Science reproducibility has been in the spotlight for a while now, with some going as far as calling it a crisis. Many of the underlying principles defended by those sounding the alarm are fair-minded. Others are just plain impossible to attain in practice. This is especially true in our modern impact-driven science-funding world. Without claiming to be a reproducibility expert, in this presentation I will offer an experience report on how we tackled a long-running reproducibility challenge, while trying to keep practical concerns in perspective–especially time (read: project funds). In more detail, this work involved re-implementing algorithms for partitioning ecosystem carbon dioxide fluxes into its components: ecosystem respiration and photosynthetic productivity. I will argue that while it is worth striving for reproducible science, you still have to deliver your science results–preferably before your deadline!
When: July 13, 2020 at 1 pm
Speaker: Gilberto Pastorello
Research Scientist
Integrated Data Frameworks Group
Data Science and Technology Department
Computational Research Division
Lawrence Berkeley National Laboratory
