Warner, T. T., 2011: Quality assurance in atmospheric modeling. Bull. Amer. Meteor. Soc., 92, 1601–1610. http://dx.doi.org/10.1175/BAMS-D-11-00054.1 [PDF]

1. Clearly define the scientific or practical objective.

2. Identify and develop a physical understanding of the atmospheric processes that must be accurately simulated.

3. Perform a thorough analysis of all available observations.

4. Prepare an experimental design.

5. Define the required horizontal and vertical resolutions of the model.

6. Avoid the tendency to prematurely run the model, before the above-listed steps have been completed.

7. Choose the model start time and the method of model initialization to allow for spin-up of the physical processes of interest.

8. Run test simulations of evaluate the sensitivity of model solution to the computation domain size.

9. Define the most appropriate physical process parameterizations.

10. Understand the limitations to the predictability of the phenomena being modeled.

11. Establish a verification plan before the model is run and perform a thorough verification, using appropriate metrics, of the model solution using all available observations.

12. Be well organized in maintaining a detailed experimental log and the files of model code and output.

13. Use good coding practices and well-documented and well-tested software.

14. Employ open-source software tools to improve the efficiency of the modeling process.

This paper follows from Warner’s 1997 tutorial on lateral boundary conditions.

Warner, T. T., R. A. Peterson, and R. E. Treadon, 1997: A Tutorial on Lateral Boundary Conditions as a Basic and Potentially Serious Limitation to Regional Numerical Weather Prediction. Bull. Amer. Meteor. Soc., 78, 2599–2617. [PDF]

Warner’s paper is also a natural companion to my paper on how to write effective case studies in meteorology.

Schultz, D. M., 2010: How to research and write effective case studies in meteorology. Electronic J. Severe Storms Meteor., 5 (2), 1-18. [PDF]

There is also complementary material in Chapter 18 in Eloquent Science.

In their 2011 Golden Bull Awards, however, the Plain English Campaign got it dead wrong.

The UK Met Office won an award for ‘empowering people to make their own decisions’ by using the technical systems for the ‘probabilities of precipitation’. Presumably, the Campaign was upset that terms such as “very likely” are replaced by something like “70% chance of rain”. This award was picked up by the UK newspapers like the Metro, Daily Mail, and Guardian. Apparently, one of the issues is that the forecasts now have an “American style” to them.

Here is the Met Office explanation for why probabilities were introduced. Specifically,

Often people want to make a decision, such as whether to put out their washing to dry, and would like us to give a simple yes or no. However, this is often a simplification of the complexities of the forecast and may not be accurate. By giving PoP we give a more honest opinion of the risk and allow you to make a decision depending on how much it matters to you. For example, if you are just hanging out your sheets that you need next week you might take the risk at 40% probability of precipitation, whereas if you are drying your best shirt that you need for an important dinner this evening then you might not hang it out at more than 10% probability. PoP allows you to make the decisions that matter to you.

Let’s face it. As good as meteorologists do with the forecasts, there is still a measure of uncertainty, even for forecasts less than a day in advance. I would like to think that communicating the uncertainty to the forecast is something that the public would appreciate, but perhaps numerical values are too much. If the American public can interpret that a 70% chance of rain is larger than a 50% chance of rain, I don’t understand why the British can’t.

Check those page proofs carefully!

The column points out something that we scientists may know but forget from time to time: factors other than the quality of the science determine whether and how often our articles get cited by others. Among the factors that scientists have found affect the citation rate:

• whether the article gets covered by the New York Times

• whether the titles of the articles asked a question

• the length of the titles

• “highly amusing” titles

• presence of an acronym or colon (:) in the title

Some of these are positive effects on the number of citations. Some are negative effects on the number of citations. Some depend on the discipline. In any case, how to get your paper cited more frequently is apparently a game that involves more than just good science.

Enjoy!

Alon, U., 2009: How to choose a good scientific problem. Molecular Cell, 35, 726-728. DOI: 10.1016/j.molcel.2009.09.013. [PDF]

Three-Month Thesis has a thoughtful post about how many thesis drafts you need. I would disagree with a few things on that page.

1. The number of drafts depends very strongly on the student. I would argue that you need as many drafts as it takes to finish the thesis.

2. In principle, the quality of the thesis is improved after each draft, although I am not sure that it will look like the graph above from his Web site. Using the writing/editing funnel approach in Eloquent Science (chapter 7), revisions on the largest scale should be performed first (those related to organization), whereas revisions on the smallest scale should be performed later (those related to grammar, punctuation, spelling). Depending again on the student (some may be better at organization than spelling), the quality may not be increasing in the way presented.

3. I think it’s OK to keep doing research while you write. Sometimes you need additional experiments or to create extra plots to make your argument more convincing. Limiting your ability to improve your argument by saying no to additional research is wrong. [That said, I agree that it may be too attractive to keep doing more work rather than writing up what you've done to date.]

Nevertheless, I want to highlight three important points that Three Month Thesis raises:

• Discuss the thesis structure with your supervisor before you start.

• Plan chapters before you sit down to write, so you know what you’re going to include before you start.

• Give chapters to your supervisor one at a time, rather than drafts of the entire thesis.

These three points I do agree with, and they provide some guidance for working with your supervisor. The last point is particularly important as it is easy for a supervisor to identify problems in your writing from one chapter that can be fixed before the next chapter is written, thereby saving everyone time.

Remember that only you, your supervisor, and your externals are going to read your thesis. It should be adequate to demonstrate that you know your field, can describe your research clearly, and have made a novel contribution to your science. Being too much of a perfectionist with this document is generally not useful. Save that level of effort for the peer-reviewed publications.

Schultz, D. M., 2011: Rejection rates for multiple-part manuscripts. Scientometrics, 86, 251-259. [PDF]

I found that although the rejection rates for multiple-part manuscripts were not that different from the rejection rates for manuscripts as a whole, the reviewer comments about multiple-part manuscripts suggested some ways that authors could avoid those problems in the first place:

“Write manuscripts that are sensibly independent of each other, make minimal reference to unsubmitted manuscripts, and have sufficient and substantiated scientific content within each manuscript.”

Since doing that research, I’ve had discussions with several authors before submission, and I am proud to say that I’ve convinced them to follow a different course: writing sensible manuscripts that are independent of each other. We still get cases of multiple-part manuscripts coming in, though. They are a pain to handle.

Recently, I became aware of a case where Part 1 was submitted to one journal, and Part 2 was submitted to another journal. Part 2 was accepted, yet Part 1 is still in review. This kind of outcome just complicates matters for everyone: authors, editors, reviewers, publishers and readers. Who is the winner here? Is labeling your papers as part of a multiple-part series really worth all this hassle at all stages: peer-review, publication, and post-publication?

What are the indications that you are going to struggle with peer review with your multiple-part paper? (With apologies to Jeff Foxworthy.)

If your two manuscripts are so intertwined that Part I refers to Part II as much as Part II refers back to Part I, you may have a problem in peer review.

If you started writing a paper, only to realize that you have more than the required word-count that you want to say and so start writing a second paper, you may have a problem in peer review.

Simply put, writing a multiple-part manuscript is not easy. Balancing material between the two parts, yet providing enough information so that the reviewers believe they have all the evidence is not easy.

Since publishing that article, I have thought more about what the results from my article in Scientometrics mean. My impression is that high-quality authors could navigate the potential minefield pretty well. They were clever enough to avoid the pitfalls. Mediocre authors struggled to get published because of the particular challenges that writing multiple-part papers required. And, low-quality authors who had trouble getting their single-submission papers published in the first place had little hope of getting a multiple-part paper published.

Personally, I can’t see the benefit in writing multiple-part papers for myself, but I see why some people do it. (Frankly, I think a lot of authors just do it because they have seen it in the literature and do the same thing. In fact, a lot of the literature has that flavor…it worked for Author A, must work for me the same way!)

Some journals ban multiple-part papers. Is it time for the AMS journals to abolish multiple-part papers?

Reread How-To guides and other inspirational books and articles. Here are some links and some recommendations ([1] and [2]) to get you started. For me, I derive inspiration from Strunk and White’s The Elements of Style, which I reread every 3-5 years. Rereading parts of Eloquent Science can also be helpful.

Maintain a list of your common weaknesses in writing. Keep this list on a sheet of paper on your desk or a bulletin board at your desk. Having those in front of you and referring to them when writing will help you shake those bad habits.

Write every opportunity you get. Even if they are short articles for Weather magazine describing an event, the practice will help you improve.

Have you read an article where the author talks about “an important process” or “the important role of another process”?

Do these sort of platitudes go in one of your ears and out the other? Are you convinced by the author’s use of the word “important” that it truly is an important process? Or, do you want to see the supporting evidence for why it is important?

How about quantifying exactly how important that particular process is? If calculations show that varying the input only led to a 20% change in the output, is that an “important” effect?

Given that multiple processes usually come together to produce a weather event, is isolating one process as “important” a realistic description of how the atmosphere works?

They say that the First Rule of Writing is to show, not tell. Thus, you should avoid telling the readers how important something is, and show them instead.

(Image from smartpassiveincome.com)

Several authors have criticized the use of the term overrunning to represent warm-frontal lifting here and here.

I don’t need to add anything to those Web pages, but I do want to point out that the definition provided in the American Meteorological Society’s Glossary of Meteorology is wrong and ambiguous.

overrunning—A condition existing when an air mass aloft is in motion relative to another air mass of greater density at the surface.
This term is usually applied in the case of warm air ascending the surface of a warm front or quasi-stationary front.

An air mass aloft will always be less dense than air at the surface. Therefore, every instance of sheared flow could classify as overrunning by that definition. As Stoelinga et al. (2002, BAMS) point out about ascent over fronts, it’s not the density or temperature that is important, but the static stability. The warm air flows over a warm-frontal zone because the warm air is less stable than the frontal zone (highly stable). With this ambiguity and the problems already stated, I recommend that everyone avoid the use of overrunning.