Urban Ecological Land – Cover Mapping for New York City

An urban ecological land-cover map containing 37 unique classes was created for New York City using object-based imagery analysis (OBIA) techniques in conjunction with multispectral orthoimagery, Light Detection and Ranging (LIDAR) data, and thematic Geographic Information System (GIS) layers. Based on a classification scheme adapted from the United States National Vegetation Classification (NVC), the new map included a mix of ecological and anthropogenic features mapped across four hierarchical levels of detail: 1) basic land cover; 2) land-cover  sub-classes; 3) NVC Group; and 4) NVC Association. All ecologically-relevant classes were mapped to the NVC Group level, and a subset of 9 classes were mapped to the NVC  Association level. An accuracy assessment conducted on the Level 2 map indicated an overall accuracy of 92%, a high classification rate attributable in part to the efficient mapping of widely-distributed upland forest classes. However, forested wetlands and other uncommon wetland features were mapped with much lower accuracy, suffering particularly from high rates of omission. Insufficient reference data were available to conduct a quantitative accuracy assessment for the Level 3 and Level 4 NVC classes, but the quality of these classes likely varied by data input (i.e., high for classes based on expert opinion, lower for classes based on spectral criteria). The current map will serve as baseline documentation for monitoring and protection of NYC’s most important ecological features, and subsequent versions could be materially improved by advances in wetlands mapping and additional field-based expert opinion.

• Author(s): Spatial Informatics Group
• Date Published: October 16, 2014