On developing an unambiguous peatland classification using fusion of IKONOS and LiDAR DEM terrain derivatives – Victor Project, James Bay Lowlands
Bogs and fens, which comprise > 90% of the landscape near the De Beers Victor diamond mine, 90 km west of Attawapiskat, ON, provide different hydrological functions in connecting water flow pathways to the regional drainage network. It is essential to define their distribution, area and arrangeme...
Main Author: | |
---|---|
Format: | Master Thesis |
Language: | English |
Published: |
University of Waterloo
2011
|
Subjects: | |
Online Access: | http://hdl.handle.net/10012/5962 |
id |
ftunivwaterloo:oai:uwspace.uwaterloo.ca:10012/5962 |
---|---|
record_format |
openpolar |
spelling |
ftunivwaterloo:oai:uwspace.uwaterloo.ca:10012/5962 2023-05-15T15:33:17+02:00 On developing an unambiguous peatland classification using fusion of IKONOS and LiDAR DEM terrain derivatives – Victor Project, James Bay Lowlands DiFebo, Antonio 2011 http://hdl.handle.net/10012/5962 en eng University of Waterloo http://hdl.handle.net/10012/5962 Peatland Classification Geography Master Thesis 2011 ftunivwaterloo 2022-06-18T22:59:06Z Bogs and fens, which comprise > 90% of the landscape near the De Beers Victor diamond mine, 90 km west of Attawapiskat, ON, provide different hydrological functions in connecting water flow pathways to the regional drainage network. It is essential to define their distribution, area and arrangement to understand the impact of mine dewatering, which is expected to increase groundwater recharge. Classification was achieved by developing a technique that uses IKONOS satellite imagery coupled with LiDAR-derived DEM derivatives to identify peatland classes. A supervised maximum likelihood classification was performed on the 1 m resolution IKONOS Red/Green/Blue without the infrared (RGB) and with the infrared (IR_RGB) band to determine the overall accuracy prior to inclusion of the DEM derivatives. Confusion matrices indicated 62.9% and 65.8% overall accuracy for the RGB and IR_RGB, respectively. Terrain derivatives were computed from the DEM including slope, vertical distance to channel network (VDCN), deviation from mean elevation (DME), percentile (PER) and difference from mean elevation (DiME). These derivatives were computed at a local (15-cell grid size) and meso (250-cell grid size) scale to capture terrain morphology. The mesoscale 250-cell grid analysis produced the most accurate classifications for all derivatives. However, spectral confusion still occurred (regardless of scale) most frequently in the Fen Dense Conifer vs. Bog Dense Conifer classes and also in the Bog Lichen vs. Bog Lichen Conifer. Despite this confusion, by combining the larger scale LiDAR DEM derivatives and the IKONOS imagery it was found that the overall classification accuracy could be improved by 13%. Specifically, the DiME derivative combined with the multispectral IKONOS (IR_RGB) produced an overall accuracy of 76.5%, and increased to 83.7% when Bog Lichen and Bog Lichen Conifer were combined during a post hoc analysis. This classification revealed the landscape composition of the North Granny Creek subwatershed, which is divided ... Master Thesis Attawapiskat James Bay University of Waterloo, Canada: Institutional Repository Attawapiskat ENVELOPE(-82.417,-82.417,52.928,52.928) |
institution |
Open Polar |
collection |
University of Waterloo, Canada: Institutional Repository |
op_collection_id |
ftunivwaterloo |
language |
English |
topic |
Peatland Classification Geography |
spellingShingle |
Peatland Classification Geography DiFebo, Antonio On developing an unambiguous peatland classification using fusion of IKONOS and LiDAR DEM terrain derivatives – Victor Project, James Bay Lowlands |
topic_facet |
Peatland Classification Geography |
description |
Bogs and fens, which comprise > 90% of the landscape near the De Beers Victor diamond mine, 90 km west of Attawapiskat, ON, provide different hydrological functions in connecting water flow pathways to the regional drainage network. It is essential to define their distribution, area and arrangement to understand the impact of mine dewatering, which is expected to increase groundwater recharge. Classification was achieved by developing a technique that uses IKONOS satellite imagery coupled with LiDAR-derived DEM derivatives to identify peatland classes. A supervised maximum likelihood classification was performed on the 1 m resolution IKONOS Red/Green/Blue without the infrared (RGB) and with the infrared (IR_RGB) band to determine the overall accuracy prior to inclusion of the DEM derivatives. Confusion matrices indicated 62.9% and 65.8% overall accuracy for the RGB and IR_RGB, respectively. Terrain derivatives were computed from the DEM including slope, vertical distance to channel network (VDCN), deviation from mean elevation (DME), percentile (PER) and difference from mean elevation (DiME). These derivatives were computed at a local (15-cell grid size) and meso (250-cell grid size) scale to capture terrain morphology. The mesoscale 250-cell grid analysis produced the most accurate classifications for all derivatives. However, spectral confusion still occurred (regardless of scale) most frequently in the Fen Dense Conifer vs. Bog Dense Conifer classes and also in the Bog Lichen vs. Bog Lichen Conifer. Despite this confusion, by combining the larger scale LiDAR DEM derivatives and the IKONOS imagery it was found that the overall classification accuracy could be improved by 13%. Specifically, the DiME derivative combined with the multispectral IKONOS (IR_RGB) produced an overall accuracy of 76.5%, and increased to 83.7% when Bog Lichen and Bog Lichen Conifer were combined during a post hoc analysis. This classification revealed the landscape composition of the North Granny Creek subwatershed, which is divided ... |
format |
Master Thesis |
author |
DiFebo, Antonio |
author_facet |
DiFebo, Antonio |
author_sort |
DiFebo, Antonio |
title |
On developing an unambiguous peatland classification using fusion of IKONOS and LiDAR DEM terrain derivatives – Victor Project, James Bay Lowlands |
title_short |
On developing an unambiguous peatland classification using fusion of IKONOS and LiDAR DEM terrain derivatives – Victor Project, James Bay Lowlands |
title_full |
On developing an unambiguous peatland classification using fusion of IKONOS and LiDAR DEM terrain derivatives – Victor Project, James Bay Lowlands |
title_fullStr |
On developing an unambiguous peatland classification using fusion of IKONOS and LiDAR DEM terrain derivatives – Victor Project, James Bay Lowlands |
title_full_unstemmed |
On developing an unambiguous peatland classification using fusion of IKONOS and LiDAR DEM terrain derivatives – Victor Project, James Bay Lowlands |
title_sort |
on developing an unambiguous peatland classification using fusion of ikonos and lidar dem terrain derivatives – victor project, james bay lowlands |
publisher |
University of Waterloo |
publishDate |
2011 |
url |
http://hdl.handle.net/10012/5962 |
long_lat |
ENVELOPE(-82.417,-82.417,52.928,52.928) |
geographic |
Attawapiskat |
geographic_facet |
Attawapiskat |
genre |
Attawapiskat James Bay |
genre_facet |
Attawapiskat James Bay |
op_relation |
http://hdl.handle.net/10012/5962 |
_version_ |
1766363748005576704 |