Sedimentology and architecture of ice stream beds

在过去的250万年里，加拿大多次被大冰盖覆盖，留下了各种各样的冰川地貌和景观遗产，这些遗产仍然没有得到很好的理解或绘制。冰川地质学的学科已经重新充满活力的发现，冰盖（现代和古代）不是缓慢流动的圆顶与稳定状态的冰流速，而是包含许多快速流动的冰流，这是长达1000公里，150公里宽的冰走廊，速度高达每年10公里。这一发现对北美和其他地方的冰川景观和沉积物的起源提出了新的问题，具有重要的应用意义，但还有待于在适当的冰盖范围内进行研究。在北美和斯堪的纳维亚半岛的最后一个劳伦蒂德和科迪勒拉冰盖中，有300多条前冰流在其河床上留下了数百万个流线型地貌（鼓形，巨型冰川线理），这些地貌由沉积物（沉积物的软床）或岩石（硬床，例如，加拿大盾）或两者的组合。到目前为止，对这些地貌的起源还没有达成共识。现在，利用提议者及其小组开发的创新机器学习技术来分析非常大的数字数据集（例如，35 Tb）的高分辨率图像，如LiDAR。这揭示了迄今为止看不见的细节，表明鼓和MSGLs组成了一个床形连续体，并随着冰盖工程师的床，以降低摩擦阻力和流动速度而演变。研究将参考人造表面如何在摩擦下演变的相关模型，以及断层和大型滑坡上流线型表面的研究，其中类似的床形在不同的尺度上发展。在项目结束时，我们将准确地绘制北美和芬兰的冰流床，其规模和详细程度是迄今无法实现的，并建立了流线型床型演变和地下地质的形成沉积学模型。冰盖模型制作者、研究现代冰流（冰流床无法进入）的冰川学家和需要冰流下矿化碎片侵蚀和迁移信息的矿产勘探行业，以及加拿大人口稠密、建在古冰流床上的城市地区的地下水资源管理人员都需要这一信息。拟议的研究将深入了解冰下过程和大型大陆冰盖下的碎片通量，并将为工业界和学术界培训机器学习和高分辨率地形数据集空间分析方面的高素质学生。长期目标是绘制一幅新的加拿大冰川图，以取代50年前使用低分辨率航空照片绘制的冰川图。

Canada was covered by large ice sheets many times in the last 2.5 million years leaving a diverse heritage of glacial landforms and landscapes which are still not well understood or mapped. The discipline of glacial geology has been re-energized by the finding that ice sheets (modern and ancient) were not slow-flowing domes with steady-state ice flow velocities throughout, but instead contained many fast-flowing ice streams which are corridors of ice as much as 1000 km long and 150 km wide with velocities as high as 10 km a year. This discovery poses new questions about the origins of glaciated landscapes and sediments in North America and elsewhere and has important applied significance yet to be fully evaluated by research on the appropriate ice sheet-wide scale. More than 300 former ice streams within the last Laurentide and Cordilleran ice sheets in North America and in Scandinavia left many millions of streamlined landforms on their beds (drumlins, megascale glacial lineations) composed of either sediment (a soft bed of sediment) or rock (a hard bed e.g., Canadian Shield) or combinations of the two. To date there is no consensus on the origin(s) of these landforms. Detailed mapping is now attainable for the first time, using innovative machine learning techniques developed by the proponent and his group to analyse very large digital data sets (e.g., 35 Tb) of high-resolution imagery such as LiDAR. This reveals hitherto unseen detail that shows that drumlins and MSGLs comprise a bedform continuum and evolve as the ice sheet engineers its bed to lower frictional drag and flow faster. Research will be informed by relevant modelling of how man-made surfaces evolve under friction and by studies of streamlined surfaces on faults and large landslides where analogous bedforms are developed on various scales. At the conclusion of the project we will have accurately mapped ice stream beds  in North America and in Finland at a scale and level of detail hitherto unachievable, and established formative sedimentological models for the evolution and subsurface geology of streamlined bedforms. This information is required by ice sheet modelers, glaciologists working on modern ice streams (where their beds are inaccessible) and the mineral exploration industry requiring information on the erosion and transport of mineralized debris under ice streams, and by groundwater resource managers in densely-populated urban areas in Canada that are built on former beds of paleo-ice streams. The proposed research will generate insights into subglacial processes and debris fluxes below large continental ice sheets and will train highly-qualified students for industry and academia in machine learning and the spatial analysis of high-resolution topographic data sets that are now becoming increasingly available. The long term objective is a new Glacial Map of Canada to replace that which was last completed 50 years ago using low resolution air photographs.