CoastSat

A global shoreline mapping toolbox

Star

Get started! View Repository

Supporting article in Environmental Modelling and Software

Kilian Vos

Kilian Vos

Earth Observation Scientist

About me

I'm a data scientist and researcher working on projects in Earth Observation and Machine Learning. I use big data cloud computing (Google Cloud Platform/Google Earth Engine) to extract useful features from time-series of publicly available satellite imagery and build near real-time monitoring applications. I like to use image processing and machine learning to solve to complex problems, create cloud-native automated pipelines and deliver webGIS apps to visualise spatio-temporal data (Django, Leaflet, PostgreSQL/PostGIS).

I also have experience in time-series regression problems (LSTMs, autoencoders) and optimisation of decision-making problems (Markov Decision Processes) using Reinforcement Learning (Deep Q-learning, multi-agent).

Open-source projects: CoastSat, Coastal WebGIS Portal, SDS_Benchmark
, RL optimisation

Interests

  • 🐍 Python programming
  • 🛰️ Earth Observation
  • 📈 Machine Learning
  • 🌊 Coastal/Water Engineering
  • 🌏 WebGIS applications

Education

  • PhD in Remote Sensing / Coastal Engineering, 2022

    University of New South Wales (UNSW)

  • MSc in Environmental Engineering (Specialisation in Remote Sensing), 2017

    Ecole Polytechnique Fédérale de Lausanne (EPFL)

  • BSc in Environmental Engineering, 2015

    Ecole Polytechnique Fédérale de Lausanne (EPFL)

Code + Data

*

CoastSat

A global shoreline mapping toolbox with Google Earth Engine

WebGIS for coastal

A web dahsboard for satellite-derived coastal data

SDS_Benchmark

A testbed for evaluation shoreline mapping algorithms against benchmark datasets

RL Fleet maintenance

An RL framework for optimising the maintenance of a fleet of aircraft

CoastSat.slopes

Beach slope estimation from satellite-derived shorelines

Pacific shoreline dataset

Dataset of time-series of shoreline change around the Pacific Rim

Data - Slopes Australia

Dataset of beach-face slopes around Australia mapped from satellites

Skills

Python programming

PyTorch, Tensorflow, scikit-learn, scikit-image, geopandas, gdal, rasterio

Earth Observation

Monitoring pipelines, Image Classification, Feature Extraction

Time-series analysis

Anomaly detection, Frequency analysis, LSTMs

WebApps

Django, Leaflet, PostgreSQL, PostGIS, Oracle

Machine Learning

Reinforcement Learning, Deep Q-learning, Gradient Boosting

Awards

Best Paper Award Coastal Dynamics 2021

Kilian Vos has received the award for Best Paper for his outstanding presentation at Coastal Dynamics 2021, entitled: ENSO CONTROLS ON INTER-ANNUAL SHORELINE CHANGES AROUND THE PACIFIC RIM.
See certificate

1st Prize at Maxar Spatial Challenge

114 participants on 44 teams leveraged Maxar’s high-resolution satellite imagery to build innovative geospatial solutions to address Australia’s biggest challenges.
See certificate

Outreach & Talks

CoastSat and its applications

In this 20min Youtube video I describe CoastSat and showcase current applications of satellite-derived shorelines

Millions of satellite images reveal how beaches around the Pacific vanish or replenish in El Niño and La Niña years

This article in The Conversation explains in plain language our findings on the impact of ENSO on coastal change around the Pacific Basin.

The CoastSat shoreline mapping toolbox (recorded talk)

In this 20min Youtube video I present the open-source Coastsat toolbox which is capable of mapping satellite-derived shorelines from publicly available Landsat and Sentinel-2 imagery.

Beach slopes from satellite-derived shorelines (recorded talk)

How to estimate beach slopes in the absence of field measurements? Here is a 12min Youtube video where I present a novel technique to estimate beach slopes from satellite-derived shorelines.

Beach slopes from satellite-derived shorelines (slides only)

This work was presented at the Coast2Coast Webinar organised by Giovanni Coco, Kristen Splinter and Mitchell Harley.