A new cloud computing platform for the sunlight reflection research community

At Reflective, a common refrain we hear from scientists in the field is that improvements in the data infrastructure available to researchers could be a powerful lever to speed-up research.

Motivated by this feedback, we interviewed various members of the global research community in late 2024, to understand how we might help. Those interviews identified consistent themes:

1. Lack of computing infrastructure. Many researchers struggle with limited computational resources. We heard about scientists doing important analysis of climate simulations on their personal laptops, downloading data first over slow internet connections, or even relying on physically sharing hard-drives to get hold of it. Getting these researchers access to 10x or 100x more compute power and memory, and removing the need to manage a local archive of data offers a clear route to increasing and improving scientific output. 
2. Fragmented sources of data. The current systems for accessing model simulation data are fragmented, and there is no common platform for sharing new simulations which don’t sit within the GeoMIP/CMIP ecosystem. In addition, accessing data via the Earth System Grid can be slow and cumbersome. A fast and easy-to-use repository of commonly used simulation data could save time and effort for researchers. 
3. Open Science opportunities. More broadly, many areas of science are currently benefiting from a move towards open science practices, through more open sharing of software, data, and information. In support of these goals, some research communities have built common cloud research environments with a shared software environment, to allow for seamless sharing of code and easier collaboration. 

Inspired by these ideas, we are now launching the Reflective Cloud – an interactive cloud computing platform, designed to support open, collaborative science for the SAI research community. The cloud hub offers significant compute resources to researchers for free, via a Jupyter notebook server accessed in the browser. It allows for analysis of climate model data entirely in the cloud, not limited by the researcher’s local computing power or internet speed. To manage the usage of these resources, users need to be granted access – fill out this form to apply. 

We also hope the cloud hub can be a home for collaboration and the sharing of resources among the research community. To this end, we have put together an initial set of resources and code examples here, including tutorials for working with the ARISE-1.5K and GeoMIP simulations in the cloud, and a new archive of post-processed and analysis-ready data. We welcome contributions to this resource, which can be made via Github. Going forwards, we will be expanding this into a new cloud-optimised repository of SAI simulations, to support quick and reproducible workflows without the need for local data archives. 
These resources are in development, and we are actively exploring how best to support the movement towards more open, collaborative, and empowered science. We would love to hear feedback – get in touch here.