Human activity in the McMurdo Dry Valleys. Rescue, knowledge and understanding our role as a vector of change.

Accessing, rescuing, and analysing the vast range of (mostly hidden) historical information about human activities in this geographically and scientifically distinctive region.

The need and impact

Although humans are increasingly impacting the environment, human activity is not a focus of Antarctic research. The McMurdo Dry Valleys, accessible from Scott Base and McMurdo Station, are geographically and scientifically distinctive and like most ice-free regions in Antarctica, increasingly vulnerable to change.

Accessing, rescuing, and analysing the vast range of (mostly hidden) historical information about human activities in this region is vital. These data will allow analysis of where humans have explored, how patterns of activity have changed, the range and scale of activities undertaken, and the impact on what is regarded as a pristine landscape.

Developing novel approaches to analyse this unique reconstructed dataset provides insights into how human activity has affected and continues to affect Antarctica. The impact of the project is manifold, with research translation for Antarctic environmental managers and policy makers, enabling informed debate and decision-making that anticipates changes in climate and environmental conditions.

This research will leverage Aotearoa New Zealand’s influence within the Antarctic Treaty system, and establish impactful processes for engaging with stakeholders and maximising the utility of research investment. Key researchers within this project currently research Antarctic human activity, historical records, and governance and policy and have regularly presented the results from their research into the Antarctic Treaty system.

The approach

We will capture grey or hidden data through archival review and snowball sampling through existing networks of scientists. Raw information (from handwritten through to machine-readable text) will be ingested by leveraging our existing expertise in Google’s Tesseract OCR to automate the conversion of the raw information to analysable data.

To improve and validate transcription outputs, we will use a citizen-science approach or utilise Amazon’s Mechanical Turk service to assist with and validate the transcriptions. Contextual scientific data will be harmonised and added to existing relevant databases such as, GBIF, for biological data.

Exploring the resulting database using computational movement analysis and a variety of spatial data mining approaches enables understanding of how people undertake wayfinding in a unique environment (the McMurdo Dry Valleys) where only topography controls the way in which they move around the landscape.

Using the data generated from the analysis, we will develop and validate a novel agent-based model (ABM) to understand the likely spatial scale of historic movement and impact. Developing a stochastic agent-based simulation of each person within a field event using the belief, desire, and intention decision-making framework, the ABM will account for the factors that constrain (slope), facilitate (existing paths), and influence (protected areas) movement within the McMurdo Dry Valleys.

Validation for the ABM will come from in situ GPS measurements of human movement and detailed records of movement, collected from representative field events. The model will provide insight into the scale of the current and potential human footprint across the ice-free areas of the Ross Sea Region based on expected increases in activity.

Using the forecasted level of human activity and the analysis of a bioregion’s potential to become a habitat for non-native species, this analysis will also highlight the potential human vectors of non-native species introduction in the Ross Sea region.

Research aims

  • Identify, gain access to, scan, and/or collate raw scans of historic records of human activity within the McMurdo Dry Valleys, and develop computational workflow to ingest, process, convert (and validate) raw scans into machine readable text.
  • Using machine learning, network analysis, and other methods, undertake a spatial and temporal analysis of the data for insights, patterns, and behavioural rules into human activity.
  • Utilising the behavioural rules generated through analysis, develop (and validate using existing novel GPS datasets) an agent-based model to attempt to mimic wayfinding patterns.
  • Using the validated ABM to provide insight into the true scale of historic and current activities in Antarctica.
  • Using external data sources, explore the expected effects of current and future impacts of this activity.


  • Dr Fraser Morgan (Project Lead)
  • Professor Rebecca Priestley
  • Dr Pierre Roudier
  • Associate Professor Claire Postlethwaite
  • Professor Thegn Ladefoged
  • Associate Professor Priscilla Wehi
  • Kristin Wilson