282264 - Towards reproducible climate research

NOAA/GFDL was founded in 1955 and is still in the forefront of climate research, contributing to the numerous policies and decisions undertaken in this world of evolving responses with respect to climate, which in turn creates an avalanche of effects in various sectors, e.g agriculture, health, GDP. The scale and magnitude of computing and data have proven to increase significantly in the last decade, thus making data delivery methods to the world a herculean research problem by itself. In addition to this, the time and efforts invested by a user in analyzing and peer-reviewing a research article is very laborious. Literature shows numerous outstanding climate studies published in International climate assessment reports, such as the Intergovernmental Panel on Climate Change (IPCC), the United Nations body for assessing the science related to climate change. The need to verify the research and make it reproducible and transparent before it gets translated into major decisions is, now more than ever, one of our most critical challenges. In this presentation, we will paint a picture of the history of climate computing and analytics with significant transformations applied in order to make meaningful, quantifiable, credible, interoperable, accessible and reusable climate research. In other words, we will draw a path towards reproducible research using Docker containers for massive data publishing and climate analytics. This paper will also discuss some of the pioneering efforts from collaborators from other laboratories and organizations (such as ESGF, Google, NASA JPL, Columbia University, PMEL, etc.) in the area of docker containers in computing and analysis on and off the cloud.