Introduction
The explosion in the availability of large and complex data sets from diverse sources (e.g. environmental monitoring; remote sensing; bio-logging, climate modelling; social media; and contributions from citizen science), together with advances in methodology and computation presents an exceptional opportunity to transform our knowledge of both the effects of environmental change and to develop solutions to environmentally related challenges.
Unlocking the full power of data will have far reaching impact across a wide spectrum of environmental areas; it will raise the bar in data-driven environment research and provide the information and insight required by sectors strongly impacted by the environment including agri-food, fisheries, water, waste, transport, health, finance and energy. It will enhance our resilience to natural hazards ability through enhanced prediction, forecasting and early warning systems, and the development of next generation, data-led, resilience planning, resilient infrastructure and financial instruments.
Environmental challenges are inherently global, and the sheer volume of data that need to be incorporated into decision making at this scale means that data science and AI will be crucial in providing the robust, decision-quality information that is required. This is especially true in the case of the Sustainable Development Goals (SDGs) for which quality, accessible, timely and reliable disaggregated data is essential for understanding the scale of the challenges and in measuring progress.
Explaining the science
The global environmental research community has developed a plethora of observational and modelling platforms designed to better understand the evolution of our environment and our role within in. As we further push the boundaries on detailed observations from molecular through to global processes, we gather more evidence deemed important to affect change. However, we have reached a crossroad of exploration that hinges on utilising this information to develop sustainable solutions to key problems, such as pollution and resource management. Likewise, implementing change requires a deep understanding of future impact, unintended consequences, behavioural change and effective policies. Numerical models of environmental systems used to predict and quantify change are traditionally containerised collections of stiff coupled ODEs with restrictions placed around computational complexity. New data driven approaches could exploit the wealth of global observations to build new modelling frameworks that cover local to global scales whilst also exploiting emerging compute hardware.
In many cases, a mechanistic framework that couples process to environmental impact does not exist. Machine learning representations could offer a potential route to develop robust tools for a range of environmental stake-holders and would allow us to make truly effective use of the rapid increase in the data from a variety of sources including:
- Data from the digitisation of specific environmental systems, e.g. water and air quality monitoring
- Earth system observations from remote sensing satellites, e.g. land-use
- Unstructured data sources e.g. text and video footage
- Personal and population-based health data
- Financial and non-financial data generated through business transactions.
Aims
Environmental data science naturally sits at the boundaries of multiple disciplines. It is the aim of this interest group to:
- Identify new solutions to key environmental challenges.
- Provide a platform for knowledge exchange on common problems identified in other fields.
- Identify funding mechanisms to build a sustainable community
- Facilitate the identity of environmental data science as a career path
In the long-term, the Environmental sustainability interest group aims to discuss mechanisms for implementing change through policy and behaviour change. This will require engagement with local, regional and national bodies to understand the key data and technology driven challenges they face to affect change.
How to get involved
The opportunity to join this group is open to academics across all universities, businesses and public and third sector organisations.
Organisers
Dr Scott Hosking
Co-director for Natural Environment, Turing Research and Innovation Cluster in Digital Twins (TRIC-DT)
Dr Emily Paremain
Manager of the Institute for Data Science and Artificial Intelligence, University of Exeter
Dr Alejandro Coca Castro
Research Fellow
Dr Alden Conner
TPS Senior Researcher, Research Applications