distance to population distribution, distance to road distribution, etc. Duplicate distribution points occurring within a 1 km radius were removed, and environmental variables with high pairwise correlation (|r| ≥ 0.8) were excluded. Only the environmental variable with stronger ecological significance among the highly correlated ones was selected for model construction. In the Maxent modeling workflow, combinations of 5 feature classification parameters and β values (regularization coefficients) ranging from 1 to 5 at 0.5 intervals were tested. The optimal parameter combination was selected based on the smoothest response curves and the lowest Akaike Information Criterion corrected (AICc) and Bayesian Information Criterion (BIC) scores. The distribution sites were randomly split into a training set (75%) and a test set (25%). The Jackknife test was employed to estimate the relative contribution of each environmental variable, with 10 replicate runs configured. Model performance was evaluated using the Area Under the Receiver Operating Characteristic Curve (AUC) value, which ranges from 0.5 to 1. A critical threshold of 0.85 was set, and models with AUC > 0.85 were reserved for subsequent analyses. The reclassification tool in ArcGIS was used to divide the habitat distribution probability (P) into four grades: P ∈ [0.6, 1] as highly suitable habitat, P ∈ [0.4, 0.6) as moderately suitable habitat, P ∈ [0.2, 0.4) as lowly suitable habitat, and P ∈ [0, 0.2) as unsuitable habitat.

搜集汇总

背景与挑战

背景概述

该数据集为2019年青海省及三江源国家公园藏羚羊适宜生境的1km空间分布栅格图，通过Maxent模型结合多类环境变量分析生成，划分了高、中、低及不适宜四个等级区域，适用于生态保护与研究。