| Autor: |
Bert Guindon, Ying Zhang, Xianfeng Jiao |
| Rok vydání: |
2015 |
| Předmět: |
|
| Zdroj: |
International Journal of Applied Earth Observation and Geoinformation. 38:193-203 |
| ISSN: |
1569-8432 |
| DOI: |
10.1016/j.jag.2015.01.007 |
| Popis: |
The landscape of Alberta’s oilsands regions is undergoing extensive change due to the creation of infrastructure associated with the exploration for and extraction of this resource. Since most oil sands mining activities take place in remote forests or wetlands, one of the challenges is to collect up-to date and reliable information about the current state of land. Compared to optical sensors, SAR sensors have the advantage of being able to routinely collect imagery for timely monitoring by regulatory agencies. This paper explores the capability of high resolution RADARSAT-2 Ultra Fine and Fine Quad-Pol imagery for mapping oilsands infrastructure land using an object-based classification approach. Texture measurements extracted from Ultra Fine data are used to support an Ultra Fine based classification. Moreover, a radar vegetation index (RVI) calculated from PolSAR data is introduced for improved classification performance. The RVI is helpful in reducing confusion between infrastructure land and low vegetation covered surfaces. When Ultra Fine and PolSAR data are used in combination, the kappa value of well pads and processing facilities detection reached 0.87. In this study, we also found that core hole sites can be identified from early spring Ultra Fine data. With single-date image, kappa value of core hole sites ranged from 0.61 to 0.69. |
| Databáze: |
OpenAIRE |
| Externí odkaz: |
|
Full Text from ScienceDirect