Thompson David R, Flannery David T, Lanka Ravi, Allwood Abigail C, Bue Brian D, Clark Benton C, Elam W Timothy, Estlin Tara A, Hodyss Robert P, Hurowitz Joel A, Liu Yang, Wade Lawrence A
1 Jet Propulsion Laboratory, California Institute of Technology , Pasadena, California.
2 Imaging Spectroscopy, Jet Propulsion Laboratory, California Institute of Technology .
Astrobiology. 2015 Nov;15(11):961-76. doi: 10.1089/ast.2015.1349.
A new generation of planetary rover instruments, such as PIXL (Planetary Instrument for X-ray Lithochemistry) and SHERLOC (Scanning Habitable Environments with Raman Luminescence for Organics and Chemicals) selected for the Mars 2020 mission rover payload, aim to map mineralogical and elemental composition in situ at microscopic scales. These instruments will produce large spectral cubes with thousands of channels acquired over thousands of spatial locations, a large potential science yield limited mainly by the time required to acquire a measurement after placement. A secondary bottleneck also faces mission planners after downlink; analysts must interpret the complex data products quickly to inform tactical planning for the next command cycle. This study demonstrates operational approaches to overcome these bottlenecks by specialized early-stage science data processing. Onboard, simple real-time systems can perform a basic compositional assessment, recognizing specific features of interest and optimizing sensor integration time to characterize anomalies. On the ground, statistically motivated visualization can make raw uncalibrated data products more interpretable for tactical decision making. Techniques such as manifold dimensionality reduction can help operators comprehend large databases at a glance, identifying trends and anomalies in data. These onboard and ground-side analyses can complement a quantitative interpretation. We evaluate system performance for the case study of PIXL, an X-ray fluorescence spectrometer. Experiments on three representative samples demonstrate improved methods for onboard and ground-side automation and illustrate new astrobiological science capabilities unavailable in previous planetary instruments.
Dimensionality reduction-Planetary science-Visualization.
新一代行星漫游车仪器，如被选入火星2020任务漫游车有效载荷的PIXL（行星X射线岩石化学仪器）和SHERLOC（用于有机物和化学物质的拉曼发光扫描宜居环境），旨在在微观尺度上原位绘制矿物学和元素组成图。这些仪器将生成大型光谱立方体，其中包含在数千个空间位置采集的数千个通道的数据，巨大的潜在科学产出主要受放置后获取测量所需时间的限制。下行链路之后，任务规划者还面临另一个瓶颈；分析人员必须快速解读复杂的数据产品，以便为下一个指令周期的战术规划提供信息。本研究展示了通过专门的早期科学数据处理来克服这些瓶颈的操作方法。在漫游车上，简单的实时系统可以进行基本的成分评估，识别感兴趣的特定特征，并优化传感器积分时间以表征异常。在地面上，基于统计的可视化可以使未校准的原始数据产品更易于解读，以用于战术决策。诸如流形降维等技术可以帮助操作人员一眼理解大型数据库，识别数据中的趋势和异常。这些车上和地面分析可以补充定量解释。我们以X射线荧光光谱仪PIXL为例评估系统性能。对三个代表性样本进行的实验展示了车上和地面自动化的改进方法，并说明了以往行星仪器所不具备的新的天体生物学科学能力。
降维 - 行星科学 - 可视化
