Document Type

Unpublished Paper

Publication Date

12-8-2025

First Page

1

Last Page

90

Abstract

This project aims to create a pathway towards a future, ten years from now, when a majority of published research across geosciences is not only FAIR (Findable, Accessible, Interoperable, and Reusable), but reproducible by a third party. Building on the Center for Open Science framework for cultural change, this project is organized around four scaffolded goals: improving infrastructure and tools, building community support, aligning incentives, and minimizing barriers to policy change. We will systematically analyze 300-500 publications from leading hydrology and water resources journal to quantify changes in reproducibility rates since 2017 and identify the tools and practices that best enable reproducible research. Through evaluation of existing tools and workflows against seven use cases, ranging from spreadsheet models to high performance computing, we will establish best practices that improve reproducibility without excessive burdens on authors. Understanding the unique challenges and opportunities posed by AI in geosciences, we will include machine learning as a dedicated use case, while also testing how large language models can be used to evaluate research code and data artifacts for compliance with reproducibility best practices. Workshops and an online course will build a cohort of researchers committed to open science and reproducibility. In addition to these “bottom-up” approaches that focus on author best practices, we will evaluate complementary “top-down” approaches, providing recommendations for incentives and statement options for journals, funders, and institutions. Although focused on hydrology and water resources, the findings, tools, and workflows developed here are designed to be applicable to the broader geosciences, due to the wide range of use cases. CloudAccess.

Download

Share

COinS