The applicability of a small satellite constellation in classification for large-area habitat mapping: a case study of DMC multispectral imagery in west-central Alberta

The small satellite constellation (SSC) has the outstanding advantages of efficient operation and cost, global surveying, and increased revisit frequency, so that it has become the cutting edge of development in remote sensing in recent years. The Disaster Monitoring Constellation (DMC) is the first successful application of the concept of an earth-observation constellation. Its mission is to address the issue of large-area habitat mapping and to display the capability of SSC. The DMC provides 32 m spatial resolution imagery with one-day revisit frequency and 600 km swath width. The objective of this research was to demonstrate the applicability of SSC in large-area habitat mapping (i.e., by applying DMC multispectral imagery in the classification of grizzly bear habitat in west-central Alberta). An object-oriented classification method was selected to classify the DMC imagery into a hierarchical land cover classification scheme composed of three levels of detail. An error matrix was tabulated based on ground reference data to assess the accuracy of the DMC-based classification. In order to learn about the capability of DMC imagery as objectively as possible, the DMC-based and Landsat-based classifications were compared. The overall accuracies of DMC-based classification were 95.6% in the most general level I, and 82.4% in the most detailed level III classifications. However, the DMC-based classification was inferior to the Landsat-based classification in accuracy, and the difference was gradually widened from level I to level III. It is believed that the DMC multispectral imagery can be a good candidate for remotely sensed data sources to describe the physiognomy of the earth's surface with more than 80% accuracy, especially for large areas.