An Automated Procedure for Extending the NWI Classification System for Wetland Functional Assessment in Minnesota, United States

Abstract There is increasing recognition for the need to evaluate wetland functions across broad geographic areas in support of strategic planning efforts for wetland protection and restoration. Such efforts rely heavily on geospatial data. The most comprehensive and widely available geospatial data for wetlands in the United States is the National Wetlands Inventory (NWI), but the standard attributes for the NWI have somewhat limited utility for functional assessments. Extending the NWI data with an alternative classification system based on hydrogeomorphic descriptors is one way to improve the utility of this data to perform landscape-scale wetland functional assessments, but adding this information has traditionally relied on a labor-intensive manual interpretation process. We developed an automated classification procedure for augmenting the NWI data with additional attributes to support wetland functional assessment. We applied this procedure to a large watershed located in southern Minnesota and compared the results to an independently developed classification based on traditional photo interpretation. The classification results for the automated method compared favorably to the results from the manual methods for landscape position (84%) and landform (85%). The agreement between our automated classification results and the manual classification exceeds the level of agreement reported by previous similar efforts. The level of agreement was lower for assignment of the water flow path (32%). Adding a secondary water flow path class increased the agreement somewhat (43%). The most important factor affecting the accuracy of the automated procedure was the quality of the geographic information systems input data.