Description: <DIV STYLE="text-align:Left;"><DIV><DIV><P STYLE="margin:0 0 11 0;"><SPAN><SPAN>One of the most fundamental information needs to support ecosystem assessment and land management planning is consistent, continuous, and up to date vegetation data of sufficient accuracy and precision. The Northern Region Existing Vegetation Mapping Program (VMap) database and map products help meet this information need and provides the Northern Region with a geospatial database of existing vegetation produced using consistent analytical methodology according to the Existing Vegetation Classification and Mapping Technical Guide (Brohman and Bryant, 2005) to support the Region 1 Multi-level Classification, Mapping, Inventory, and Analysis System, R1-CMIA (Berglund et. al., 2009). </SPAN></SPAN><SPAN><SPAN>The Idaho Panhandle and Kootenai National Forests (IPKNF)</SPAN></SPAN><SPAN><SPAN>VMap database provides four primary map products; lifeform, tree canopy cover class, tree size class, and tree dominance type to support mid and base-level analysis and planning. VMap uses the Region 1 Existing Vegetation Classification System (R1-ExVeg) (Barber, et.al. 2009) in its map unit design. The R1-ExVeg system describes the logic for grouping entities by similarities in their floristic characteristics. This has been an iterative process in Region 1 as different classification schemes have been tested and evaluated for utmost utility by end users. The system was designed to allow consistent applications between Regional inventory and map products within the R1-CMIA framework. VMap is a remote sensing derived product. As such, it uses a combination of airborne imagery and a nationally available digital elevation model (DEM). The </SPAN></SPAN><SPAN><SPAN>IPKNF is located in a region that is often and persistently cloaked in clouds. In addition to being obscured by cloud cover, the area of interest was also obscured by forest fire smoke in 2015. For these reasons, high resolution NAIP imagery was not available with full coverage for the area. Thus, in order to obtain contemporary and full coverage imagery of the mapping area, Rapid Eye high resolution satellite data was sourced between July 18 and August 8 to capture exiting and relevant vegetation patterns. The imagery was delivered with 5 meter pixel resolution, and five spectral bands of radiometric resolution, including red, green, blue, and infrared components. However, even with a custom collection of image data, cloud cover was still present. To reveal could obscured areas, cloud patches in the Rapid Eye data were masked, and coded as no data. Those areas of no data were then supplemented with cloud free Landsat 8 data. In 2016, no entirely cloud free Landsat data were available either. Nonetheless, a full area composite Landsat 8 scene was assembled with image data captured between June 4 and August 16, 2016. Areas obscured by clouds in this dataset were substituted with could free data acquired June 16, 2015. Finally, a composite of could free Rapid Eye and Landsat 8 </SPAN></SPAN><SPAN><SPAN>image data, </SPAN></SPAN><SPAN><SPAN>were</SPAN></SPAN><SPAN><SPAN>put through a process of aggregation to derive spatially cohesive units (i.e., polygons)</SPAN></SPAN><SPAN><SPAN>, that ultimately resemble stand boundaries</SPAN></SPAN><SPAN><SPAN>.</SPAN></SPAN></P><P STYLE="margin:0 0 11 0;"><SPAN><SPAN>In the field, reference information is collected and used to make spatial predictions of the vegetation attributes contained in the database. Predicted raster surfaces of the attributes are summarized to the delineated polygons.</SPAN></SPAN></P><P STYLE="margin:0 0 11 0;"><SPAN><SPAN>Draft map products are then reviewed and appropriate changes are made in the labeling algorithms. Final results are then used to populate the VMap database. An accuracy assessment was conducted to provide a validation of the data, giving an indication of reliability of the map products, so that managers are fully informed throughout the decision making process. Estimates of overall map accuracy and confidence of individual map classes can be inferred from the accuracy assessment error matrix derived from the comparison of known reference sites to mapped data. These accuracy assessment results are relevant to the entire </SPAN></SPAN><SPAN><SPAN>IPKNF</SPAN></SPAN><SPAN><SPAN>as a whole ranging from 60-90%, depending on the </SPAN></SPAN><SPAN><SPAN>particular </SPAN></SPAN><SPAN>attribute.</SPAN></P></DIV></DIV></DIV>