A sequence of Landsat TM/ETM+ scenes capturing the substantial surface water variations exhibited by prairie pothole wetlands over a drought to deluge period were analyzed in an attempt to determine the general hydrologic function of individual wetlands (recharge, flowthrough, and discharge). Multipixel objects (water bodies) were clustered according to their temporal changes in water extents. We found that wetlands receiving groundwater discharge responded differently over the time period than wetlands that did not. Also, wetlands located within topographically closed discharge basins could be distinguished from discharge basins with overland outlets. Field verification data showed that discharge wetlands with closed basins were most distinct and identifiable with reasonable accuracies (user's accuracy=97%, producer's accuracy=71%). The classification of other hydrologic function types had lower accuracies reducing the overall accuracy for the four hydrologic function classes to 51%. A simplified classification approach identifying only two hydrologic function classes was 82%. Although this technique has limited success for detecting small wetlands, Landsat-derived multipixel-object clustering can reliably differentiate wetlands receiving groundwater discharge and provides a new approach to quantify wetland dynamics in landscape scale investigations and models.
Bibliographical noteFunding Information:
This collaborative work was supported by the U.S. Geological Survey (USGS) Integrated Landscape Monitoring initiative. Funding was provided by USGS Central Region Integrated Studies Program (CRISP), Carbon Cycle Research (Land Remote Sensing), Climate Effects Network (Global Change Office), and USGS Earth Resources Observation and Science Center over-guide. We thank John Melack and Gabriel Senay for their reviews and Lei Ji for assisting with graphics. Use of brand names in this manuscript does not constitute nor imply endorsement by the U.S. Government.
- Cluster analysis
- Object-oriented image analysis
- Wetland classification