Geodetic mass balance of Mýrdalsjökull ice cap, 1999−2021: DEM processing and climate analysis

This repository gathers the data I used and produced during my master thesis at the University of Iceland from April to September 2022 with the financial support of the Landsvirkjun. The geodetic mass of Mýrdalsjökull, the fourth largest Icelandic ice cap, was investigated over the period 1999-2021. The untapped SPOT5 archive (2002−2015), the lidar data, the Pléiades imagery (2011−present), aerial photographs from 1999 [1] and the ArcticDEM dataset (2010−2019) [2] were used to create Digital Elevation Models (DEMs) of the ice cap. A pre-processing of the DEMs was first performed: co-registration, filtering and interpolation. Then, applying a Gaussian Process regression (GP) [3], a state-of-the-art method in DEM processing, a spatially and temporally continuous DEM dataset was created, in 15 x 15 m resolution and 1-month interval from 1999 to 2021. Volume and mass changes based on the synthetic GP-generated DEMs were computed and analyzed in 5-year and annual intervals between 1999 and 2019. A local analysis of three glacierized catchments of Mýrdalsjökull (southern catchment, northern catchment and Kötlujökull outlet) was also performed. Errors were estimated using the method from [4]. Tools from the following repositories were used: demcoreg (https://doi.org/10.5281/zenodo.5733347): DEM co-registration xdem (https://doi.org/10.5281/zenodo.4809698): uncertainties computation and DEM manipulation pyddem (https://pypi.org/project/pyddem/): Gaussian Process regression The complete master thesis can be accessed at : The repository contains the following data: 1 – DEM_coregistered All DEMs have been coregistered considering the Islandsdem v1.0 as a reference (atlas.lmi.is/dem) The DEM naming works as follow: Glaciername_DEM_date_sensor_resolution_zmae_projection_otherinformation.tif The files ending with *_filtered.tif (SPOT5, AerialPhotographs) have been filtered using the filtering combination described in 3.1.3. The files ending with *_mosaic.tif (SPOT5, Pléiades) are the result of the mosaicking of several DEMs. 2 – Shapefiles Outlines of Mýrdalsjökull in 1999 [1], 2003, 2010 and 2019 [6] Outlines of the three catchments (South, North and Kötlujökull) in 1999. Reference buffer around Mýrdalsjökull used to crop all DEMs to the same extent. Equilibrium Line Altitude (ELA) from 2004-10-05. 3 – Gaussian_Process_regression 2 netcdf files: the stack of DEMs (*_DEMstack_*) and the result of the Gaussian Process regression (*_GPregression_*). Both files were obtained thanks to pyddem tools. The Gaussian Process regression was run at a spatial resolution of 15 x 15m and a temporal resolution of 1 month, starting in January 1999 and ending in December 2022. 4 – Mass_balance_results 1 csv file containing mass balance results: Annual mass balance for the ice cap & the 3 catchments 4-year mass balance for the ice cap & the 3 catchments Results from the comparison with survey dates mass balance (Fig 11(a)) Results from the comparison with [3] (Fig 11(b)) Mass balance overview (Fig 11(c)) References [1] Belart, J., Magnússon, E., Berthier, E., Gunnlaugsson, Á. Þ., Pálsson, F., Aðalgeirsdóttir, G., Jóhannesson, T., Thorsteinsson, T., and Björnsson, H. (2020). Mass balance of 14 Icelandic glaciers, 19452017: spatial variations and links with climate. Frontiers in Earth Science, page 163. [2] Porter, C., Morin, P., Howat, I., Noh, M., Bates, B., Peterman, K., Keesey, S., Schlenk, M., Gardiner, J., et al. (2018). ArcticDEM. Harvard Dataverse, 1. [3] Hugonnet, R., McNabb, R., Berthier, E., Menounos, B., Nuth, C., Girod, L., Farinotti, D., Huss, M., Dussaillant, I., Brun, F., et al. (2021). Accelerated global glacier mass loss in the early twenty-first century. Nature, 592(7856):726731. [4] Hugonnet, R., Brun, F., Berthier, E., Dehecq, A., Mannerfelt, E. S., Eckert, N., and Farinotti, D. (2022). Uncertainty analysis of digital elevation models by spatial inference from stable terrain. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing. [5] Hannesdóttir, H., Sigurðsson, O., Þrastarson, R. H., Guðmundsson, S., Belart, J. M., Pálsson, F., Magnússon, E., Víkingsson, S., Kaldal, I., and Jóhannesson, T. (2020). A national glacier inventory and variations in glacier extent in Iceland from the Little Ice Age maximum to 2019. Jökull 2020: 1, 34. Acknowledgements Pléiades images were acquired at research price thanks to the CNES ISIS programme (http://www.isiscnes.fr). This study uses the lidar mapping of the glaciers in Iceland, funded by the Icelandic Research Fund, the Landsvirkjun research fund, the Icelandic Road Administration, the Reykjavík Energy Environmental and Energy Research Fund, the Klima- og Luftgruppen research fund of the Nordic Council of Ministers, the Vatnajökull National Park, the organization Friends of Vatnajökull, LMÍ, IMO, and the UI research fund. Dataset Attribution This dataset is licensed under a Creative Commons CC BY-NC 4.0 International License (Attribution-NonCommercial).

本数据集收纳了我于2022年4月至9月在冰岛大学攻读硕士学位期间所使用与生成的全部数据，本研究得到了Landsvirkjun（冰岛国家电力公司）的经费支持。本次研究针对冰岛第四大冰盖——米达尔斯冰原（Mýrdalsjökull）1999年至2021年间的大地测量质量变化展开了调查。

研究用到了未被充分利用的SPOT5档案影像（2002−2015）、激光雷达（lidar）数据、普莱赛（Pléiades）卫星影像（2011年至今）、1999年的航空摄影资料[1]以及ArcticDEM数据集（2010−2019）[2]，以此构建该冰盖的数字高程模型（Digital Elevation Models, DEMs）。

首先对所有DEMs开展预处理工作，包括共配准、滤波与插值。随后采用当前DEM处理领域的主流方法——高斯过程回归（Gaussian Process regression, GP）[3]，构建了时空连续的DEM数据集，其空间分辨率为15×15米，时间分辨率为1个月，覆盖时段为1999年至2021年。

基于该高斯过程回归生成的合成DEM数据集，我们计算并分析了1999年至2019年间以5年和年为单位的冰体体积与质量变化。此外，还针对米达尔斯冰原的三个冰川汇水区（南部汇水区、北部汇水区与科特卢约库尔（Kötlujökull）分支冰川）开展了局部尺度分析。误差估算采用文献[4]提出的方法。

本研究使用了以下开源工具库：
- demcoreg（https://doi.org/10.5281/zenodo.5733347）：用于DEM共配准
- xdem（https://doi.org/10.5281/zenodo.4809698）：用于不确定性计算与DEM操作
- pyddem（https://pypi.org/project/pyddem/）：用于高斯过程回归

完整硕士论文可于以下链接获取：

本数据集包含以下内容：
1. DEM_coregistered文件夹
所有DEM均以Islandsdem v1.0为参考进行了共配准（数据源地址：atlas.lmi.is/dem）。DEM命名规则如下：Glaciername_DEM_date_sensor_resolution_zmae_projection_otherinformation.tif。其中，以*_filtered.tif结尾的文件（对应SPOT5影像、航空摄影资料）已采用3.1.3节描述的滤波组合进行了滤波处理；以*_mosaic.tif结尾的文件（对应SPOT5影像、普莱赛影像）为多幅DEM拼接后的成果。

2. Shapefiles文件夹
- 1999年[1]、2003年、2010年与2019年[6]的米达尔斯冰原轮廓
- 1999年三个汇水区（南部、北部与科特卢约库尔）的轮廓
- 用于将所有DEM裁剪至统一范围的米达尔斯冰原周边参考缓冲区
- 2004年10月5日的平衡线高度（Equilibrium Line Altitude, ELA）

3. Gaussian_Process_regression文件夹
包含2个netcdf格式文件，分别为DEM堆叠文件（*_DEMstack_*）与高斯过程回归成果文件（*_GPregression_*），二者均通过pyddem工具生成。本次高斯过程回归的空间分辨率为15×15米，时间分辨率为1个月，覆盖时段为1999年1月至2022年12月。

4. Mass_balance_results文件夹
包含1个csv格式文件，其中存储了以下质量平衡结果：
- 冰原及三个汇水区的年际质量平衡
- 冰原及三个汇水区的4年尺度质量平衡
- 与实测日期质量平衡的对比结果（图11(a)）
- 与文献[3]结果的对比结果（图11(b)）
- 质量平衡总览（图11(c)）

参考文献
[1] Belart, J., Magnússon, E., Berthier, E., Gunnlaugsson, Á. Þ., Pálsson, F., Aðalgeirsdóttir, G., Jóhannesson, T., Thorsteinsson, T., Björnsson, H. (2020). 1945–2017年冰岛14座冰川的质量平衡：空间变异特征及其与气候的关联. 《地球科学前沿》(Frontiers in Earth Science), 页码163.
[2] Porter, C., Morin, P., Howat, I., Noh, M., Bates, B., Peterman, K., Keesey, S., Schlenk, M., Gardiner, J., 等. (2018). ArcticDEM. 哈佛数据文库, 1.0版.
[3] Hugonnet, R., McNabb, R., Berthier, E., Menounos, B., Nuth, C., Girod, L., Farinotti, D., Huss, M., Dussaillant, I., Brun, F., 等. (2021). 21世纪初全球冰川质量损失加速. 《自然》(Nature), 592(7856): 726–731.
[4] Hugonnet, R., Brun, F., Berthier, E., Dehecq, A., Mannerfelt, E. S., Eckert, N., Farinotti, D. (2022). 基于稳定地表空间推断的数字高程模型不确定性分析. IEEE 应用地球观测与遥感精选主题期刊(IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing).
[5] Hannesdóttir, H., Sigurðsson, O., Þrastarson, R. H., Guðmundsson, S., Belart, J. M., Pálsson, F., Magnússon, E., Víkingsson, S., Kaldal, I., Jóhannesson, T. (2020). 冰岛国家冰川名录及从小冰期鼎盛期至2019年的冰川范围变化. 《Jökull》2020: 1, 34.

致谢
本研究的普莱赛影像通过CNES ISIS项目以科研优惠价格获取（http://www.isiscnes.fr）。本研究使用的冰岛冰川激光雷达测绘数据，得到了冰岛研究基金、Landsvirkjun研究基金、冰岛道路管理局、雷克雅未克能源公司环境与能源研究基金、北欧部长理事会气候与空气集团研究基金、瓦特纳冰原国家公园、瓦特纳冰原之友组织、LMÍ、IMO以及冰岛大学研究基金的资助。

数据集归属
本数据集采用知识共享署名-非商业性使用4.0国际许可协议（Creative Commons CC BY-NC 4.0 International License）进行授权。