Errors In Terrain-Based Model Predictions Caused By Altered Forest Inventory Plot Locations In The Southern Appalachian Mountains, USA

Abstract

Forest modeling applications that cover large geographic areas can benefit from the use of widely-held knowledge about relationships between forest attributes and topographic variables. A noteworthy example involves the coupling of field survey data from the Forest Inventory Analysis (FIA) program of USDA Forest Service with digital elevation model (DEM) data in attempts to explain how topographic characteristics influence forest productivity, vegetation composition, fire behavior, and other phenomena. Because U.S. federal law prohibits the release of actual FIA plot coordinates, only altered coordinates are released to the public. Here, terrain-based variables derived from a 10 m DEM using actual FIA plot locations were compared to those from altered plot locations in a region of the Southern Appalachian Mountains of western North Carolina, USA. Variables examined included simple terrain attributes such as percent slope and azimuth of aspect, and composite attributes such as terrain shape index, flow accumulation, slope position and forest site quality index. Results showed little correspondence between variables from altered plot locations compared to those derived using actual locations. Further, FIA field measurements of percent slope and azimuth of aspect showed little correspondence with corresponding DEM-derived estimates from the actual plot coordinates. In order to effectively link FIA plot data with DEM-derived topographic variables in mountainous regions like the Southern Appalachians, access to actual plot coordinates or terrain variables derived from them may be required.