Open Oceanography - FOSS science
Abstract
As a remote sensing scientist and physical oceanographer, software has always been central to my science. For many years I worked on algorithm development for passive microwave radiometers carried on several different satellites. Within the geophysical algorithm development field, the entire code base is generally considered proprietary, resulting in multiple (sometimes 8 or more) groups working completely independently to produce the same geophysical variable from a single instrument. Even simple data screening techniques arent shared, much less more complicated algorithms. This has resulted in substantial replication of effort, which in a budget constrained environment, slows new development and innovation. To put it simply, closed code is slowing science. Like many people in my field, I worked exclusively in Fortran for data processing and analysis and used Matlab mostly for data exploration and visualizations. In 2018, just a few years ago, I transitioned from Fortran/Matlab to Python and began using GitHub to publish all my code as FOSS from inception. I walked away from 20 years of code that I had developed and tailored to my specific needs and analysis methodologies. Now, my science is faster, better, and more collaborative. When a feature isnt provided or I find a problem with software, I can raise an issue and work directly with FOSS engineers on GitHub. This separation of labor and collaborative workflow allows us each to focus on our area of expertise and results in better, faster science results. I can publish my software with a digital object identifier in open-access journal publications so that anyone can quickly and easily reproduce my results, figures, and tables. FOSS can transform science, increase reproducibility, and advance innovation. In this talk, I will illustrate these ideas by looking at how in a more recent project we opened up our code. I will discuss both the benefits we gained from this approach, as well as some areas where challenges remain, some which pertain to policies, others to community practices, and others to technical limitations. I hope this discussion will spur ideas to address each of these.
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2021
- Bibcode:
- 2021AGUFMIN55C..02G