DATA of "Resolving the 2D temporal evolution of subglacial water flow with dense seismic array observations."

The data set contains all data presented in the paper: Observing the subglacial hydrology network and its dynamics with a dense seismic array published in PNAS ( https://doi.org/10.1073/pnas.2023757118 ) See our online presentation of this dataset: https://meetingorganizer.copernicus.org/EGU2020/EGU2020-10710.html. The present data and code concerns the source location obtained with matched-field-processing analysis and the hydraulic potential calculation (Shreve, R. L. Movement of Water in Glaciers. J. Glaciol. 11, 205–214 (1972)). We perform source location over 1-sec long signal segment of the vertical component only. We filter the signal within the [3-7] Hz frequency range and coherently apply the MFP each 0.1 Hz within this range. To maximize our algorithm efficiency and minimize computational costs we use a gradient-based minimization algorithm (Nelder-Mead optimization) to converge to the best match between the trial and the observed phase delays rather than an exhaustive grid-search exploration. The convergence criterion is reached when the variance of values obtained over the last 5 iterations of the optimization is smaller than 1e-2 with a maximum of 3000 iterations. Our 29 different starting points used for optimization are located 250 m below the glacier surface and they uniformly cover an area of 800 x 800 m2 centered on the array. We set the initial velocity to 1800 m.sec -1. The 29 punctual locations found per signal segment (1 sec) after convergence are located all in the same place if a clear global convergence exists (i.e. high MFP output) or at up to 29 different locations if up to 29 local minima exist (i.e. low MFP output). Timeseries of physical quantities can be found here https://doi.org/10.5281/zenodo.3701520 Spatial observations acquired during the same period can be found here https://doi.org/10.5281/zenodo.3971815 The RESOLVE project has been supported by a grant from LabEx OSUG@2020 (Investissement d’avenir – ANR10LABX56) and by the IDEX Université Grenoble Alpes. Most of the computations presented in this paper were performed using the GRICAD infrastructure (https://gricad.univ-grenoble-alpes.fr), which is supported by Grenoble research communities, and with the CiGri tool (https://github.com/oar-team/cigri) developed by Gricad, Grid5000 (https://www.grid5000.fr) and LIG (https://www.liglab.fr/). You can find more information on the method and seismic dataset used in this paper here: https://zenodo.org/deposit/5645545

本数据集包含发表于《美国国家科学院院刊》（PNAS，https://doi.org/10.1073/pnas.2023757118）的论文《利用密集地震阵列观测冰下水文网络及其动力学》中的全部数据。该数据集的在线展示页面为：https://meetingorganizer.copernicus.org/EGU2020/EGU2020-10710.html。
本数据集及配套代码涉及通过匹配场处理（matched-field-processing）分析得到的震源位置与水力势计算（参考Shreve, R. L. 冰川内水的运动. 《冰川学杂志》, 11, 205–214 (1972)）。本次分析仅针对垂直分量的1秒时长信号片段开展震源定位工作：我们将信号滤波至[3-7] Hz频段内，并在该频段内以0.1 Hz为步长执行相干匹配场处理。为提升算法效率、降低计算成本，我们采用基于梯度的最小化算法（Nelder-Mead优化）来迭代收敛至试算信号与观测相位延迟间的最优匹配，而非使用穷举网格搜索。当优化过程最后5次迭代所得值的方差小于1e-2，且迭代次数不超过3000次时，即达到收敛准则。本次优化共设置29个不同的初始点，所有初始点均位于冰川表面下方250米处，且均匀覆盖以该地震阵列中心为原点的800×800平方米区域，初始速度设为1800 m·s⁻¹。每个1秒信号段经收敛后可得到29个离散震源位置：若存在明确的全局收敛（即匹配场处理输出值较高），则所有位置均重合于同一点；若存在多达29个局部极小值（即匹配场处理输出值较低），则可得到最多29个不同的位置。
相关物理量的时间序列数据可通过以下链接获取：https://doi.org/10.5281/zenodo.3701520。同期获取的空间观测数据可通过以下链接获取：https://doi.org/10.5281/zenodo.3971815。
本研究依托RESOLVE项目开展，得到LabEx OSUG@2020（未来投资计划——ANR10LABX56）与格勒诺布尔阿尔卑斯大学IDEX项目的资助。本文中的大部分计算工作依托GRICAD计算基础设施（https://gricad.univ-grenoble-alpes.fr）完成，该平台由格勒诺布尔研究共同体支持，同时使用了由Gricad、Grid5000（https://www.grid5000.fr）及LIG（https://www.liglab.fr/）开发的CiGri工具（https://github.com/oar-team/cigri）。更多关于本文所用方法与地震数据集的信息可通过以下链接查阅：https://zenodo.org/deposit/5645545。