Please note that all opinions below are my own, and not necessarily those of my institution, funders, or colleagues.
I aim to minimise the carbon footprint of the lab's activities wherever possible.
The climate data show that we have very little time to radically reduce our carbon emissions to secure a sustainable future, by limiting global temperature increases to 1.5C (Intergovernmental Panel on Climate Change 2018 report).
I believe that as a scientist, I am morally responsible for acting in accordance with what our data show. Furthermore, I believe that all scientists, whether climate specialists or neuroscientists, need to take action on this critical issue. The reason is that climate change will affect our economy (and therefore research funding), societal stability, and energy security, all of which we rely on to conduct our research. These effects will happen in our scientific lifetimes, and those of our trainees.
There are some useful summaries elsewhere on the carbon footprint of scientific research (Nature editorial 2017, Nathans & Sterling, 2016, eLife), and what scientists can do to mitigate this. Here I outline the specific issues for my lab, and how I am trying to tackle these.
I have also written a guide for academic societies that are looking to generate an Environment Policy, focusing in particular on how to reduce the climate impact of conferences and meetings.
This is an ongoing and evolving attempt to list and manage my lab's carbon footprint: please do get in touch with any suggestions.
- Talks and presentations
Sussex Sustainability Assembly, February 2020
"The climate impact of academic flying...and what can we do about it?" (scroll to 1:27)
In this talk at the first Sussex Sustainability Assembly, I gave an overview of why academic air travel is a problem, and what we could do about it locally at Sussex.
An outcome of this presentation is that I am currently developing a Sustainable Staff Business Travel Policy with academic and professional services colleagues at Sussex.
Open Science Room at Organization for Human Brain Mapping, June 2020
In this talk at the OHBM 2020 Open Science Room, I examined the environmental impact of open science, with a particular focus on neuroimaging datasets. I explained why open neuroimaging has environmental consequences, assessed the green credentials of popular repositories, and proposed that we might need to more fundamentally reduce scientific consumption first and foremost, before mitigating the footprint of data acquisition and sharing.
An outcome of this talk is that I am currently establishing an OHBM Environment Special Interest Group (email me if you are a member of OHBM and would like to join).
- Air travel
The biggest carbon footprint of academic research probably comes from air travel - flying to conferences and meetings.
Much has been written elsewhere about how big a problem flying is, why we need to reduce it, and why academic scientists are among the 'worst offenders' (Kimberly Nicholas 'Data on academic flying', Nathans & Sterling, 2016, eLife, Nature editorial 2015). Parke Wilde and Joseph Nevins of 'Flying Less: Reducing Academia's Carbon Footprint' have produced a very useful FAQ covering these points.
I am restricting my air travel, because I feel strongly that the carbon footprint is too high. This means that I focus my conference attendance on meetings that I can travel to by train (in the UK and Europe). I have not flown in 5 years, and in that time have attended an average of 1.25 conferences and meetings a year. It is possible to remain active in conference participation without flying.
I believe #flyingless is the one action that can have the biggest impact on the carbon footprints of academic scientists.
Interestingly, recent data show that there is no relationship between air travel and metrics of academic productivity, such as publications or h-index (Wynes et al, 2019).
I have written this guide for academic societies that are looking to generate an Environment Policy, focusing in particular on how to reduce the climate impact of conferences and meetings.
- Open data and code repositories
Uploading our data and code to public repositories has a carbon footprint, because this requires servers to store the information. Energy is required to manufacture the servers, and then to run them.
I use the Open Science Framework (OSF) to share my data and code. OSF uses Google Cloud storage, which is run using renewable energy (at least, purchasing the equivalent of the energy usage from renewable sources).
I share summary fMRI results on NeuroVault. NeuroVault uses Amazon Web Services, only 50% of which is run using renewable energy. After I enquired, NeuroVault are aiming to switch their AWS server location to one that is guaranteed to run on renewables.
Also, see my Open Science Room talk at OHBM 2020 for a longer discussion on the environmental impacts of open neuroimaging.
- Purchase of scientific equipment
The manufacture and transport of goods has an associated carbon footprint. This includes scientific equipment. I use various bits of kit to run my experiments, such as pulse oximeters to measure participant's heartbeats, and actiwatches to measure sleep and activity patterns.
I am minimising the carbon impact of my lab's physical hardware by purchasing items only when it is really necessary for a defined experiment, and only purchasing the amount I need. This is also ethical usage of our limited research funds.
- Web searching
Each time we search the web there is an associated carbon footprint, because running servers takes electricity. Some search engines now run their operations using renewable energy (e.g. Google have since 2017).
You can go one better and use a search engine that not only runs on renewable energy, but does something useful with its profits. I use Ecosia, which plants trees. This is the default search engine across the University of Sussex campus, but it is easy to install as your defaut search bar off-campus.
- MRI brain scanning
MRI brain scanning has a carbon footprint. This comes from the manufacture, transport, and installation of the scanner; its day-to-day operation - which requires electricity; and the liquid helium that cools the super-conducting elements. This calculation puts the carbon footprint of one scan at 160kg.
The 2 MRI scanners we have in the Clinical Imaging Sciences Centre are Siemens models. Siemens has a large decarbonisation program, and aims to be carbon neutral by 2030. For example, the energy involved in the manufacture of scanners comes from renewable sources.
The day-to-day operation of the scanner requires electricity. The University of Sussex is currently undergoing a major revamp to replace the current systems that generate energy on campus with more sustainable alternatives.
MRI scanners require liquid helium to cool the super-conducting elements. Liquid helium is a naturally-occuring substance in the geological environment. Unfortunately, it exists almost entirely in reserves of natural gas. This means it is extracted as part of fossil fuel mining for natural gas - but we need to 'keep it in the ground'.
I am yet to find an answer to the climate impact of liquid helium, so I am applying the 'reduce, reuse, recycle' mantra.
Reduce: Only acquire (good quality) data that you intend to use.
Reuse: Consider if you can use data that already exist, for example from the Human Connectome Project.
Recycle: Share your summary data on Neurovault so others can use it for meta-analyses.
- MRI data analysis
Analysing our MRI brain scan data requires servers with large storage capacity. There is a carbon footprint from the manufacture of such servers, and then the energy required to run them.
We use the Sussex High Performance Cluster to store and analyse our MRI data. Sussex is currently making major changes to its high performance computing resourcing, and I was invited to advise on the environmental implications as a 'Sustainability Champion'. Watch this space.
Reducing the amount of storage space you use will reduce both the energy requirements to store your data, and require less overall server storage - thus reducing the manufacture footprint. Large datafiles for MRI scans are inevitable, but we can reduce the total amount of storage used by deleting intermediary files that result from our image processing steps. For example, a typical fMRI analysis pipeline in SPM involves preprocessing stages such as realignment and normalisation. You can delete the outputs of these stages and keep only the final, fully preprocessed fMRI data. Processing multimodal Human Connectome Project style acquisitions also creates a lot of intermediary files, not all of which may be necessary to keep.
One can also reduce the number of files created by running only analyses one really needs to, in order to answer hypotheses. This is also just good scientific practise.
- Personal carbon footprint
If you are interested in calculating a personal carbon footprint, this WWF tool may be useful.