Wetland Cover Resources in the Prairie Pothole Region

The density and distribution of breeding waterfowl is primarily influenced by wetland class and abundance. Consequently, knowledge of the wetland resources in the Prairie Pothole Region
is critical in evaluating its importance to breeding waterfowl as well as other wetland dependent migratory birds.

During the late 1970’s, the USFWS initiated a National Wetland Inventory (NWI) program to map
the wetland resources in the United States. A wetland and deepwater habitat classification system was developed by Cowardin et al. (1979) and applied to wetlands photo-interpreted from high-altitude, color-infrared photography acquired by the National Aeronautics and Space Administration and the National High Altitude Photography Program. In the Prairie Pothole Region, images were acquired in the spring during the late 1970’s – mid 1980’s. Point, linear, and polygon wetlands were delineated onto mylar overlays and eventually transferred to USGS 7.5 minute quadrangle maps. Following protocols developed by the NWI, digital wetland information was produced by digitizing wetland boundaries from stable-base USGS quadrangle maps. Digital data has been available for Prairie Pothole Regions of North and South Dakota since the early 1990’s, and for Northeast Montana since 1998.

Process

The NWI classification system separated many wetland basins into several wetland habitat zones. Because information collected for wetland dependent biota generally relates to an entire basin as opposed to habitat regions within a basin, we reclassified wetlands to a simple “basin” class to complement these data. Guidelines for reclassifying NWI wetland polygons to basins was described by Cowardin et al. (1995). During the early 1990’s and again in 1999, HAPET integrated the Cowardin et al. (1995) guidelines into geographic information system (GIS) processes that reduced the > 250 NWI wetland habitats into 6 basin classes. The reclassification process primarily took advantage of NWI regime modifiers to reclassify NWI wetlands that comprised a basin to a single wetland basin reclassified to the deepest water regime. The end result was 6 basin classes that include temporary, seasonal, semipermanent, lake, and river. Point and linear wetland features were buffered to convert them to polygons.

Upland Land Cover Resources in the Prairie Pothole Region

During the mid-1990’s, a number of HAPET planning processes and evaluation activities were concluding that suggested that regional digital landcover data would be a valuable layer of information. Preliminary results of the Conservation Reserve Program (CRP)
Evaluation Study indicated that daily survival rates (DSR) of duck nests located in CRP was related to the amount of perennial cover in the surrounding 4-square mile landscape. Additionally, during the MAAPE planning process, protection of all existing native grassland was identified as an objective in each Wetland Management Districts. The development of landcover would allow HAPET to model predicted DSR for nesting waterfowl and quantify the distribution and acreage of remaining native grasslands.

In many states including Montana and South Dakota, the GAP program that was initiated by the
USFWS and eventually USGS, was producing landcover information, however, no landcover information was available for North Dakota and an evaluation of the utility of available GAP produced data was necessary.

Pilot Project
During summer 1996, the HAPET Office initiated a pilot project (using funds made available by the Mississippi and Central Flyways) to develop and test methodology for using Landsat Thematic Mapper (TM) Satellite Imagery (30 meter resolution) to classify perennial cover in central NorthDakota. Eight landcover classes were of interest and included 1) wetland, 2) native grassland, 3) undisturbed grassland (primarily CRP), 4) alfalfa hayland, 5) cropland, 6) forest/shelterbelt, 7) urban, and 8) barren. May 1993 and September 1995 image dates were acquired and ground data was collected for both training the classification process and conducting a subsequent accuracy assessment. Within class accuracies for the land covers of major interest were adequate and the decision was made to expand the landcover project to include all of east-river North and South Dakota. GAP landcover data was acquired from Montana for use in the Northeast Montana WMD.

Operational Project
In 1997, the Great Plains Regional Office of Ducks Unlimited joined the effort and partnered with the HAPET Office. Satellite images were purchased with funds provided by the Region 6 ARD for Refuges, the PPJV Coordinator, and Ducks Unlimited. Collection of ground data was shared. During the summers of 1997-99, 954,865 acres of ground data were collected (61% cropland, 25%grassland, 9% undisturbed grass, 3% hayland, and 1% each barren and forested. The ground data was divided into 2 sets with one half of the data used for training the classification process and the remaining one half used to conduct an accuracy assessment. Overall classification accuracy for a final mosaic of landcover exceeded 80%. Users accuracy for individual TM scenes ranged from 80-92%. Wetlands were classified from USFWS National Wetlands Inventory (NWI) data. A processing mask of wetland areas was incorporated into the raw imagery prior to the upland classification process.

A small acreage of open water was detected from the TM imagery outside of NWI wetlands. Landcover classes were cross-walked to conform to the National Vegetation Classification Standard to the formation level

Classification Results - Acres