Wetlands provide many ecological services, but are under threat from climate change and land modification amongst other stressors. The Ontario Ministry of Natural Resources (OMNR) currently uses the field-based Ontario Wetland Evaluation System (OWES) to assign scores to wetlands for planning and conservation purposes. These evaluations have been primarily from the field observer’s viewpoint, but were often augmented using analog air photo and/or digital ortho-photo interpretation. With such an approach overall spatial and temporal wetland dynamics were often overlooked or under represented.
This research evaluated attributes in four wetland complexes through three seasons using remote sensing data. Landsat 5 Thematic Mapper (TM, 30m) Radarsat-2 (8m), and WorldView-2 (0.5-2m) imagery were acquired, and several types of image metrics (e.g. vegetation indices, texture, and object metrics) were evaluated in mapping 14 OWES attributes. Differences were found in overall and specific class related accuracies for all 14 attributes of interest depending upon time of year, location, and/or data used. Eight attributes were successfully assessed using existing data or data developed using
the methods of this research. Scores derived for four of those attributes were equivalent to the OWES field-measured scores. Some general technical conclusions from the research were that high resolution optical imagery provided higher overall accuracies for most attributes of interest. Coarse resolution optical imagery had higher overall accuracy for the attribute Open Water Type. Radar-based variables did not improve overall accuracies, but the addition of these variables to the optical imagery object-based image analysis (OBIA) improved some individual class accuracies. With respect to
season of image acquisition, spring or summer imagery produced the highest accuracies. These results support an overview perspective with a top-down investigative approach for wetlands analysis in that they expose the inconsistencies and some consistencies in results between sites, between imagery types and/or at different times of year.