Viewing COS Data

Learning Goals

This Notebook is designed to walk the user (you) through:

1. Reading in the data using Python

- 1.1. Investigating the Data - Basics

- 1.2. Reading in the x1d/x1dsum Main Data

- 1.3. The Association (asn) file

2. Displaying the data using common plotting techniques

- 2.1. Plotting an FUV Spectrum

2.1.1. Our First Plot

2.1.2. A Complex Look at the Entire FUV

2.1.3. Looking Closer at Parts of the FUV Spectrum with a Cautionary Tale

2.1.4. Reading and plotting the data with specutils (optional)

- 2.2. Plotting an NUV Spectrum

2.2.1. Examining the first-order spectrum

2.2.2. Examining the second-order spectrum

3. Making a quick assessment of the data with tools to bin and measure the data's SNR and resolution

- 3.1. Understanding and Using Data Quality Flags

- 3.2. Binning the Data

3.2.1. Bringing in Some Useful Functions for Binning

3.2.2. Binning the FUV Data

- 3.3. Calculating the Signal-to-Noise Ratio

3.3.1. Defining a Useful Function for Estimating SNR

3.3.2. Choosing a Region to calculate SNR

3.3.3. Calculating the SNR

If you're confident with Reading-in and Plotting data in Python, you may run the first few cells defining directories and importing modules and then click the link when it appears to go ahead and skip to Section 3.

0. Introduction

The Cosmic Origins Spectrograph (COS) is an ultraviolet spectrograph on-board the Hubble Space Telescope (HST) with capabilities in the near ultraviolet (NUV) and far ultraviolet (FUV).

This tutorial aims to prepare you to begin analyzing COS data of your choice by walking you through reading and viewing a spectrum obtained with COS, as well as obtaining a few diagnostic measurements of your spectrum.

Notes for those new to Python/Jupyter/Coding:

  • You will frequently see exclamation points (!) or dollar signs (\$) at the beginning of a line of code we are telling you to run. These are not part of the actual commands. The exclamation points tell a jupyter Notebook to pass the following line to the command line, and the dollar sign merely indicates the start of a terminal prompt.

We will import the following packages:

  • numpy to handle array functions (version $\ge$ 1.17)
    • We also insert a line telling numpy to ignore a certain type of warning about divide-by-zeros.
      • This is not always a good idea; seriously consider it before copying it into your code!
  • sys, os, and pathlib for managing system variables and paths.
  • fits for accessing FITS files
  • astropy.table Table for creating tidy tables of the data
  • astropy.units and astropy.visualization.quantity_support for dealing with units
  • matplotlib.pyplot for plotting data
  • astroquery.mast Mast and Observations for finding and downloading data from the MAST archive

Later on, we will import some functions from a local file,, which is installed as part of the same GitHub repository as this Notebook. If you do not see this Python file in this directory, you can find it here.

In [1]:
# This line causes matplotlib plots to appear in the Notebook 
 # instead of possibly showing up in separate windows
%matplotlib inline

# Import for: Manipulating arrays
import numpy as np

# Import for: Managing system variables and paths
import sys 
import os
from pathlib import Path

# Imports for: Reading in data
from import fits
from astropy.table import Table

# Import for: Plotting
import matplotlib.pyplot as plt

# Imports for: Dealing with units and plotting them
from astropy import units as u
from astropy.visualization import quantity_support

# Imports for: Downloading data from archive
from astroquery.mast import Mast
from astroquery.mast import Observations
from astroquery.mast import Catalogs

The brief cell below will prevent numpy from giving us a warning when we divide elements of an array by zero. This warning is not helpful for this Notebook, but you should be careful before suppressing warnings in general.

In [2]:
# Our arrays will encounter divide by 0 warnings, which is OK here, 
# but we want to suppress the warning text with the next code line.
   # DO NOT COPY the next line into your own code without careful consideration

We will also define a few directories in which to place our data and plots, as well as a few colors we will use in plots later.

In [3]:
# These will be important directories for the Notebook

datadir = Path('./data')
outputsdir = Path('./output/')
plotsdir = Path('./output/plots')
# Make the directories if they don't exist
datadir.mkdir(exist_ok=True), outputsdir.mkdir(exist_ok=True), plotsdir.mkdir(exist_ok=True)

# Specifying a few arbitrary colors to correspond to COS segments for plotting by their hex code
# Many search engines will show you the hex color - i.e. Google "#BC8C5B" to see this orange color
segment_colors = {'FUVA':'#BC8C5B', 'FUVB':'#4B6CA4', 

We will be working with some FUV and NUV datasets downloaded in the cell below.

These two datasets contain FUV observations of the QSO 3C48 and NUV observations of the Star WD1057 + 719, respectively.

Searching for and downloading data is out of the scope of this tutorial. If you wish to learn more, please see our tutorial on downloading COS data.

In [4]:
# Download the NUV data on WD1057+719 (with the G230L grating); only the _x1dsum and _asn
nuv_downloads = Observations.download_products(Observations.get_product_list(Observations.query_criteria(obs_id = 'lbbd01020')), 
                                           download_dir=str(datadir) , extension='fits', mrp_only=True, cache=False)
# Download the FUV data on QSO 3C48; only the _x1dsum and _asn
fuv_downloads = Observations.download_products(Observations.get_product_list(Observations.query_criteria(obs_id = 'lcxv13050')),
                                           download_dir=str(datadir) , extension='fits', mrp_only=True, cache=False)
Downloading URL to data/mastDownload/HST/lbbd01020/lbbd01020_asn.fits ... [Done]
Downloading URL to data/mastDownload/HST/lbbd01020/lbbd01020_x1dsum.fits ... [Done]
Downloading URL to data/mastDownload/HST/lcxv13050/lcxv13050_asn.fits ... [Done]
Downloading URL to data/mastDownload/HST/lcxv13050/lcxv13050_x1dsum.fits ... [Done]

If you're confident Reading-in and Plotting data in Python, now go ahead to Section 3

1. Reading in the data

Calibrated COS 1-dimensional spectra are stored in fits files with the suffix x1d or x1dsum (1-dimensional here means that the cross-dispersion axis has been collapsed). x1d files contain a spectrum processed from a single exposure, while x1dsum files contain summed data from multiple exposures at different fixed pattern noise position settings(FP-POS's). You may also encounter x1dsumN files, i.e. x1dsum1/x1dsum2/x1dsum3/x1dsum4, which are intermediate files containing data from multiple exposures at the same FP-POS. All of these files are sub-types of x1d files and thus share the same basic data structure. What you learn with one filetype should be transferable to working with another. We will be working with x1dsum files in this tutorial; however, please note that we often use "x1d" as part of a variable name when programming, regardless of whether it is a summed file or not.

The calibrated spectrum data has been downloaded onto our local machine as: <current-working-directory>/data/mastDownload/HST/<Obs_id>/<Obs_id>_x1dsum.fits, where the NUV and FUV Data are contained in the obs_ids:

Spectral Region Obs_id Object Name Object Type filepath
FUV LCXV13050 QSO 3C48 ./data/mastDownload/HST/lcxv13050
NUV LBBD01020 Star WD1057 + 719 ./data/mastDownload/HST/lbbd01020

1.1. Investigating the Data - Basics

We want to learn the basics about this file, then read in the data.

We can learn a great deal about our data from its primary fits header (see cell below).

In [5]:
# Make sure these filepath variables point to your new FUV data
# Note, We'll often refer to the x1dsum file with the prefix x1d for convenience
## However, the files are x1dsum files, not the related x1d files - more info in the Data Handbook
fuv_x1d_filepath = './data/mastDownload/HST/lcxv13050/lcxv13050_x1dsum.fits' 
# This is the association file, used for processing spectra with CalCOS:
fuv_asn_filepath = './data/mastDownload/HST/lcxv13050/lcxv13050_asn.fits'

# Make sure these point to your new NUV data
nuv_x1d_filepath = './data/mastDownload/HST/lbbd01020/lbbd01020_x1dsum.fits'
# This is the NUV association file
nuv_asn_filepath = './data/mastDownload/HST/lbbd01020/lbbd01020_asn.fits'

nuv_header_x1d = fits.getheader(nuv_x1d_filepath)
nuv_header_asn = fits.getheader(nuv_asn_filepath)

fuv_header_x1d = fits.getheader(fuv_x1d_filepath)
fuv_header_asn = fits.getheader(fuv_asn_filepath)

fuv_header_x1d[:18],"...",fuv_header_x1d[45:50] #This is the main, 0th header of the calibrated nuv spectrum; 
# This is *not* the whole header...
#...The [:18] and [45:50] tell Python to print the first 18 lines, as well as lines 45-50
(SIMPLE  =                    T / conforms to FITS standard                      
 BITPIX  =                    8 / array data type                                
 NAXIS   =                    0 / number of array dimensions                     
 EXTEND  =                    T                                                  
 NEXTEND =                    1 / Number of standard extensions                  
 DATE    = '2021-04-14'         / date this file was written (yyyy-mm-dd)        
 FILENAME= 'lcxv13050_x1dsum.fits' / name of file                                
 FILETYPE= 'SCI      '          / type of data found in data file                
 TELESCOP= 'HST'                / telescope used to acquire data                 
 INSTRUME= 'COS   '             / identifier for instrument used to acquire data 
 EQUINOX =               2000.0 / equinox of celestial coord. system             
               / DATA DESCRIPTION KEYWORDS                                       
 ROOTNAME= 'lcxv13flq                         ' / rootname of the observation set
 IMAGETYP= 'TIME-TAG          ' / type of exposure identifier                    
 PRIMESI = 'COS   '             / instrument designated as prime                 ,
               / DIAGNOSTIC KEYWORDS                                             
 OPUS_VER= 'HSTDP 2021_1      ' / data processing software system version        
 CSYS_VER= 'caldp_20210323'     / Calibration software system version id         
 CAL_VER = '3.3.10  '           / CALCOS code version                            )

For instance, we notice that the FUV data was taken in TIME-TAG mode and calibrated with CalCOS version 3.3.10 (at the time of writing this - it may have been reprocessed by the time you read this).

However, some metadata information, such as the time of observation and calculated exposure time, can be found in the 1-th header rather than the 0th. We will read and print this below:

In [6]:
with as hdu:
    fuv_header1_x1d = hdu[1].header
    fuv_date = fuv_header1_x1d['DATE-OBS']
    fuv_time = fuv_header1_x1d['TIME-OBS']
    fuv_exptime = fuv_header1_x1d['EXPTIME']
    #It's also perfectly valid to access the 1-th extension header using 'fits.getheader(fuv_x1d_filepath, ext=1)'

print(f"This FUV data was taken on {fuv_date} starting at {fuv_time} with a net exposure time of {fuv_exptime} seconds.")
with as hdu:
    nuv_header1_x1d = hdu[1].header
    nuv_date = nuv_header1_x1d['DATE-OBS']
    nuv_time = nuv_header1_x1d['TIME-OBS']
    nuv_exptime = nuv_header1_x1d['EXPTIME']

print(f"This NUV data was taken on {nuv_date} starting at {nuv_time} with a net exposure time of {nuv_exptime} seconds.")
This FUV data was taken on 2016-06-13 starting at 23:56:29 with a net exposure time of 6532.512 seconds.
This NUV data was taken on 2009-08-14 starting at 06:03:56 with a net exposure time of 999.136 seconds.

1.2. Reading in the x1d/x1dsum Main Data

The simplest way to read in the x1d data from fits extension #1 is using the astropy.table.getdata command. We can then display all the fields contained in this data table using the .colnames method. You can ignore the warnings about multiple slashes in the units that come up while reading in the data. The proper units are displayed in LaTex as:

  • 'erg /s /cm**2 /angstrom' ==> $$\ \ erg\ s^{-1}\ cm^{-2}\ \mathring{A}^{-1}$$
  • 'count /s /pixel' ==> $$\ \ counts\ s^{-1}\ pixel^{-1}$$

In the case of the FUV data, we see an astropy style table of 2 rows (corresponding to FUVA and FUVB - see next python cell). These rows contain data from the 2 segments of FUV Detector (see figure 1.1).

Fig. 1.1 from COS DHB Fig. 1.6

Layout of the COS FUV detector. Note that FUVB corresponds to shorter wavelengths than FUVA.

In the case of the NUV data, we see a similar astropy style table of 3 rows (corresponding to NUVA, NUVB, and NUVC - see next python cell). These rows contain data from the 3 stripes of the NUV spectrum (see figure 1.2).

Fig. 1.2 from COS DHB Fig. 1.10

An example COS NUV spectrum. The spectrum itself, taken with the Primary Science Aperture, is in the lower three stripes labeled 'PSA'. The upper stripes, labeled 'WCA' are for wavelength calibration.

The columns of these tables include some scalar values which describe the data (i.e. EXPTIME), while the columns containing actual data hold it in lists of equal length (i.e. WAVELENGTH, FLUX, etc., where that length = NELEM).

An important thing to note about this NUV data in particular is that with the grating used here (G230L), stripe C is actually a 2nd order spectrum with a higher dispersion (x2) and ~5% contamination from the 1st order spectrum. See the COS Data Handbook, especially Fig. 1.3, for more information.

In [7]:
fuv_x1d_data =
columns = fuv_x1d_data.colnames
# Print basic info about the table's columns
print("\n\nTable of FUV data with columns:\n",columns, "\n\n")
# Display a representation of the data table:

Table of FUV data with columns:

Table length=2
sAngstromerg / (Angstrom cm2 s)erg / (Angstrom cm2 s)ct / sct / sctctctctct / sct / s
FUVA6532.512163841610.2408186811224 .. 1810.94484897820480.0 .. 0.00.0 .. 0.00.0 .. 0.00.0 .. 0.00.0 .. 0.00.0 .. 0.00.0 .. 0.00.0 .. 0.00.0 .. 0.00.0 .. 0.0128 .. 1280.0 .. 0.0
FUVB6532.512163841421.9464548043077 .. 1622.59581755181670.0 .. 0.00.0 .. 0.00.0 .. 0.00.0 .. 0.00.0 .. 0.00.0 .. 0.00.0 .. 0.00.0 .. 0.00.0 .. 0.00.0 .. 0.0128 .. 1280.0 .. 0.0

1.3. The Association (asn) file

It's also likely we will want to see what observations went into making this calibrated spectrum. This information is contained in the Association (asn) file, under the MEMNAME column.

In [8]:
fuv_asn_data =
Filename: ./data/mastDownload/HST/lcxv13050/lcxv13050_asn.fits
No.    Name      Ver    Type      Cards   Dimensions   Format
  0  PRIMARY       1 PrimaryHDU      43   ()      
  1  ASN           1 BinTableHDU     25   5R x 3C   [14A, 14A, L]   


-------------- -------------- --------
     LCXV13FLQ         EXP-FP        1
     LCXV13FXQ         EXP-FP        1
     LCXV13G4Q         EXP-FP        1
     LCXV13GXQ         EXP-FP        1
     LCXV13050        PROD-FP        1

We see that our data has MEMTYPE = PROD-FP, meaning it is an output science product (see COS DHB Table 2.6.)

This particular association file lists four EXP-FP (input science exposure), with the MEMNAME values (Dataset IDs) [LCXV13FLQ, LCXV13FXQ, LCXV13G4Q, LCXV13GXQ]. We could look for these datasets, if we wished to inspect the exposures individually.

Exercise 1.1. Finding Metadata for the NUV

  1. Read in the NUV data, just as we did with the FUV data.
  2. From the 0th header of the x1dsum file, determine the time (in MJDs) that the data was processed. (keyword = PROCTIME) and from the 1th header, determine how many wavelength calibration "flashes" were used (keyword = NUMFLASH)
  3. From the asn file, determine how many input science exposures went into the NUV x1dsum file.
In [9]:
### Your code here

2. Plotting the Data

2.1. Plotting an FUV Spectrum

Let's select the simplest data we need to plot a spectrum: WAVELENGTH, FLUX, and ERROR.

  • Note that here, ERROR is flux error.
  • Also note that somewhat counterintuitively, the FUVB segment extends over shorter wavelengths than the FUVA.

2.1.1. Our First Plot

We begin with a simple plot: simply a line plot of a single segment (FUVA) without its error. The goal of this show is to demonstrate some of the common parameters of making a simple plot with matplotlib which you are likely to often need. The comments explain what each line does.

In [10]:
# [0] Access FUVA data, the longer wvln segment and gets the data we need to plot a spectrum:
wvln, flux, segment  = fuv_x1d_data[0]["WAVELENGTH", "FLUX", "SEGMENT"] 

# Set up the plot as a single box with size of 10x4 inches, and with a dpi of 100, relevant should we choose to save it:
fig1, ax = plt.subplots(1,1,figsize=(10,4), dpi = 100)  

# The next few lines are the core of the cell, where we actually place the data onto the plot:
ax.plot(wvln, flux, # First two arguments are assumed to be the x-data, y-data
        linestyle = "-", linewidth = 0.25, c = 'black', # These parameters specify the look of the connecting line
        marker = '.', markersize = 2, markerfacecolor = 'r', markeredgewidth = 0, # The marker parameters specify how the data points will look... 
                                                                                  # ... if you don't want dots set marker = ''
        label = segment) # The label is an optional parameter which will allow us to create a legend 
                         # this label is useful when there are multiple datasets on the same plot

# The lines after this are all about formatting, adding text, and saving as an image
ax.set_title("Fig. 2.1\nSimple COS Segment FUVA Spectrum", size = 20) # Adds a title of fontsize 20 points
ax.set_xlabel('Wavelength [$\AA$]', size = 12) # Adds x axis label
ax.set_ylabel('Flux [$erg\ s^{-1}\ cm^{-2}\ Angstrom^{-1}$]', size = 12) # Adds y label

plt.xlim(1605,1815) # These two lines set the x and y bounds of the image in whatever units we are plotting in 
plt.ylim(-1E-15, 1.85E-14)

plt.legend(loc = 'upper right') # Adds a legend with the label specified in the plotting call
plt.tight_layout() # Trims blank space
plt.savefig(str(plotsdir / "Fig2.1.png")) # Optionally you can save the plot as an image # Shows all the plot calls in this cell and "clears" the plotting space - must come after any saving you want to do

Exercise 2.1: A clearer plot

Plot the data from Segment FUVB, similar to the Segment FUVA plot above, but this time, normalize flux to a max of 1, and don't plot the red markers, which can be distracting.

Note what can you simply copy over, and what you have to make sure to change.

In [11]:
# Your code here

2.1.2. A Complex Look at the Entire FUV

Now that we have an idea for how matplotlib works, let's make a more complicated graph showing both FUV segments - independently and together.

One of the most important steps to creating a plot is planning out how it will look and convey its information. We'll begin by planning this out below:

Panel Contents Information Conveyed Notes
top Entire FUV Spectrum as a simple plot Overview of the entire spectrum we have, coarse look without much detail Color by segment
middle Shorter Wavelength FUVB Spectrum as an errorbar plot Closer look at the shorter wavelength spectrum with an idea of error Color errorbar by segment, central line in black
bottom Longer Wavelength FUVA Spectrum as an errorbar plot Closer look at the longer wavelength spectrum with an idea of error Color errorbar by segment, central line in black
In [12]:
fig, (ax0, ax1, ax2) = plt.subplots(3, 1, figsize = (16, 16)) # ax0, ax1, ax2 are our 3 vertically-aligned panels: top, middle, and bottom

for i in range(2): # Repeats for i = [0,1] to apply to each segment's data at a time
    wvln, flux, fluxErr, segment = fuv_x1d_data[i]["WAVELENGTH", "FLUX", "ERROR", "SEGMENT"] # Selects all useful data for the chosen segment
    # This section applies the top (0th) panel's plotting and formatting:
    ax0.plot(wvln, flux,
                linestyle = "-", label = segment, c = segment_colors[segment])
    ax0.legend(fontsize = 20 , loc = 'upper left')
    ax0.set_title("Fig 2.2\nFUV Spectra with G160M Grating", size = 35)

    if i == 0: # This indented code applies only to segment FUVA data in bottom Panel 
        markers, caps, bars = ax2.errorbar(x = wvln, y = flux, yerr = fluxErr,
                    linestyle = "-",  label = segment, marker = '', markersize = 1,
                                           c = 'k', alpha = 1, ecolor = segment_colors[segment] )
        ax2.legend(fontsize = 20 , loc = 'upper left')
        ax2.set_xlabel('Wavelength [$\AA$]', size = 30)
        [bar.set_alpha(0.75) for bar in bars]
    if i == 1: # This indented code applies only to segment FUVB data in middle Panel 
        markers, caps, bars = ax1.errorbar(x = wvln, y = flux, yerr = fluxErr,
                    linestyle = "-",  label = segment, c = 'k', ecolor =  segment_colors[segment] )
        ax1.legend(fontsize = 20 , loc = 'upper left')
        ax1.set_ylabel('Flux [$erg\ s^{-1}\ cm^{-2}\ Angstrom^{-1}$]', size = 30)
        [bar.set_alpha(0.75) for bar in bars]
plt.savefig(str(plotsdir/ 'Fig2.2.png'), dpi = 200)

2.1.3. Looking Closer at Parts of the FUV Spectrum

It can be very difficult to get any insights on small-scale details in the above plots because data is too dense to parse at once. Below, we'll show examples of:

  • Plotting a small region around an absorption line feature.
  • Plotting an entire segment's spectrum in segments of wavelength space.

Let's begin with showing a region around a line, in this case, we see a sharp line around 1670 Å.

In [13]:
line1670 = 1670.79 # An aluminium spectral line looks to be in the right place to be our line
                  # SOURCE: Leitherer, Claus, et al. The Astronomical Journal 141.2 (2011): 37.

wvln, flux, fluxErr, segment = fuv_x1d_data[0]["WAVELENGTH", "FLUX", "ERROR", "SEGMENT"] # Selects all useful data for the chosen segment

wvln_extent = 2 # How many Angstroms in each direction around line1670 do we want to look at?

lineRegion_mask = (wvln>line1670 - wvln_extent) & (wvln<line1670 +wvln_extent) # Mask the data to within +/- wvln_extent Angstrom of the line - speeds up the plotting
wvln_region, flux_region, fluxErr_region = wvln[lineRegion_mask], flux[lineRegion_mask], fluxErr[lineRegion_mask] #applies the created mask

plt.figure(figsize=(8, 6)) # how big should the plot be
plt.axvline(x = line1670, c = 'k', linewidth = 4 , linestyle = 'dotted', # vertical line at x position line1670; color = 'k' AKA black, line thickness = 4 points, type of line is dotted
            alpha = 0.8,  label = "$\lambda_{lab}$ of Al II") # alpha sets transparency where 1 is opaque; label for the legend

plt.errorbar(x = wvln_region, y = flux_region, yerr = fluxErr_region, label = 'FUVA', linestyle = '', marker = '.') # marker tells the point to look like a dot

plt.xlabel('Wavelength [$\AA$]', size = 20)
plt.ylabel('Flux [$erg\ s^{-1}\ cm^{-2}\ Angstrom^{-1}$]', size = 20)

plt.legend( fontsize = 15)
plt.savefig(str(plotsdir / 'Fig2.3.png'), dpi = 200)
plt.title("Fig. 2.3\nZoom in on the line at 1670 $\AA$", size = 25)
Text(0.5, 1.0, 'Fig. 2.3\nZoom in on the line at 1670 $\\AA$')

We note that this is around the lab-frame wavelength of Al II (1670.79 Å), an Aluminium spectral line sometimes seen in UV spectra. It's tempting to assume that because this line is close to a somewhat common spectral line (Al II) that it must correspond to that line...

But caution!

Remember the source we're looking at - the quasar 3C48. This quasar can be expected to have a substantial redshift, and indeed it does. $$z \approx 0.37$$

Source: Simbad search for 3C48

Indeed, at that redshift, we see this line to be an entirely different wavelength: $\approx 1218$ Å. This is close to, and may correspond to, the Lyman-$\alpha$ transition, rather than Al II. This somewhat misleading line provides an excellent reminder to take what you know about the source into consideration.

Now we will create a much more complex plot, allowing us to visualize an entire segment's spectrum in fine detail We'll split the spectrum into "segments" of ~10 Angstroms, and plot these segments in a vertical series. These plots may take a minute to create. If you want them to run quicker, hit the Interrupt the Kernel square-shaped button to stop the current cell from running (you will get an error message), and run this cell with a smaller number of rows.

In [14]:
# Let's see how long it takes with the above "Cell magic" ...
# (Cell magic is a Jupyter/iPython series of utilities:
for i in range(2): # Repeats for i = [0,1] to apply to each segment's data at a time
    wvln, flux, fluxErr, segment = fuv_x1d_data[i]["WAVELENGTH", "FLUX", "ERROR", "SEGMENT"] # Selects all immediately useful data for the chosen segment
    minx, maxx = min(wvln), max(wvln)
    miny, maxy = min(flux), max(flux)
    rangex = maxx - minx
    fig = plt.figure(figsize = (14,20))

    nRows = 15 # How many segments we wish to split the spectrum into

    for i in range(nRows):
        min_ = minx + i*rangex/nRows
        max_ = minx + (i+1)*rangex/nRows
        ax = plt.subplot(nRows,1, i+1)
        if i == 0: # A way to set Title, xlabel, and ylabel that will work independent of number of rows
            ax.set_title(f"Fig. 2.4{segment[-1]} \nSegment {segment} Spectrum split into segments", size = 30)
        if i == nRows - 1:
            ax.set_xlabel("Wavelength [$\AA$]", size = 30)
        if i == int(nRows/2):
            ax.set_ylabel('Flux [$erg\ s^{-1}\ cm^{-2}\ Angstrom^{-1}$]', size = 30)
        # Create the plot itself
        ax.errorbar(wvln,flux,fluxErr, c =, alpha = 0.8, 
                    marker = '.', markerfacecolor = 'k', markersize = 2, mew = 0)

        ax.set_xlim(min_, max_)
    plt.savefig(str(plotsdir / f'Fig2.4{segment[-1]}_{nRows}Rows_seg{segment}.png'), dpi = 200)