Accuracy of timescales retrieved from diffusion modeling in olivine: A 3D perspective
Accuracy of timescales retrieved from diffusion modeling in olivine: A 3D perspective
复制标题
DOI：
10.2138/am-2015-5163
复制
发表时间：
2015-10
影响因子：
3.1
通讯作者：
T. Shea;F. Costa;D. Krimer;J. Hammer
中科院分区：
文献类型：
--
作者：
T. Shea;F. Costa;D. Krimer;J. Hammer
摘要橄榄石中的扩散模拟是解决各种岩浆作用时间尺度的有效工具。实际的橄榄石地质速度测量应用采用从给定的3D晶体群中随机抽样的跨剖面的1D化学横断面，但从该过程中检索时间尺度的准确性和精确度并没有受到很好的限制。在这里，我们使用数值三维扩散模型的Fe-Mg评估和量化的不确定性与他们的一维同行。三维扩散模型建立使用简单和现实的橄榄石形态，并纳入扩散各向异性以及不同的分区风格。沿沿着理想或随机平面对3D模型晶体进行切片，用于进行1D模型和时间尺度比较。结果表明，从1D配置文件检索的时间尺度是高度不准确的，可以改变0.1-25的因素，如果不考虑扩散各向异性。即使校正了各向异性，由于晶体形状和切片效应，时间尺度仍然可以在真实3D扩散时间的0.2-10倍之间变化。描述了简单的晶粒选择程序，以减少计算和实际扩散时间之间的失配，并分别实现~5%和~15-25%的相对准确度和精密度。只要仔细选择颗粒，大约20个浓度分布图和相关的1D模型就足以达到这种精度。
Abstract Diffusion modeling in olivine is a useful tool to resolve the timescales of various magmatic processes. Practical olivine geospeedometry applications employ 1D chemical transects across sections that are randomly sampled from a given 3D crystal population, but the accuracy and precision with which timescales can be retrieved from this procedure are not well constrained. Here, we use numerical 3D diffusion models of Fe-Mg to evaluate and quantify the uncertainties associated with their 1D counterparts. The 3D diffusion models were built using both simple and realistic olivine morphologies, and incorporate diffusion anisotropy as well as different zoning styles. The 3D model crystals were sectioned along ideal or random planes, which were used to perform 1D models and timescale comparisons. Results show that the timescales retrieved from 1D profiles are highly inaccurate and can vary by factors of 0.1-25 if diffusion anisotropy is not taken into account. Even when anisotropy is corrected for, timescales can still vary between 0.2-10 times the true 3D diffusion time due to crystal shape and sectioning effects. Simple grain selection procedures are described to reduce the misfit between calculated and actual diffusion times, and achieve an accuracy and precision of ~5% and ~15-25% relative, respectively. Provided that the grains are carefully selected, about 20 concentration profiles and associated 1D models suffice to achieve this accuracy.