Analysis of Glacier Hazard Potentials By Knowledge-Based Remote Sensing Fusion for GIS Modeling (AGREG)

Snow, glaciers and permafrost in cold mountain areas such as the Swiss Alps are
especially sensitive to changes in environmental conditions due to their
proximity to melting conditions. In addition, mass wasting is most intensive in
those mountain areas with high relief energy. Environmental changes in high
mountain regions substantially influence the potential for glacial and
periglacial hazards. Ice- and moraine-dammed lakes represent a widespread
hazard potential closely related to glacier fluctuations. Magnitude and
frequency of ice avalanches from steep glaciers - in principle a normal
expression of mass exchange under such topographic conditions - are coupled
with stability conditions affected by glacier advance/retreat and, hence, with
long-term atmospheric impacts. Steep and unstable reservoirs of loose debris, a
potential source of debris flows, are often the result of glacier shrinkage. In
a similar way, changes in the stress regime due to vanishing glaciers lead to
potential destabilization of adjacent valley flanks.

Since the Alps are among the most densely populated high mountain areas in the
world, Switzerland is particularly impacted by glacial and periglacial hazards
but, on the other hand, also has an extensive and well-recognized tradition in
investigating such processes. A number of specific monitoring and modeling
studies related to single hazardous situations have been performed, mainly
based on recent catastrophes or imminent hazard situations. An urgent need
exists for area-wide modeling of glacier hazard potentials with a view to
establishing an integrated and adequate information base for planning and
detailed monitoring, but a corresponding systematic approach is, for the
present, still lacking.

The proposed project aims at closing this gap in several ways: Work Package
(WP) (1): By developing techniques for detection of glacier hazard potentials
based on optical spaceborne remote sensing data which rarely has been used to
date in Swiss glacier monitoring; multispectral analyses and multitemporal and
multiscale fusion will play a major role in this, with a special focus on
recent or upcoming high resolution sensors. WP (2): By integrating empirical
models for glacier hazard assessment into geographical information systems
(GIS) which have proven to be successful for hazard simulation but have not
been used yet for determining glacier hazard potentials; GIS modeling
especially allows for the fusion of remote sensing and elevation data for
spatial (3D) analyses. To ensure high synergy, WPs (1) and (2) will be closely
related to the ongoing SNF project "The Swiss Glacier Inventory 2000" (SWI
2000) (no. 21-54073.98) and the international project "Global Land Ice
Monitoring from Space" (GLIMS). WP (3): By applying the methods from WPs (1)
and (2), an initial attempt will be undertaken to implement an area-wide model
for integrating glacier hazard potentials of extensive regions in the Swiss
Alps following a downscaling strategy with varying resolution and accuracy
levels, both with respect to data and to models. As hazard management in
Switzerland is the domain of local and regional authorities, the proposed
project does not aim at preparing detailed local hazard maps (Gefahrenkarten),
but rather will provide new remote sensing and modeling techniques for
decision support. It should demonstrate the usefulness of these techniques for
overview mapping (Gefahrenhinweiskarten) as a basis for decision-making and for
scenario simulations in connection with climate change effects. The efforts
made in this project will contribute to handle economically complex
mathematical and physical models and represent a decision basis for the
specific need of further detailed case studies. A further outcome will be a
documentation of historical glacier catastrophes in the Swiss Alps, which will
- among others - be used for model calibration and verification.

[Summary provided by Christian Huggel, University of Zurich.]

以瑞士阿尔卑斯山为代表的寒冷山地中的积雪、冰川与多年冻土（permafrost），由于其所处环境接近融点阈值，对环境变化尤为敏感。此外，块体运动（mass wasting）在地形起伏强烈的山地中最为活跃。高海拔山地的环境变化，会显著影响冰川与冰缘灾害（periglacial hazards）的发生潜力。冰碛湖与冰坝湖是一类分布广泛的灾害隐患，其形成与冰川波动密切相关。陡峭冰川产生的冰崩规模与频次，本质上是该地形条件下物质交换的常规表现，其发生与冰川进退引发的冰体稳定性状态紧密相关，进而受长期大气变化的影响。松散碎屑组成的陡峭不稳定堆积体——即泥石流的潜在物源——通常是冰川退缩的产物。同理，冰川消失引发的应力场变化，会导致相邻谷坡出现潜在失稳风险。

阿尔卑斯山是全球人口密度最高的高海拔山地之一，因此瑞士深受冰川与冰缘灾害的影响；但与此同时，瑞士在该类灾害过程的研究方面拥有悠久且广受认可的传统。学界已开展多项针对单一灾害场景的专项监测与模拟研究，这些研究大多基于近期发生的灾害或迫在眉睫的灾害隐患。当前亟需开展全区域的冰川灾害潜力模拟研究，以构建一套完整且充足的信息基础，用于灾害规划与精细化监测，但目前仍缺乏相应的系统性研究方法。

本拟建项目旨在通过多路径填补这一空白：工作包1（Work Package, WP1）：开发基于光学星载遥感数据的冰川灾害潜力识别技术——这类数据在瑞士冰川监测中迄今极少应用；其中多光谱分析、多时间尺度与多尺度融合技术将发挥核心作用，研究将重点关注当前及未来的高分辨率遥感传感器。工作包2（Work Package, WP2）：将用于冰川灾害评估的经验模型集成至地理信息系统（Geographic Information System, GIS）中——这类系统在灾害模拟中已被证实效果良好，但尚未用于冰川灾害潜力评估；GIS建模尤其可实现遥感数据与高程数据的融合，从而开展空间（三维）分析。为保障高度协同性，WP1与WP2将紧密关联于正在实施的瑞士国家科学基金会（Swiss National Science Foundation, SNF）项目“2000年瑞士冰川名录（SWI 2000）”（编号21-54073.98），以及国际项目“全球陆地冰空间监测计划（Global Land Ice Monitoring from Space, GLIMS）”。工作包3（Work Package, WP3）：将应用WP1与WP2的技术方法，首次尝试基于“降尺度策略”构建瑞士阿尔卑斯山大范围区域的冰川灾害潜力集成模型，该策略在数据与模型层面均采用不同分辨率与精度等级的参数设置。鉴于瑞士的灾害管理工作由地方与区域当局主导，本项目并非旨在制作精细化的本地灾害风险图（Gefahrenkarten），而是将提供新型遥感与建模技术以支撑决策制定。本项目将验证这些技术在概览式灾害风险图（Gefahrenhinweiskarten）制作中的实用性，这类概览图可作为决策依据，同时可用于开展与气候变化影响相关的情景模拟。本项目的研究工作将助力复杂数学与物理模型的高效应用，同时可为后续开展精细化案例研究的特定需求提供决策依据。本项目的另一项成果将是瑞士阿尔卑斯山历史冰川灾害的整编记录，该记录可用于模型的率定与验证等工作。

[苏黎世大学克里斯蒂安·胡格尔（Christian Huggel）提供摘要]