You are here:
Tying Together Methods to Estimate Wetland Connectivity: Tests within the Pipestem, ND
Citation:
Christensen, J., G. Evenson, M. Vanderhoof, S. Leibowitz, H. Golden, AND L. Alexander. Tying Together Methods to Estimate Wetland Connectivity: Tests within the Pipestem, ND. AWRA Spring Conference, Snowbird, UT, May 01 - 03, 2017.
Impact/Purpose:
Different methods currently being investigated to determine the degree of surface connection between wetlands
Description:
Surface connectivity of wetlands in the Prairie Pothole Region (PPR) can occur through fill-spill and fill-merge mechanisms, with some wetlands eventually spilling into stream/river systems. These wetland-to-wetland and wetland-to-stream connections vary both spatially and temporally in PPR watersheds. The likelihood of spatial connection is influenced by the topology, the spatial distribution, and the distance between waterbodies. Temporal connectivity varies both seasonally, as fill-spill may be driven by spring snow-melt or by late summer rains, and inter-annually, driven by multi-decadal drought-deluge cycles which fill, merge or draw down wetlands over multiple years. Understanding the spatial and temporal patterns and frequency of connecting events is important in order to better frame and map the continuum of aquatic connectivity in PPR watersheds. We investigated spatial and temporal connectivity patterns in a PPR watershed, the Pipestem Creek, North Dakota, using two different methods: a long-term dataset of remote sensing and a detailed process model across a drought to deluge cycle (1990-2011). Remote sensing via Landsat with a 30 m resolution provides snap shots across the long-term drought deluge cycle that accurately identifies landscape scale wetland expansion and contraction patterns. Due to the resolution constraints, Landsat struggles to identify very small wetlands or linear water features that may connect wetlands and streams and may miss critical seasonal events when cloud cover is more likely. Meanwhile, process models such as the Soil and Water Assessment Tool (SWAT) model can be modified to account for spillage from small wetlands and linear connections and it simulates a continuous temporal record instead of relying on snapshots through time. However, SWAT is also parameter-rich and computationally intensive and thus is difficult to process over broader landscape scales. In this study, 19 cloud-free Landsat images from the 22 yr drought to deluge time period were processed to identify the percent of water inundation across the watershed. A modified SWAT model was developed for the Pipestem watershed and run over the 22 yr period to simulate fill-spill processes of individual wetlands. The percentage of wetland spillage events to neighboring wetlands, spillage events to streams, and the contributing area to streams were recorded for time windows encompassing the Landsat dates. We compare and discuss the various SWAT model fill-spill outputs to the Landsat percent saturation measurements during times of drought and deluge in the watershed. By finding commonalities in connections between Landsat and SWAT, we can improve our confidence in both approaches when mapping wetland connectivity. In parallel, where the two methods diverge in describing wetland connectivity, these instances help us better understand the limitations of the approaches and potentially can be used to improve the complete picture of connectivity in the PRR. Ultimately, combining both the methods of remote sensing and process modeling allows us to visualize both the spatial and temporal continuum of connections across the watershed, the limitations of the approaches and how we can study and map wetland connectivity in other areas of the PPR and beyond.
