The potential of land cover classification techniques for conservation of Wedholme Flow, a lowland raised peat bog in Cumbria

Abstract

Land cover classification of a lowland raised bog, (Wedholme Flow) Cumbria was undertaken using remotely sensed fine spatial-resolution Ikonos data coupled with airborne LiDAR data. The aim of this work was to provide spatial data which could be used to aid future conservation of the site. Conventional Maximum Likelihood classification was compared to fuzzy object-oriented classification techniques to assess which technique is most suited to spatial assessment of semi-natural peat environments. Object-orientated classification using eCognition software was used to aggregate spatial and contextual information into a hierarchical configuration for land cover class description. This enabled investigation of morphological, topological and spatial structure of the site. This paper demonstrates that the technique can be used to distinguish between areas of extracted peat and original remnant bog surface and identify various levels of disturbance. The fuzzy nature of this classification enables assignment of each object within the image to several classes that best represents the gradual boundaries in semi-natural environments. The classification is used to recommend suitable restoration techniques for Wedholme Flow in comparison to current management and emphasises the importance of remote sensing in future scientific modelling.

More information

Identifiers

Catalogue record

Export record