For more information on the terms and concepts discussed here, see NASA's online remote sensing tutorial or the UNBC gislab homepage.
If you are looking for 2003 photos, please be patient. I'm working on it!
~Bob
|
|
Study Area |
 |
|
The research area subset is located northwest of Fort St. John, British Columbia. It frames the Besa-Prophet Pre-Tenure planning area a special management area of the Muskwa Kechika Management area. CLICK ON THE MAP TO VIEW THE STUDY AREA SUBSET It extends approximately from 57°11’ N to 57° 50’ N, which encloses the study area between the Sikanni Chief and Prophet Rivers. East-west extent ranges approximately from 122° 51’ to 124° 31’, stretching from the Alaska Highway to the headwaters of the Akie River. The borders of the subset have been extended beyond the border of the Besa-Prophet area to include known areas of large mammal migration. The subset covers three biogeoclimatic zones: Alpine Tundra, Spruce-Willow-Birch, and Boreal White and Black Spruce. Mammals found in the study area include the following: elk (Cervus elaphus), moose (Alces alces), white-tailed (Odocoileus viginianus) and mule deer (Odocoileus hemionus), Stone’s sheep (Ovis dalli stonei), mountain goats (Oreamnos americanus), caribou (Rangifer tarandus caribou), wolves (Canis lupus), and grizzly (Ursus arctos) and black bears (Ursus americanus) (Sims 1999). Planned development threatens to alter the landscape of the Besa-Prophet area. Baseline vegetation information is critical to current research efforts in the study area. It is the goal of this project to provide the most accurate, current, and temporally-dynamic vegetation information possible.
|
|
Project Objectives
|
Besa Prophet 3D Perspective CLICK ON THE IMAGE TO ZOOM TO REDFERN AND TRIMBLE LAKES |
|
The project uses Landsat TM data to generate landscape classifications that may be used to help predict the presence or movement of wildlife species in the study area. Multi-temporal image data (and derivatives such as vegetation indices) will show change over seasons due to natural changes in the reflectance of vegetation with time. The primary objectives of the research are as follows: 1) Refine and assess seasonal vegetation ‘greenness’ (onset, peak and duration of vegetation growth) values using multi-temporal (multi-image) vegetation metrics. 2) Assess the improvement of ecosystem classification with temporally-dynamic, satellite-derived vegetation information. 3) Accurately identify patterns of vegetation that have been recognized as related to the primary needs of ungulate species in the Besa-Prophet area, and report habitat potential as it relates to changes in vegetation type and phenology (onset, peak, duration, min, mean, and max greenness). |
|
|
Project Status | |
| Relative Radiometric Normalizaion Multi-date and/or multi-path images are often radiometrically normalized to remove the effects of non-surface related differences in radiance values and create a common radiometric response (Hall et al. 1991). Differences may be the result of solar zenith angle, sensor calibration, or the seasonal and daily differences in atmospheric conditions. Relative radiometric normalization techniques are being applied to the data so that seasonal differences in illumination and changes in atmospheric conditions do not confound change detection procedures. Successful normalization may allow change detection to reflect only actual changes in vegetation. If the outputs of the normalization techniques are deemed suitable, it will be assumed that vegetation features are directly comparable (i.e. sensor and atmospheric interference have been removed or minimized). |
||
| Vegetation Classification Field work was neccessary to collect information about the characteristics of the vegetation in the study area. However, the cost associated with these data-gathering trips is considerably less than what would need to be incurred with traditional mapping endevours. 227 sites have been visited. At each site, species and percent cover of all layers of vegetation were recorded. This information serves as 'training' data during the supervised classification. Data was organized from these plots into 28 discrete landscape types or classes. Using these groups as a guide, the classification process categorizes all pixels in an image into a discrete land cover classes, based on their unique spectral signature. See the UNBC GIS lab pages for a description of the supervised classifcation procedure, or the PCI geomatics help gateway for a description of the maximum liklihood algorithm. BELOW IS A COLOUR COMPOSITE OF LANDSAT BANDS 5,4&3. ROLLOVER THE IMAGE TO DISPLAY THE OUTPUT OF THE MAXIMUM LIKLIHOOD CLASSIFICATION. |
||
|
Classification accuracy will be assessed using a stratified random sampling scheme, as accuracy may differ within classes as well as among them.
|
||
| 'Greennes' Tracking Seasonal differences in vegetation can be monitered using vegetation metrics. Vegetation indices such as the NDVI have been generated on monthly Landsat TM images.
|
||
| Combining the temporal information of the monthly NDVI images and the supervised classification allows profiles (vegetation metrics) to be drawn for each class. The graph below shows the preliminary results for several vegetation classes. Each colour represents a different type of vegetation. The x-axis represents time, while the y-axis indicates the NDVI value. Note the dramatic dropoff in values after September.
This information can be displayed spatially to be incorporated into habitat suitablity assessments currently being used by other researchers. |
| Research Team | |
| Roberta Lay MNRES Candidate, University of Northern British Columbia BA (Geography) 2002 University of Northern British Columbia New
Lab 8-123 Roberta is most interested in the practical aspects of remote sensing –those techniques that can be directly applied in management planning and/or decision making. |
 |
Dr. Roger Wheate Associate
Professor/GIS Lab Coordinator (Geography), University of Northern
British Columbia New
Lab 8-307 Dr.
Wheate's interests cover the application of remote sensing and GIS
across the spectrum of NRES (Natural Resource and Environmental Studies)
faculty areas. His main focus lies in the integration of the geomatic
sciences, cartographic output, feature extraction and terrain visualisation;
special interests include mountain cartography / and glacier mapping
using remote sensing. |
 |
| |
|
| Dr. Janet Marsh |  |
This project has been made possible with funding from the Muskwa-Kechika Trust Fund.