Abstract Object-based image analysis and supervised machine learning on a digital elevation model derived from airborne laser scanning data were conducted to assess the aggregate raw material potential over the Kemijarvi hummocky moraine field in Finnish Lapland. Geomorphometric analysis revealed surface stoniness to be diagnostic of moraine hummocks. It was utilized to detect true moraine hummocks from among the potential hummocks delineated using object-based image analysis with logistic regression of curvature features derived from the digital elevation model (area under the receiver operating characteristics curve, AUC = 0.77). Based on the electrical-sedimentary anisotropy data, the Kemijarvi moraine is polygenetic such that the ridges are diagonally and erosionally fluted (from sector 280–290°) on their surfaces, while the ridge cores are deposited from the NW (from sector 310–340°). The cores are composed of infill gravels and crudely stratified diamict sequences. We consider the core of the Kemijarvi hummocky moraine field to represent pre-existing (Early Weichselian) subglacial ridges rather than being generated through monogenetic (Late Weichselian) compressional shearing and stacking, frozen-bed extension, or bed ribbing instability mechanisms, as presented for ribbed moraines that are transverse to ice flow. Recently, evidence has been presented of multiple slip events (earthquakes with moment magnitudes Mw ≈ 6.7–7.0) up-ice from Kemijarvi. We therefore propose that the genesis of the hummocks may be associated with past (Early Weichselian) fault instability within glacio-isostatic adjustment(s), which was able to generate subglacial water bodies (possibly lakes) and saturated materials that were forced towards the SE. Based on this new geomorphological evidence from mapping using a digital elevation model derived from airborne laser scanning data and azimuthal conductivity surveys combined with the reinterpretation of previously presented geophysics and sedimentological observations, we suggest the Kemijarvi hummocky moraine field to be reconsidered as the Kemijarvi subglacial mass-flow field, which holds significant potential for construction aggregate materials.

中文翻译：

---

LiDAR 数据和沉积物各向异性的模式识别提倡芬兰北部 Kemijärvi 丘陵冰碛场的多基因冰下质量流起源

摘要 对源自机载激光扫描数据的数字高程模型进行了基于对象的图像分析和监督机器学习，以评估芬兰拉普兰 Kemijarvi 丘陵冰碛油田的综合原材料潜力。地貌测量分析显示表面石质可以诊断冰碛丘陵。它被用来从使用基于对象的图像分析描绘的潜在小丘中检测真正的冰碛小丘，该分析具有来自数字高程模型的曲率特征的逻辑回归（接收器操作特征曲线下的面积，AUC = 0.77）。根据电沉积各向异性数据，Kemijarvi 冰碛是多基因的，因此山脊在其表面上呈对角线和侵蚀凹槽（从扇区 280-290°），而脊核则从西北方向沉积（从扇区 310-340°）。岩心由充填砾石和粗分层混叠层序组成。我们认为 Kemijarvi hummocky 冰碛场的核心代表预先存在的（早期 Weichselian）冰下山脊，而不是通过单基因（晚 Weichselian）压缩剪切和堆积、冻床延伸或床肋不稳定机制产生的，如所提出的横向于冰流的棱纹冰碛。最近，有证据表明 Kemijarvi 冰上发生多次滑动事件（矩震级 Mw ≈ 6.7-7.0 的地震）。因此，我们提出小丘的起源可能与冰川均衡调整中过去（早期魏克塞尔）断层的不稳定性有关，它能够产生冰下水体（可能是湖泊）和被迫流向东南的饱和物质。基于使用从机载激光扫描数据和方位电导率调查得出的数字高程模型进行测绘的这一新地貌证据，并结合对先前提出的地球物理学和沉积学观测的重新解释，我们建议将 Kemijarvi 丘陵冰碛场重新考虑为 Kemijarvi 冰下物质-流场，在建筑骨料方面具有巨大的潜力。