Data of two sites in Somalia underlying the paper: Multistability of model and real dryland ecosystems through spatial self-organization

For both sites, topographical data was retrieved from the ALOS World 3D 30m (AW3D30, v. 2.1) digital raster elevation model. This model describes the height above sea level (in m, rounded to the nearest integer), at a ground resolution of approximately 30m at the equator. The elevation data was preprocessed for the removal of artifacts by applying a global soft-thresholding on its dual tree complex wavelet transform. Specifically, we set a threshold of 0.9 on the first five dual-tree complex wavelet transform levels. From the preprocessed data, we calculated the slope gradient (in %) and slope aspect (in degrees). We first extracted square DEM windows of 33 by 33 cells (i.e. approximately 990m × 990m), centered on the image windows. We then applied a least squares fitting procedure of an unconstrained quadratic surface on the unweighted elevation values. This gave the following datasets. <br>

针对两个研究站点，地形数据均取自ALOS全球30米地形数据（ALOS World 3D 30m，简称AW3D30，版本2.1）数字栅格高程模型。该模型记录了海拔高度（单位：米，四舍五入至整数），在赤道处的地面分辨率约为30米。为去除伪影，我们对该高程数据的双树复小波变换（dual tree complex wavelet transform）执行全局软阈值处理以完成预处理。具体而言，我们在前五级双树复小波变换层级上设置了0.9的阈值。基于预处理后的数据，我们计算了坡度（单位：%）和坡向（单位：度）。我们首先以图像窗口为中心，提取尺寸为33×33个栅格单元的正方形数字高程模型（Digital Elevation Model，DEM）窗口（对应约990米×990米的范围）。随后，我们对未加权的高程值执行无约束二次曲面的最小二乘拟合流程。由此得到如下数据集。