Addenda and Errata

This page lists addenda and errata to the First Edition (2012) of the book A Hands-On Introduction to Using Python in the Atmospheric and Oceanic Sciences, by Johnny Wei-Bing Lin. (Note that minor typos are not listed.)

• p. vi: Change the sentence:

  If this is all you want, the first half of the book (Chs.1-6), should be all you need.

  to

  If this is all you want, the first half of the book (Chs.1-6), plus Ch. 9, should be all you need.

  Also add a footnote to the corrected sentence saying:

  Ch. 9 covers visualization using the matplotlib package. You can use much of matplotlib without knowing anything besides procedural programming, but contour plotting in matplotlib, for AOS applications, makes more sense if you understand the object-oriented framework. This is why I put Ch. 9 in the second half of the book.

  (Added Aug 26, 2012.)
• p. viii: Instead of distributing the course_files directory by putting a tarball on my website, as the book suggests, I've decided to upload that directory to GitHub. This provides multiple ways of obtaining the files. I've redirected the pertinent URLs in the book to automatically go to the page with instructions on how to download the files. Thus, those online instructions, rather than any implied instructions in the book, should be followed to obtain the files. (Updated Aug 25, 2012.)
• p. 8 (and others): I didn't realize this when I was writing the book, but if you're using the Enthought Python Distribution (EPD), whether the free version or the full version, the ScientificPython package is not included. (I had assumed ScientificPython was part of it without checking; sorry about that.) All the code examples I provide, both in the book and in the course files, assume you have the ScientificPython package. Thus, if you use EPD, any code I provide that will access netCDF files will not run.

  One workaround for this issue is to use one of the other options I mention for getting and installing Python (p. 8 and 9). Another workaround is to use the netCDF4 package instead of ScientificPython; the netCDF4 package is part of the full version of EPD (thought not the free version), and has an API very similar to that used in ScientificPython. The changes you need to make to make my code work with the netCDF4 package are to change the import line (e.g., the bottom of p. 81) to:

  import netCDF4 as S

  and instead of using the NetCDFFile constructor, use the Dataset constructor; the example on p. 81 then becomes:

  fileobj = S.Dataset('file.nc', mode='r')

  Other occurrences of the import and/or NetCDFFile calls would likewise be changed. Finally, the netCDF4 package's getValue method does not work for arrays (it does, however, work for scalars), unlike ScientificPython's getValue method which works for both arrays and scalars. In array cases, use array slicing syntax to obtain the data. Thus, on the bottom of p. 81, the example becomes:

  data = fileobj.variables['Ts'][:]

  This works regardless of the rank of the data. (Array slicing syntax, without calling getValue, also works for arrays in both the ScientificPython and netCDF4 packages.) Once these changes are made in all applicable code samples, the rest of the code in those samples should work. Again, I'm so sorry that I didn't catch this error prior to publication of the book. (First updated Nov 13, 2012. Last revised December 8, 2012.)

• p. 48: In Example 24, the definition of mylist is incorrect; the line should be:

  mylist = [[2, 3, -5],[21, -2, 1]]

  (Added Feb 4, 2014.)
• p. 53: In the solution to the second problem of Exercise 13, the text says that a[1:4,0:2] selects the last two rows and first three columns. It should instead read that it selects the last two rows and the first two columns as a subarray. (Added Oct 8, 2015.)
• p. 55–56: In the second bulleted example on p. 55, the one that begins "Flatten the array ...", the command should read:

  N.ravel(c)

  rather than N.ravel(a). This is because a is already a 1-D array. The output on p. 56 for N.ravel(c) should then read:

  [ 2. 1. 3. 3.2 22. 1. 5.5 4. 2.1 -6.4 0.1 21. -2.2 5.3 1.1 2.4 -9. -2. ]

  (Added Oct 21, 2014. Thanks to Andres Sepulveda!)
• p. 60: Near the bottom of the page, the phrase, "Additionally, the output array c is automatically created ..." should be changed to specify the variable product_ab rather than c. (Added Oct 21, 2014. Thanks to Andres Sepulveda!)
• p. 76: In the final code block at the bottom of the page, insert: a = ['3.4', '2.1', '-2.6'] after the import numpy as N line. (Added Oct 30, 2014. Thanks to Andres Sepulveda for pointing out the issue!)
• p. 96: The tokens N.skew and N.kurtosis should read scipy.stats.skew and scipy.stats.kurtosis, an there should be an import scipy.stats line at top. (Added Jan 4, 2015. Thanks to my 2015 AMS Beginner's Python short course for catching the error!)
• p. 103: In Exercise 21, question 1, the question should ask to create a "3 row, 4 column" array instead of a "3 column, 4 row" array. (Added Feb 2, 2014.)
• p. 108: In the make_authoryear method, there should be a space before the parenthesis in line 13 of the Exercise 22, question 2 solution. Thus, line 13 should read:

  + ' (' + self.year + ')'

  (Added Feb 2, 2014.)
• p. 110: "Realizations of an object are called instances," should read, "Realizations of a class are called instances." (Added July 3, 2020.)
• p. 130–137: In my Atmosphere class skeleton, I use a raise statement to catch the absence of the quantities needed to calculate a derived quantity. The way the code is written (on p. 132, lines 12–13 and p. 136, lines 12–13 and 23–24), the raise statements will actually prevent the method from ever calculating your quantity of interest. The text on p. 134 referring to the raise (the penultimate sentence of the penultimate paragraph) should, accordingly, be deleted.

  Instead of including the else statements that contain the raise statements, an if test should be added for any and all quantities that are primary, and cannot be calculated, and an exception thrown if those primary attributes are undefined. Thus, on p. 132, if pressure is a primary attribute, there should be a separate if statement (not an elif) testing to see if the attribute p exists, and if not, a raise is called, i.e.:

  if not hasattr(self, 'p'):
    raise ValueError, "cannot obtain given initial quantities"