Evolution of wetland monitoring from inventory to functional assessment and modelling: A case study from a US catchment

Yeo I., Lang M., Lee S., Huang C., Yetemen Ö.

22nd International Congress on Modelling and Simulation: Managing Cumulative Risks through Model-Based Processes, MODSIM 2017 - Held jointly with the 25th National Conference of the Australian Society for Operations Research and the DST Group led Defence Operations Research Symposium, DORS 2017, Hobart, Australia, 3 - 08 December 2017, pp.1537-1543 identifier

  • Publication Type: Conference Paper / Full Text
  • Doi Number: 10.36334/modsim.2017.l1.yeo
  • City: Hobart
  • Country: Australia
  • Page Numbers: pp.1537-1543
  • Keywords: Inundation, Remote sensing, Spatial data, Wetland classification, Wetland modelling
  • Istanbul Technical University Affiliated: Yes


Wetlands provides important ecosystem services. Despite increasing recognition for their importance as a natural resource, wetlands are continuously facing a high risk of loss and degradation. Geographically isolated wetlands (GIWs) and headwaters are most vulnerable. They are temporary water ways with relatively small size. They can be easily filled to support other uses without permits and generally unprotected under the legal framework in the US [US EPA, 2015]. However, there is increasing pressure to consider cumulative influence of certain types of GIWs on downstream water. While debates on protecting temporary waterways continue, improving understanding on the wetland hydrology and stream-wetland connection is crucial. This requires reliable information to update the status and changes on wetlands and rapid assessment tool to investigate the strength and frequency of the wetland connectivity to other water bodies. The U.S. Fish and Wildlife Service (US FWS) provides information on the extent and status of wetlands, through National Wetlands Inventory (NWI) [Cowardin & Golet, 1995]. The NWI is a spatial dataset that features wetlands and deep water habitats in consistent standardized ecological classification. However, the NWI, similar to other regional land use maps, is a categorical map. It does not provide information on inundation extent. Inundation is highly dynamic and can vary remarkably time to time, in response to multiple drivers and the local hydrological condition. It is a key factor controlling the ecological functioning of a wetland. In this study, we first demonstrated a practical and effective regional framework to develop long-term wetland inundation record. Using Landsat time records and airborne Light Detection and Ranging (LiDAR) intensity data, we generated a set of temporally consecutive maps of subpixel water fraction (SWF). The SWF maps indicate the percent of surface water within every 30-m Landsat pixel at an annual time basis over 1985-2011. They can provide crucial information on change dynamics and inundation extent of wetlands. When the mapping was demonstrated for the Coastal Plain of the Chesapeake Bay Watershed (CBW), comprehensive accuracy assessments of the SWF maps resulted in an estimated root mean square error (RMSE) of 7.78% for open water area. Moreover, a separate accuracy assessment targeting inundation in wetlands (i.e. presence or absence of water) yielded an overall accuracy of 93%. These results indicated that Landsat data can be calibrated to accurately extract long-term water information at the regional scale. We then demonstrated how SWF maps and NWI can be used to assess the cumulative impacts of headwater wetlands on downstream water, and how such data could assist us to overcome the challenges in modelling wetland hydrology and assessing the hydrological connection to downstream water at the local landscape level. The study was conducted on the coastal catchment in the upper region of Choptank River in CBW. The study area included a dense network of wetlands, which made up for ~ 30 % of the catchment area in aggregate. When assessed at the local scale, it was evident that the SWF maps showed inundation changed in response to the weather variability, and the change trend was consistent with daily stream flow (r = 0.81; p-value < 0.01) and base flow (r = 0.57; p-value < 0.1). Furthermore, the change patterns followed the characteristics of the hydrological regimes (i.e., hydroperiod, the seasonal variation of inundation) described by NWI. The catchment-scale, cumulative impacts of GIWs was further investigated using the catchment scale simulation model, Soil and Water Assessment Tool (SWAT), with improved wetland extension. Results showed significant, cumulative, catchment-scale hydrological impacts of GIWs. GIWs changed the partitioning of precipitation between actual evapotranspiration (AET) and stream flow, and the major transport pathway of water delivery into the stream flow. Wetland dominated catchment produced lower AET, but maintained higher streamflow mainly delivered by the groundwater. This study demonstrates the evolution of mapping and monitoring wetlands using remote sensing, and the progress toward modelling wetland function using improved water information and a catchment scale model.