Terrestrial laser scanner data covering the summit craters of Láscar Volcano, Chile

Georeferencing (local coordinate system) In total, four TLS scans were acquired on two days in November 2013 (two at each day to overcome shadowing effects). The two point clouds from each view point were combined using tie points, i.e. reflectors that were placed in the field, and the RiSCAN Pro Software (http://www.riegl.com). For the two point clouds from day 1, we achieved a standard deviation of 0.0023 m using 6 tie points, while for the two point clouds acquired on day 2 we reached a standard deviation of 0.0052 m using 3 tie points. In addition to the TLS measurement, the reflectors’ positions were also measured using a total station. This additional data allowed us to 1) orientate each of the two point clouds to a local geodetic reference frame in the XY plane using a 3D affine transformation with a remaining RMSE of ∼1 cm and 2) estimate the orientation about Z and the full translation parameters using hand-held GPS coordinates of a common point and the individual tie points. Following this procedure we produced a combined point cloud of all four TLS scans in a local geodetic reference frame. Georeferencing (global coordinate system) In order to derive the coordinates of the TLS point cloud in a global coordinate system, we used the open-source software Minuit2 5.18/00 which was developed at CERN (James and Winkler, 2004 and references therein). This tool finds the minimum value of multi-parameter functions and was in our case employed to find the minimum root mean square residuals (in elevation) between the TLS point clouds and a reference DEM featuring a 1 m pixel spacing that was calculated from tri-stereo optical Pléiades-1 satellite imagery. When applying this minimization technique, the data are transferred to the same coordinate system as the reference data (WGS 1984 UTM Zone 19 South). In a first step, we minimized the two TLS point clouds from the two different acquisition dates separately. We masked out areas from the Pléiades reference DEM that we know are very different when compared to the TLS point data. For instance, areas along the steep crater walls are interpolated to a high degree in the Pléiades DEM, while the scanner-facing crater walls are expected to have comparably precise point values in the TLS dataset. Thereafter, we combined the TLS point clouds and ran another Minuit RMSE minimization onto the masked Pléiades DEM.

地理配准（局部坐标系）
本研究于2013年11月分两日开展地面激光扫描（Terrestrial Laser Scanning, TLS）数据采集，每日完成2站扫描以规避阴影干扰。每站点云数据均通过布设的野外反射靶标（即同名控制点），结合RiSCAN Pro软件（http://www.riegl.com）完成拼接融合。首日采集的2站点云通过6个同名控制点配准后，标准差为0.0023 m；次日采集的2站点云通过3个同名控制点配准后，标准差为0.0052 m。除地面激光扫描外，研究团队还使用全站仪测定了所有反射靶标的空间坐标。该靶标坐标数据支持完成两项配准工作：1）通过三维仿射变换将单日的2站点云统一至XY平面内的局部大地参考框架，配准后剩余均方根误差（Root Mean Square Error, RMSE）约为1 cm；2）利用公共点的手持GPS坐标与各同名控制点，解算绕Z轴的旋转参数与完整平移参数。通过上述流程，最终得到统一至局部大地参考框架下的4站地面激光扫描拼接点云。

地理配准（全局坐标系）
为将地面激光扫描点云转换至全局坐标系，研究团队采用了欧洲核子研究中心（CERN）开发的开源软件Minuit2 5.18/00（James与Winkler，2004及相关参考文献）。该工具可实现多参数函数的最小值寻优，本研究中用于求解地面激光扫描点云与参考数字高程模型（Digital Elevation Model, DEM）之间的高程方向最小均方根残差。该参考DEM空间分辨率为1 m，由三视立体光学昴宿星-1（Pléiades-1）卫星影像反演得到。通过该最小化算法，可将点云数据转换至与参考数据一致的坐标系——WGS 1984 UTM南19区（WGS 1984 UTM Zone 19 South）。研究首先分别对两日采集的2站地面激光扫描点云进行最小化配准。研究人员对昴宿星-1参考DEM进行掩膜处理，剔除与地面激光扫描点云数据差异显著的区域。例如，陡峭的陨石坑壁区域在昴宿星-1 DEM中存在较高程度的插值误差，而面向扫描仪的陨石坑壁在地面激光扫描数据中则具备更高的点位精度。随后将所有地面激光扫描点云进行拼接，并基于掩膜后的昴宿星-1参考DEM再次使用Minuit工具进行均方根误差最小化配准。