Assessing Landsat TM Imagery For Mapping and Monitoring Prairie Dog Colonies

At the turn of the 20th century, prairie dog colonies (PDCs) covered between 40 and 100 million hectares of the prairies of western North America (Marsh 1984, Anderson et al. 1986). However because of government eradication programs, habitat loss, and more recently Sylvatic plague epidemics, PDCs have experienced a decline of nearly 98% in the last 100 years. The goal of this research was to evaluate the effectiveness of Landsat Thematic Mapper (TM) imagery for directly mapping PDCs using a two-stage classification process. Specific objectives of this research were to: (1) determine how well PDCs can be distinguished from uncolonized prairie (UP); and (2) determine whether active PDCs can be distinguished from inactive PDCs, as a result of plague or other factors.
Research focused on a portion of one TM scene (Path 37/Row 27) in north central Montana (Figure 3). This study area covers part of the Charles M. Russell National Wildlife Refuge (CMR), Fort Belknap Indian Reservation, and the Bureau of Land Management’s (BLM) Phillips County Resource Area. The PDCs in this area have been extensively mapped and monitored by land management agencies since the late 1970's, and is currently the sites of black-footed ferret reintroduction efforts.
My research shows that PDCs can be mapped fairly accurately using Landsat TM imagery and this methodology; however, accurately distinguishing PDCs from IPDCs was impossible. The overall tendency of my classifications was to overpredict PDCs and IPDCs while consistently missing the very small colonies. Although commission errors were high, a good percentage of these errors were due to confusion between active and inactive PDCs. In addition, some of the commission error may be attributed to the correct identification of active PDCs, especially on the nearly 700,000 acres of private lands, much of which has never been surveyed for PDCs. The spatial resolution of Landsat TM imagery is quite adequate for the identification of PDCs, and any higher resolution imagery would, in my opinion, create more problems than benefits.

Project completion date: August 31,2002

For the complete thesis Click to download