LIBRARY: VEGETATION LAYER NAME: VeGSIZE LOCATION: \GARCIA\USFS OVERVIEW of Vegetation Theme derived from Landsat TM Imagery 1994: (Provided by USFS) Vegetation databases have a minimum mapping unit of 1 hectare. There is strong reasoning for this. Most significantly, representatives of the Forest Service feel that vegetation characteristics being mapped can only be developed to a rough level at the 30 m pixel level using the methods and technologies we employ. These veg attributes are best characterized at the stand level, especially the structural components. Most people look at the landscape with an inherent recognition of landscape features which are generally much larger than a 30 m pixel. Its true that an aggregate of pixels can effectively say something about the vegetated landscape and that a pixel level product provides some analytical flexibility, but mapping to that resolution with the accuracy goals of the Forest Service is an inefficient way to map vegetation in their view. Also, for many types of analysis people inevitably aggregate the data to fit the scale of analysis. The providers of the data argue that aggregating data based on thematic similarity from the pixel level is problematic and ensures that error in the map labels translates to spatial error in the map. Without going into great detail about the mapping techniques, USFS adheres to two fundamental concepts: 1. Independently derived stand delineations that meet the minimum mapping unit but are of fine enough resolution to capture significant landscape and vegetation features. 2. A considerable amount of time is spent reviewing each of the veg attributes in the field and with resource photography to reduce map classification error. That translates to editing individual stands, which can only be done effectively if the analyst can clearly identify the location of the stand. Editing pixels is a tedious and frequently ambiguous process. It is very difficult to locate individual pixels on the ground or on photos. So...that means that any pixel level products produced by the Forest Service are not stand based, have not been reviewed and edited, and therefore can not reliably be used for analysis. BRIAN SCHWIND Remote Sensing Specialist, Vegetation Mapping Project Leader 1920 20th Street Sacramento CA 95814 PH (916)454-0805 FX (916)454-0820 bschwind@fs.fed.us DESCRIPTION: Ninety meter buffers were placed around all rivers and streams within the project area, using stream blue line data at 1:24k scale. Thus, the buffer data have a cross section diameter of 180 meters. The buffer coverage was converted to a grid and then used to clip the vegetation grid data to produce a grid layer of vegetation categories within 90 meters of all rivers and streams. It is important to note that when buffering stream data, where buffers overlap (where streams are less than 180 meters apart), islands of extraneous data are produced, and can cause serious errors when clipping vegetation data. These data islands must be flagged and omitted before clipping vegetation data with the buffers. VITAL STATISTICS: Datum: NAD 27 Projection: UTM Units: Meters 1st Std. Parallel: 34 00 00 (34.0 degrees N) 2nd Std. Parallel: 40 30 00 (40.5 degrees N) Longitude of Origin: -120 00 00 (120.0 degrees W) Latitude of Origin: 00 00 00 (0.0 degrees) False Easting (X shift): 0 False Northing (Y shift): 0 Source: USGS DLG Source Projection: Universal Transverse Mercator Zone 10 Source Units: Meters Source Scale: 30 meters For more metadata, see; \metadata\usfs\raster\eveg.rtf \metadata\usfs\raster\methods1.doc \metadata\usfs\raster\methods2.doc \metadata\usfs\raster\table_39a.rtf DOCUMENTATION DATES: November 2001 Vegetation layer clipped to fit project study area Contact Name: P. J. Trichilo Institute for Fisheries Resources 791 Eighth Street, Ste. N Arcata, CA 95521 (707) 822-9428