SDSS-IV MaNGA DR17 Principcal Component Analysis spaxel classifcations

The zip files contains 10120 fits.gz files and two Python .p files. The files spaxel_properties_master_DR17.p and elliptical_radii_params_DR17.p contain all of the PCA values of spaxels from all galaxies and all the fraction of spaxels of a particular type in an ellipse (so looping over all of the maps to get this information is not necessary). There is one map per MaNGA galaxy. The data structure of each fits.gz file is: HDU 0: [image] primary header from the DAP MAPS file. HDU 1: [image] 'PC1' - PC1 amplitude. HDU 2: [image] 'PC2' - PC2 amplitude. HDU 3: [image] 'PC3' - PC3 amplitude. HDU 4: [image] 'PC1ERR' - PC1 error. HDU 5: [image] 'PC2ERR' - PC2 error. HDU 6: [image] 'PC3ERR' - PC3 error. HDU 7: [image] qualmask – Mask applied to PC1 map, has mask=(snr_4000A.T < 4.) | (pc1_map_reshaped.T < -10.) | (nocov) | (lowcov) | (donotuse) | (deadfiber) | (forestar). i.e. it excludes low S/N spaxels, weird PCA values and bad spaxels. HDU 8: [image] 'snr4000A' - Median signal-to-noise in the 4000A break region. HDU 9: [image] 'norm' - normalisation of the spectrum in the PCA. Used for reconstruction of the spectrum. HDU 10: [image] 'class_map' - map of PCA classifications. 1=quiescent, 2=star-forming, 3=starburst, 4=green valley, 5-post-starburst, 0=unclassified (do not use) HDU 11: [image] 'spx_bin_mask' - Mask accounting for identical values in a bin (see below for more details). The PCA code (see https://github.com/KateRowlands/MaNGA-PCA, Rowlands et al. 2018, based on Wild et al. 2007) is run on the HYB10-MILESHC-MASTARSSP cubes, where the stellar continuum is binned but the emission line measurements are done on the unbinned spectra (see SDSS DR17 DAP documentation for more details). In these maps the spectra in each stellar continuum bin are identical, so the PC amplitudes are identical. The analysis is done in this way to preserve the shape of the maps for comparing to other quantities. The identical nature of spaxels in the same bin needs to be accounted for in some analysis e.g. those which count spaxels of a certain PCA class. The spx_bin_mask accounts for this double counting by providing a mask which has the central spaxel in the Voronoi bin set to 1. For spaxels with unique PCA values, set spx_bin_mask==1. If plotting 2D maps of the PCA classes then spx_bin_mask should not be applied otherwise there will be gaps in the maps. To flag out poor quality spaxels, reject anything with snr4000A < 4, although different S/N cuts may be applied depending on your science case. Furthermore, the PCA parameters are affected by dust. PCA classifications of PSBs in regions with visible dust lanes e.g. in edge-on and or/ dusty galaxies should be closely examine by hand to ensure robustness. Values of -99 and 99 indicate no data or bad data and should be excluded.本压缩包包含10120个fits.gz格式文件与2个Python格式.p文件。文件"spaxel_properties_master_DR17.p"与"elliptical_radii_params_DR17.p"收录了所有星系的空间像素（spaxel）的全部主成分分析（PCA）值，以及椭圆内特定类型空间像素的占比——因此无需遍历所有数据图即可获取该统计信息。每个MaNGA星系对应一张专属数据图。 单份fits.gz文件的数据结构如下： HDU 0：[图像] 来自数据处理管线（DAP, Data Analysis Pipeline）MAPS文件的主表头； HDU 1：[图像] "PC1"——PC1振幅； HDU 2：[图像] "PC2"——PC2振幅； HDU 3：[图像] "PC3"——PC3振幅； HDU 4：[图像] "PC1ERR"——PC1误差； HDU 5：[图像] "PC2ERR"——PC2误差； HDU 6：[图像] "PC3ERR"——PC3误差； HDU 7：[图像] qualmask——应用于PC1图的掩膜，掩膜条件为：(snr_4000A.T < 4.) | (pc1_map_reshaped.T < -10.) | (nocov) | (lowcov) | (donotuse) | (deadfiber) | (forestar)，即该掩膜会剔除低信噪比空间像素、异常PCA值空间像素与无效空间像素； HDU 8：[图像] "snr4000A"——4000埃跃迁区的中位信噪比； HDU 9：[图像] "norm"——PCA所用光谱的归一化因子，用于光谱重构； HDU 10：[图像] "class_map"——PCA分类图，其中1代表宁静星系，2代表恒星形成星系，3代表星暴星系，4代表绿谷星系，5代表后星暴星系，0代表未分类（请勿使用）； HDU 11：[图像] "spx_bin_mask"——用于抵消同一分箱内重复计数的掩膜（详见下文说明）。 本数据集所用的PCA代码（详见https://github.com/KateRowlands/MaNGA-PCA，引自Rowlands等人2018年研究，基于Wild等人2007年的工作）运行于HYB10-MILESHC-MASTARSSP数据立方体，其中恒星连续谱已完成分箱处理，但发射线测量基于未分箱光谱（更多细节可参考SDSS DR17 DAP官方文档）。在上述数据图中，同一恒星连续谱分箱内的光谱完全一致，因此对应的PC振幅也完全相同。采用该分箱分析方式旨在保留数据图的原始形态，以便与其他物理量进行对比。部分分析场景（如统计特定PCA分类的空间像素数量）需要抵消同一分箱内的重复计数问题，"spx_bin_mask"通过将Voronoi分箱中的中心空间像素设为1来实现该功能：对于拥有唯一PCA值的空间像素，"spx_bin_mask"需设为1。若绘制PCA分类的二维分布图，则不应应用"spx_bin_mask"，否则数据图将出现空白间隙。 若要标记剔除低质量空间像素，可直接剔除"snr4000A < 4"的数据，但可根据具体科研需求调整信噪比截断阈值。此外，PCA参数会受到尘埃消光的影响。对于存在可见尘埃带的区域（如侧视星系或富含尘埃的星系）中的后星暴星系分类结果，需手动进行细致核查以确保分类的可靠性。值为-99和99的数据表示无有效数据或数据损坏，应予以剔除。