BurnMan: A lower mantle mineral physics toolkit
BurnMan: A lower mantle mineral physics toolkit
复制标题
DOI：
10.1002/2013gc005122
复制
发表时间：
2014-04-01
影响因子：
3.5
通讯作者：
Unterborn, Cayman
中科院分区：
文献类型：
--
作者：
Cottaar, Sanne;Heister, Timo;Unterborn, Cayman
我们提出BurnMan，一个开源的矿物物理工具箱，通过求解状态方程（EoS）来确定下地幔中指定成分的弹性特性。该工具箱用Python编写，可用于评估新矿物物理数据或地球动力学模型的地震速度，并作为地幔成分反演的正演模型。用户可以从为下地幔提供的矿物列表中定义成分，或者轻松地包括自己的成分。BurnMan为EoS和多相平均方案提供了方法上的选择。结果可以直观地或定量地与观测到的地震模型进行比较。示例用户脚本显示了如何完成这些步骤。本文包括几个例子实现BurnMan：首先，我们基准的计算检查的正确性。其次，我们排除了两个陷阱，在EoS建模：使用一个不同的EoS比一个用来推导矿物物理参数或使用一个不正确的平均方案。这两个陷阱导致了文献中关于下地幔成分和温度的错误结论。我们进一步说明，拟合弹性速度单独或联合导致不同的Mg/Si比的下地幔。然而，我们发现，在矿物物理的不确定性，热解组合物可以匹配PREM非常好。最后，我们发现，具体的输入参数的不确定性，结果在地震速度的幅度和梯度的变化相当大的量。
We present BurnMan, an open-source mineral physics toolbox to determine elastic properties for specified compositions in the lower mantle by solving an Equation of State (EoS). The toolbox, written in Python, can be used to evaluate seismic velocities of new mineral physics data or geodynamic models, and as the forward model in inversions for mantle composition. The user can define the composition from a list of minerals provided for the lower mantle or easily include their own. BurnMan provides choices in methodology, both for the EoS and for the multiphase averaging scheme. The results can be visually or quantitatively compared to observed seismic models. Example user scripts show how to go through these steps. This paper includes several examples realized with BurnMan: First, we benchmark the computations to check for correctness. Second, we exemplify two pitfalls in EoS modeling: using a different EoS than the one used to derive the mineral physical parameters or using an incorrect averaging scheme. Both pitfalls have led to incorrect conclusions on lower mantle composition and temperature in the literature. We further illustrate that fitting elastic velocities separately or jointly leads to different Mg/Si ratios for the lower mantle. However, we find that, within mineral physical uncertainties, a pyrolitic composition can match PREM very well. Finally, we find that uncertainties on specific input parameters result in a considerable amount of variation in both magnitude and gradient of the seismic velocities.