If you know how to swim, you must have learned it in the water. The same is true for writing. You have to keep on writing in order to learn how to write well.
If you are writing a paper of more than 3,000 words, you should consider putting together an outline first.
An outline defines the logic flow of your paper. If you add titles for subsection, sub-subsection, and even sub-sub-subsection, you should see a well-paved path. If not, you need to revise it. Then, you can add bits and pieces to where they belong to. You may take out the sub-sub-subsection titles when you are ready to submit the manuscript.
Here is an example (for a paper using a numerical ocean model and data assimilation technique):
I. Introduction
1.0 Opening paragraph
1.1 Background (review publications)
1.1.1 Observational background
1.1.2 Theoretical background
1.1.3 Modeling background
1.2 Present research
II. Methodology
2.1 Data
2.1.1 Atmospheric data
2.1.2 Air-sea flux data
2.1.3 Ocean data
2.1.3.1 Data used for assimilation
2.1.3.2 Data used for validation
2.2 Models
2.2.1 Model basics (equations, etc.)
2.2.2 Forcing field
2.2.3 Basin, resolution, and boundary conditions
2.2.4 Initial conditions
2.2.5 Output sampling and averaging
III. Results
3.1 Main run
3.1.1 Mean state
3.1.2 Seasonal variability
3.1.3 Interannual variability
3.2 Sensitivity to winds
3.3 Sensitivity to mixing
…
IV. Summary and discussion
4.1 Summary
4.2 Discussion
4.2.1 Unresolved issues
4.2.2 Future work
Remember:
Scientific writing is hard work.
Scientific writing is not science.
Scientific writing is a craft. It is a skill that must be developed through practice, practice and more practice.
References: The craft of scientific writing by Michael Alley (1987, 1996)