Best practices in software development for robust and reproducible geoscientific models based on insights from the Global Carbon Project models
Bibliographic record
Abstract
Abstract. Computational models play an increasingly vital role in scientific research, by numerically simulating processes that cannot be solved analytically. Such models are fundamental in geosciences and offer critical insights into the impacts of global change on the Earth system today and in the future. Beyond their value as research tools, models are also software products and should therefore adhere to certain established software engineering standards. However, scientists are rarely trained as software developers, which can lead to potential deficiencies in software quality like unreadable, inefficient, or erroneous code. The complexity of these models, coupled with their integration into broader workflows, also often makes reproducing results, evaluating processes, and building upon them highly challenging. In this paper, we review the current practices within the development processes of the state-of-the-art land surface models used by the Global Carbon Project. By combining the experience of modelers from the respective research groups with the expertise of professional software engineers, we bridge the gap between software development and scientific modeling to outline key principles and tools for improving software quality in research. We explore four main areas: 1) model testing and validation, 2) scientific, technical, and user documentation, 3) version control, continuous integration, and code review, and 4) the portability and reproducibility of workflows. Our review of current models reveals that while modeling communities are incorporating many of the suggested practices, significant room for improvement remains in areas such as automated testing, documentation, and reproducible workflows. For instance, there is limited adoption of automated documentation and testing, and provision of reproducible workflow pipelines remains an exception. This highlights the need to identify and promote essential software engineering practices within the scientific community. Nonetheless, we also discuss numerous examples of practices within the community that can serve as guidelines for other models and could even help streamline processes within the entire community. We conclude with an open-source example implementation of these principles built around the LPJ-GUESS model, showcasing portable and reproducible data flows, a continuous integration setup, and web-based visualizations. This example may serve as a practical resource for model developers, users, and all scientists engaged in scientific programming.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.070 | 0.165 |
| Meta-epidemiology (narrow) | 0.002 | 0.002 |
| Meta-epidemiology (broad) | 0.001 | 0.003 |
| Bibliometrics | 0.009 | 0.008 |
| Science and technology studies | 0.001 | 0.003 |
| Scholarly communication | 0.008 | 0.009 |
| Open science | 0.006 | 0.005 |
| Research integrity | 0.002 | 0.006 |
| Insufficient payload (model declined to judge) | 0.002 | 0.002 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".