Macquarie Island Terrain Class Model

A spatially explicit terrain class model (TCM) was produced of the full extent of Macquarie Island to determine areas on Macquarie Island with broadly the same microclimate and ensure that potential microrefugia could be identified. Non-correlated terrain variables likely to affect evapotranspiration rates were included in the model (elevation, surface curvature, topographic wetness index, solar radiation and topographically deflected wind speed). Terrain variables were derived from the Macquarie Island digital elevation model (DEM, Brolsma 2008), see Bricher et al. 2013 for methods. A fuzzy c-means clustering was applied to identify broadly similar climatic conditions and three different metrics (Xie-Beni Index, cluster stability, PCA) determined the optimum number of terrain clusters to be five. See Dickson et al. 2019 and Baker et al. in prep for full methodology. A ‘2nd Membership’ was also produced that only contains cells with at least double that of the second highest cluster centre membership association, which acknowledges that the highest membership ‘hard clustering’ contains some noise from the DEM and cells that may exist on the boundary of two memberships.ArcGIS shapefiles (.cpg, .dbf, .prj, .sbn, .sbx, .shp, .shx) in a zipped file:MI_TCM_2nd_Membership.zipMI_TCM_Highest_Membership.zipMore details in the referenced publication.

为确定麦夸里岛全域内微气候大致相同的区域并识别潜在微避难所，构建了空间显式地形分类模型（TCM）。模型纳入了可能影响蒸散速率的非相关地形变量，包括海拔、表面曲率、地形湿度指数、太阳辐射及地形偏转风速。地形变量源自麦夸里岛数字高程模型（DEM，Brolsma 2008），方法详见Bricher等（2013）。
应用模糊c均值聚类（fuzzy c-means clustering）识别大致相似的气候条件，通过三种指标（谢-贝尼指数（Xie-Beni Index）、聚类稳定性、主成分分析（PCA））确定最优地形聚类数为5。完整方法参见Dickson等（2019）及Baker等（待发表）。
此外，还生成了“第二成员资格”（2nd Membership）数据集，仅包含聚类中心成员资格关联度至少为第二高值两倍的单元；这一处理旨在规避最高成员资格“硬聚类”（hard clustering）中存在的DEM噪声及可能位于两个成员资格边界的单元。

压缩包内包含ArcGIS形状文件（.cpg、.dbf、.prj、.sbn、.sbx、.shp、.shx）：
MI_TCM_2nd_Membership.zip
MI_TCM_Highest_Membership.zip

更多细节参见引用文献。