Individual based modelling to study the effect of retreating sea ice on arctic seals.

The Polar regions are particularly vulnerable to global warming and are therefore rapidly changing: it is expected that the Arctic Ocean is seasonally ice free before 2050. Many populations of Arctic pinnipeds (true seals, eared seals and walruses) are dependent on the sea-ice for pupping, resting and moulting and will consequently suffer when the sea-ice is absent. Monitoring these populations (to study the effects of climate change) is extremely challenging, due to the inaccessibility of the Arctic. The usage of remote sensing techniques – such as drone imagery and High-Resolution satellite imagery – in combination with Machine Learning and Computer Vision, can aid in conducting animal census in this vast and remote region. To this end, in this project we will develop and employ new methodologies to detect and count seals. Firstly, this will be done in the well-studied Dutch Wadden Sea, where many years of monitoring have provided a unique set of annotated training data. Secondly, similar approaches will be used on Arctic seals, where available data is scarcer.