Search the EPA Research Database

Project title

Habitat mapping, assessment and monitoring with high resolution imagery

Primary Funding Agency

Environmental Protection Agency

Co-Funding Organisation(s)

n/a

Lead Organisation

University of Dublin, Trinity College (TCD)

Lead Applicant

John Connolly

Project Abstract

iHabiMap will develop a dynamic software framework to facilitate assessment of data from several high spatial and spectral resolution optical sensors including RGB, multispectral (UAV and Sentinel-2), LiDAR and hyperspectral for mapping, assessing and monitoring three ecosystems. iHabiMap will assess the use of fixed wing and octocopter UAV using a variety of sensors including: RGB cameras; multispectral and hyperspectral sensors; and LiDAR data. COPERNICUS Sentinel-2 imagery will be used in parallel. We will focus on three Annex-I habitats: grassland; upland and coastal. The data will be classified and assessed using machine learning techniques including: Random Forest and Semantic Image Segmentation. These methods can give rapid and accurate image classifications. A key part of this project is the multi-disciplinary collaboration between geospatial scientists, data analytics and ecologists to train and test the machine learning classification using in-situ ground truth data. BEC ecologists have led many of the NPWS ecological surveys over the last decade and therefore deliver highly specific expertise to the project. The close collaboration between BEC, DCU and Orbas will enable detailed assessment of the accuracy of the machine learning approaches in classifying UAV derived imagery for mapping, assessing and monitoring remote habitats. The iHabiMap project with deliver several key outputs including: Peer-reviewed publications in the following domains: Remote Sensing, Land management and Ecology; Conference presentations � both oral and poster presentations will be delivered at national (Irish Earth Observation Symposium) and International (EGU) conferences. Several policy accessible documents will be produced including annual Policy Briefs (outline the projects progress and impact); A practitioner�s Manual/Guide outlining standard methodology for mapping habitats with UAVs in Ireland building on existing research; A synthesis report with clear non-technical summary and recommendations. A website and social media outputs will be also developed. Other outputs from the project will include infographics for popular dissemination on social media. It is envisaged that maps consistent with national mapping methodologies, (i.e. PRIME2 boundaries), will available in common formats and some UAV data may be released for academic research. The datasets derived in the project will be tagged with structured markup to increase their findability and deposited in the EPA-SAFER depository. This multi-disciplinary project with increase research capacity in Ireland by training a PhD with skill sets developed in geospatial analysis, machine learning and ecological survey techniques. Collaborative research capacity will also be developed among the partners in the research team.

Grant Approved

�484,037.08

Research Hub

Natural Environment

Research Theme

Natural capital and ecosystem services including soils and biodiversity

Start Date

19/03/2019

Initial Projected Completion Date

18/03/2023