Citizen scientists and machine learning to monitor coastal erosion with drones: a statewide approach

Pucino N1, Allan B1, Carlvalho R1, Kennedy D2, Ierodiaconou D1

1Deakin University,

2The University of Melbourne


Beach topographic data has historically been obtained using many different techniques. Recently, Unmanned Aerial Vehicle based Structure From Motion (UAV-SfM) has gained considerable popularity in coastal research, being the best compromise in terms of cost, precision, reproducibility and simplicity. However, time or budget constraints often limit UAV-SfM surveys to be operated solely by professional researchers or commercial operators with obvious trade-offs in terms of monitoring geographical coverage and frequency of revisits.

In 2018, the Victorian Coastal Monitoring Program (VCMP), a government funded joint project led by Deakin University and the University of Melbourne, recruited, trained and engaged over 100 citizen scientists (non-technical volunteers) to autonomously use UAVs to map 16 beaches every 6 weeks for 3 years. This not only allows coastal communities to actively contribute to the study of their coast, but also produces an enormous in-flow of spatial data at an unprecedented temporal resolution and geographical coverage.

Here, I report the results from the first year of monitoring, focusing on how we validated Citizen Science (CS) data and combined advanced geostatistical analysis and unsupervised machine learning to effectively monitor 10 sandy beaches erosion/deposition cycles across Victoria. This project demonstrates how the CS protocol can be implemented to effectively upscale UAV-SfM beach surveys from the meso to the macro-scale, allowing government, academia and coastal communities to work together towards a better understanding of our beaches.


Nicolas Pucino could be defined as a geospatial coastal scientist. He obtained his B.Sc in Physical Geography from the University of Lausanne (Switzerland) and a M.Sc with distinction in coastal science from the University of Wollongong (Australia).

Currently, he applies different remote sensing and geospatial techniques to the study of multi-scale coastal erosion and beach sediment properties as part of his PhD at Deakin University, with the Deakin Marine Mapping group.

His main interests are in:

– Open Source geospatial analysis

– Beach sediment properties retrieval by integrating field and UAV-based hyperspectral signatures

– Large-scale coastal monitoring from optical satellites (multi and hyperspectral)

– Gesopatial implementation of Machine Learning algorithms