Vulnerability of Antarctica’s ice shelves to meltwater-driven fracture

This data set includes the result presented in Lai et al. (2020), including the 125m-resolution fracture map, the spatial distribution of fracture depths and the initial flaw sizes calculated using linear elastic fracture mechanics (LEFM) according to the stress fields and ice-shelf thickness. We calculated the dimensionelss stress (Rxx_bar, defined in Lai et al. (2020)) which governs fracture behaviors. We obtained a dimensionless stress criteria which determines the ice-shelf areas vulnerable to hydrofracture if inundated with melt water (Rxx_bar >Rxx*_bar). Input data source: The input sources are "SUMER Antarctic Ice-shelf Buttressing, Version 1" (https://doi.org/10.5067/FWHORAYVZCE7; Fürst et al. (2016)), "MOA 2009" (https://doi.org/10.7265/N5KP8037; Haran et al. (2014), Scambos et al. (2007)), and "Bedmap2" (https://www.bas.ac.uk/project/bedmap-2/; Fretwell et al. (2013)). Data processing: The dimensionless stress Rxx_bar is calculated using the along-flow strain rates exx (calculated from the result of Fürst et al. (2016)), the viscosity calculated from surface temperature (from the regional climate model RACMO2.3p2), and the ice-shelf thickness (from Bedmap2). For di_dry, di_water, ds, exx, Rxx_bar, PSI, non-ice shelf areas (ocean or ice sheet) are marked NaN, using the same mask applied by Fürst et al. (2016). The code for training a neural-network to identify fracture locations is available at https://github.com/chingyaolai/Antarctic-fracture-detection Grid Specifications: All parameters except for "frac_loc_125m" are generated on the same grid used by Fürst et al. (2016) with a grid resolution of 1 km and grid dimensions (x: 5501 pixels and y: 5501 pixels). The x and y coordinate of the center of the upper-left pixel are -2,750,000 m and 2,750,000 m, respectively. The x and y coordinate are provided. "frac_loc_125m" is produced on the same grid as MOA 2009 (Haran et al. (2014)) with a grid resolution of 125 m and grid dimensions (x: 48333 pixels and y: 41779 pixels). The x and y coordinate of the center of the upper-left pixel are -3,174,450 m and 2,406,325 m, respectively. Parameters: frac_loc_125m: Fracture locations classified by the neural network on the MOA 2009 125m map. Fracture and non-fracture locations are denoted 1 and 0, respectively. No Unit. di_water: Depth of the initial flaws required to destabilize fractures fully filled with water (hydrofractures). Places marked 0 are where no hydrofractures can form. Unit [m] di_dry: Depth of the initial flaws required to form dry fractures. Places marked 0 are where no fractures can form. Unit [m] ds: Depth of the stable dry fractures. Places marked 0 are where no dry fractures can form. Places marked -9999 are where dry fractures are predicted to be unstable. Unit [m] exx: Along-flow strain rates calculated, according to Glen's law, with the along-flow stress and the viscosity factor derived by Fürst et al. (2016) from data assimilation with the Elmer/Ice ice flow model. Unit [1/year] Rxx_bar: Dimensionless extensional stress, defined as Rxx/(rhoi g H). The regions vulnerable to hydrofracture satisfie Rxx_bar > Rxx*_bar. No unit. PSI: Places marked 1 are the "passive shelf ice" identified by Fürst et al. (2016). Other locations are marked 0. No unit. frac_loc_1km: Parameter "frac_loc_125m" downscaled to 1km resolution (see Lai et al. (2020)). No unit. x: x coordinate. Unit [m] y: y coordinate. Unit [m]

本数据集包含Lai等人（2020）发表的研究成果，涵盖125米分辨率冰裂缝分布图、裂缝深度空间分布数据，以及依据应力场与冰架厚度，通过线性弹性断裂力学（linear elastic fracture mechanics, LEFM）计算得到的初始缺陷尺寸。本研究计算了控制冰裂缝发育行为的无量纲应力Rxx_bar（定义详见Lai等人（2020）），并得到了无量纲应力判据，用于判定当冰架被融水淹没时易发生水力破裂的冰架区域（即满足Rxx_bar > Rxx*_bar的区域）。
输入数据源：本数据集的输入数据来源于《SUMER Antarctic Ice-shelf Buttressing, Version 1》（https://doi.org/10.5067/FWHORAYVZCE7; Fürst等人，2016）、《MOA 2009》（https://doi.org/10.7265/N5KP8037; Haran等人，2014；Scambos等人，2007）以及《Bedmap2》（https://www.bas.ac.uk/project/bedmap-2/; Fretwell等人，2013）。
数据处理流程：无量纲应力Rxx_bar的计算依托于沿流应变率exx（由Fürst等人（2016）的研究结果推导得到）、由地表温度（来自区域气候模型RACMO2.3p2）计算得到的冰粘度，以及冰架厚度（来自Bedmap2）。针对di_dry、di_water、ds、exx、Rxx_bar、PSI这几个参数，本数据集采用与Fürst等人（2016）一致的掩膜，将非冰架区域（海洋或冰盖）标记为NaN。用于训练神经网络以识别冰裂缝位置的代码可于https://github.com/chingyaolai/Antarctic-fracture-detection获取。
网格规格说明：除"frac_loc_125m"外，其余所有参数均采用与Fürst等人（2016）一致的网格进行生成，该网格分辨率为1千米，网格维度为x方向5501个像素、y方向5501个像素。左上角像素中心的x坐标与y坐标分别为-2750000米与2750000米，数据集已提供x、y坐标信息。"frac_loc_125m"则采用与MOA 2009（Haran等人，2014）一致的网格进行生成，该网格分辨率为125米，网格维度为x方向48333个像素、y方向41779个像素，其左上角像素中心的x坐标与y坐标分别为-3174450米与2406325米。
参数说明：
"frac_loc_125m"：通过神经网络在MOA 2009的125米分辨率影像上分类得到的冰裂缝位置，其中裂缝区域与非裂缝区域分别以1和0标记，无单位。
"di_water"：完全充水裂缝（水力裂缝）失稳所需的初始缺陷深度，标记为0的区域无法形成水力裂缝，单位为米[m]。
"di_dry"：形成干裂缝所需的初始缺陷深度，标记为0的区域无法形成干裂缝，单位为米[m]。
"ds"：稳定干裂缝的深度，标记为0的区域无法形成干裂缝，标记为-9999的区域预测干裂缝处于不稳定状态，单位为米[m]。
"exx"：沿流应变率，依据格伦定律（Glen's law），结合沿流应力与由Fürst等人（2016）通过Elmer/Ice冰流模型数据同化推导得到的粘度因子计算得到，单位为1/年[1/year]。
"Rxx_bar"：无量纲拉张应力，定义为Rxx/(ρi g H)，满足Rxx_bar > Rxx*_bar的区域易发生水力破裂，无单位。
"PSI"：标记为1的区域为Fürst等人（2016）识别的“被动陆架冰（passive shelf ice）”，其余区域标记为0，无单位。
"frac_loc_1km"：将"frac_loc_125m"降尺度至1千米分辨率得到的参数（详见Lai等人，2020），无单位。
"x"：x坐标，单位为米[m]。
"y"：y坐标，单位为米[m]。

搜集汇总

背景与挑战

背景概述

该数据集专注于评估南极冰架对融水驱动裂缝的脆弱性，包含高分辨率（125米）的裂缝图、裂缝深度分布和基于线性弹性断裂力学计算的初始缺陷尺寸。数据集提供了多个关键参数，如无量纲应力Rxx_bar，用于识别易受水力压裂影响的区域，并支持使用神经网络进行裂缝检测，旨在帮助研究冰架稳定性和气候变化影响。