White Mountain Apache Tribe Fort Apache Remote Sensing
MB&G teamed with Quantum Spatial to assist the White Mountain Apache Tribe in producing maps of existing vegetation types across the Fort Apache Reservation. The Existing Vegetation (EVT) Map was based upon 1 meter 2013 color aerial imagery as well as other geospatial data such as 2013 Landsat 8 imagery and ancillary GIS data.
The map supports forest management planning processes and other land management activities on the reservation. Object oriented imagery classification techniques were used to classify the imagery into 26 non-forest, woodland, and timberland vegetation strata, derived from 148 habitat types (1997, USDA SW Habitat Typing Guides). MB&G led the planning, coordination, and management of field data collection efforts. We utilized MB&G MobileMap, a proprietary mobile technology, and web based platforms (ArcGIS Online), to efficiently manage multiple field crews collecting data over many weeks in remote areas throughout the reservation. MB&G MobileMap helped achieve the field sampling goals, because it allowed access to high resolution imagery and vegetation segments. MB&G MobileMap, combined with web services, allowed project managers, team members, and the client to monitor field progress on a daily basis.
By leveraging these technologies MB&G was able to greatly exceed field sampling targets and ensure all EVTs were well represented across the landscape. The rich training dataset produced from the field sampling efforts provided an extraordinary foundation to build Classification and Regression Tree (CART) Models. CART models, which work best when given large amounts of training data, were then used to classify the entire 1.4 million acre project area. Overall accuracies for the final map were between 82 – 85%. Given the complexity of the classification system, these accuracies are considered very good for this level of cover type classification and the client was highly satisfied with all delivered products.