Metadata: National Wetland Inventory for Minnesota

National Wetland Inventory for Minnesota

This page last updated: 06/19/2019
Metadata created using Minnesota Geographic Metadata Guidelines

Go to Section:
  1. Overview
  2. Data Quality
  3. Data Organization
  4. Coordinate System
  5. Attributes
  6. Distribution - Get Data
  7. Metadata Reference

Section 1: Overview

Originator:Minnesota Department of Natural Resources

Title: National Wetland Inventory for Minnesota

Abstract: National Wetland Inventory (NWI) data for Minnesota provide information on the location, extent, and type of Minnesota wetlands. Natural resource managers use NWI data to improve the management, protection, and restoration of wetlands. Wetlands provide many ecological benefits including habitat for fish and wildlife, reducing floods, recharging, improving water quality, and supporting recreation.

These data were updated through a decade-long, multi-agency collaborative effort under leadership of the Minnesota Department of Natural Resources (MNDNR). Major funding was provided by the Environmental and Natural Resources Trust Fund.

This is the first statewide update of the NWI for Minnesota since the original inventory in the mid-1980s. The work was completed in phases by dividing the state into five project areas. Those project areas have all been edgematched into a final seamless statewide dataset.

Ducks Unlimited (Ann Arbor, MI) and St. Mary’s University Geospatial Services (Winona, MN) conducted the wetland mapping and classification under contract to the MNDNR. The Remote Sensing and Geospatial Analysis Laboratory at the University of Minnesota provided support for methods development and field validation. The DNR Resource Assessment Office provided additional support for data processing, field checking, and quality control review.

The updated NWI data delineate and classify wetlands according to the system developed by Cowardin et al. (1979), which is consistent with the original NWI. The updated data also contain a simplified plant community classification (SPCC) and a simplified hydrogeomorphic (HGM) classification. Quality assurance of the data included visual inspection, automated checks for attribute validity and topologic consistency, as well as a formal accuracy assessment based on an independent field verified data set. Further details on the methods employed can be found in the technical procedures document for this project located on the project website ( ).

DOWNLOAD NOTE: NWI data are only provided in either ESRI File Geodatabase or OGC GeoPackage formats. A Shapefile is not available because the size of the NWI dataset exceeds the limit for that format. If you are unable to use the File Geodatabase or GeoPackage, you can view data through Wetland Finder, an interactive mapping application on the DNR’s website ( ).

SYMBOLOGY NOTE: The ESRI File Geodatabase download includes four layer files that symbolize the data using four different wetland classification systems. The symbology layer files for the Cowardin class and the simplified HGM class are grouped into a smaller number of classes than the full elaborated classifications. Detail is available in the Minnesota Wetland Inventory User Guide and Summary Statistics report ( ). The layer files for these data have been set up to restrict drawing of the data when zoomed out beyond 1:250,000 scale. This is, in part, to prevent problems with slow performance with this large dataset.

Purpose: NWI data support effective wetland management, protection, and restoration. The data provide a baseline for assessing the effectiveness of wetland policies and management actions. These data are used at all levels of government, as well as by private industry and non-profit organizations for wetland regulation and management, land use and conservation planning, environmental impact assessment, and natural resource inventories.

Time Period of Content Date: 05/23/2019

Currentness Reference: These data correspond to the ground conditions at the time of the base imagery acquisition (2009 – 2014). Time Period of Content Date indicates when dataset was last updated.

Progress: Complete

Maintenance and Update Frequency: Annually

Spatial Extent of Data: Statewide

Bounding Coordinates: -96.862

Place Keywords: Minnesota

Theme Keywords: Wetlands, Deepwater habitats, Surface water, Swamp, Marsh, Bog, Fen, Cowardin, NWI, National Wetland Inventory

Theme Keyword Thesaurus: ISO 19115 Category

Access Constraints: None

Use Constraints: None

Contact Person Information: Steve Kloiber, Wetlands Monitoring Coordinator
Minnesota Department of Natural Resources
Minnesota Department of Natural Resources
St. Paul, MN  55155-4025
Phone: 651-259-5164

Browse Graphic: Click to view a data sample.

Associated Data Sets:

Section 2: Data Quality

Attribute Accuracy: Attributes for the draft data were checked by using visual inspection as well as automated verification routines. Mapping contractors conducted an internal review of all draft data with a senior photo-interpreter before providing the data to the DNR. The DNR reviewed the draft data and also provided access to project stakeholders to review and comment on the draft data via an online review tool developed specifically for this project. Attribute errors were forwarded to the mapping contractor to address. The final draft data were run through a series of automated checks in the US Fish and Wildlife Service (USFWS) data validation tool. These includes checks for valid attributes, same-adjacent attributes, and area checks for lake and pond attributes. Any errors found through these checks were addressed by the mapping contractor.

The final data were tested for accuracy by comparing them to an independent set of validation points. Validation points were selected by a stratified-random sampling process, classified through a field review conducted by the University of Minnesota, and reviewed by the DNR using 0.5-meter resolution imagery, 1-meter LiDAR derived digital elevation models, and other ancillary data. The mapped wetland class was associated with the validation class using a spatial join process in ArcGIS (points to polygons). The results of this accuracy test on the final data are summarized below. A summary is provided for the statewide dataset as well as the individual project areas.

Wetland-Upland Classification Accuracy (level one): Points within 3 meters of a wetland boundary were excluded to prevent potential positional uncertainty in the field GPS and base aerial imagery from leading to apparent errors in classification accuracy.
Wetland Cowardin Classification Accuracy (level two): Points within 3 meters of a class boundary were excluded along with all upland points.
The producer's accuracy, the user's accuracy, and the overall accuracy were calculated for the level one (wetland-upland) classification. The level-two (Cowardin class) assessment calculated the overall accuracy.

Level One: Wetland-Upland Accuracy
Number of validation points: 6957
Wetland Producer's Accuracy: 88%
Wetland User's Accuracy: 95%
Overall Wetland-Upland Accuracy: 91%

Level Two: Wetland Cowardin Class Accuracy
Number of validation points: 6679
Overall Class Accuracy = 74%

Many of the classification differences between the two datasets were associated with confusion between the L1 system (limnetic) and L2 system (littoral) as well as confusion between aquatic bed (AB) and unconsolidated bottom (UB) classes. The limnetic-littoral boundary is generally defined by the 2.5-meter depth contour by the Cowardin system; however, data on water body depth is not universally available, so differentiating between these systems is difficult to do with precision. The difference between AB and UB systems is based on the presence of macrophytic aquatic vegetation. Differentiation between these two classes is confounded by the dynamic nature of vegetation patterns.
These two classes of error can be excluded from the analysis by consolidating those classes. Because this involves dissolving some class boundaries, the number of points used in the analysis increases to 6698. If this consolidated class assessment is used, the overall photo-interpreted wetland classification (level two) accuracy for the statewide NWI would be 81% (Consolidated Class Accuracy).

Level One: Wetland-Upland Accuracy
CentralEast CentralNortheastNorthwestSouth
Number of Validation points169486080211532448
Wetland Producer's % Accuracy8689849390
Wetland User's % Accuracy9496979494
Overall Wetland-Upland % Accuracy8689869395

Level Two: Wetland Cowardin Class Accuracy
CentralEast CentralNortheastNorthwestSouth
Number of Validation points160080476211152398
Overall Class % Accuracy6774697884
Consolidated Class % Accuracy7684718488

Logical Consistency: Polygon and chain-node topology are present. Every polygon has a label.

Completeness: This data set represents the extent of wetlands and deepwater habitats that can be determined with the use of remotely sensed data and within the timeframe of the collected imagery. The target map unit (TMU) for this effort is to map all wetlands and deepwater habitats larger than ½-acre. Features smaller than ½-acre, but larger than 1/20-acres are occasionally mapped. Wetlands smaller than 1/20-acre are not generally mapped. The accuracy of image interpretation depends on the quality of the imagery, the experience of the image analysts, the amount and quality of the collateral data, and the amount of ground truth verification work conducted.

Horizontal Positional Accuracy: The base imagery used to develop the NWI data has a horizontal RMSE of 1.53 meters for the east-central region, 2.01 meters for the northeast region, 0.58 meters for the central and northwest region, and 0.82 meters for the southern region. Wetland boundaries typically exist along a gradient of hydrology, soils, and vegetation and determining precise locations can be difficult even in the field. A limited examination of well-defined feature boundaries (e.g. edge of water) indicate that the positional accuracy for these well-defined features is within ±15 meters of the true horizontal position.

Vertical Positional Accuracy: Not Applicable

Lineage: Since two separate vendor partners were responsible for the NWI update mapping and classification, there are some differences in methodology, but the end result is generally consistent. Ducks Unlimited completed work on the Central, East-Central, and Northeast project areas while St. Mary's University completed work on the Northwest and South project areas.

If a line were roughly drawn diagonally across the state from the northwest to southeast, north and east of that line relied more on semi-automated methods for classification while south and west of that line used a combination of automated and manual classification techniques. Both are appropriate and yield quality results but there may slight variations in the final product based on those differences in technique.

After the final data was produced for each of the five different project areas, the data were edge-matched into a single statewide database and an additional round of quality control was undertaken to ensure consistency across the dataset. Issues addressed included:
1) Inconsistent application of Temporarily Flooded (A) and Seasonally Flooded (C) Water Regimes for floodplains between project areas. Floodplain wetlands were reviewed and recoded by project region.
2) There was inconsistent application of the Intermittently Exposed (G) and Permanently Flooded (H) Water Regimes. All G Water Regimes were changed to H Water Regimes.

Once the statewide dataset was complete, the following systematic changes were addressed in the final product:
1) During the project, the Federal standard for Cowardin Classification of saturated water regimes changed. All Seasonally Saturated (B) Water Regimes were changed to Continuously Saturated (D) Water Regimes.
2) The DNR standard uses special modifier of ‘q’ for bogs/peatlands and is stored that way on the data available on the Minnesota Geospatial Commons. For inclusion in the Federal dataset, this code was converted to a ‘g’ to comply with the US Fish and Wildlife Service code schema.
3) For the Hydrogeomorphic Classification Water Flow Path, all isolated (IS) codes were merged with vertical (VR)

For more detailed information on the process for each project area, please see the technical documentation:

Input Data for all Project Areas
• Farm Service Agency NAIP Imagery – 1 meter (see project area detail for years used)
• FEMA Floodplains
• MN DNR LiDAR Data (see project area detail for resolution)
• MN DNR Minnesota County Biological Survey Native Plant Communities
• MN DNR Public Water Inventory
• MN DNR Spring Aerial Imagery – 0.5 meter (see project area detail for years used)
• MN DNR Streams
• USFWS National Wetland Inventory
• USGS National Hydrography Data

Additional Input Data
• Farm Service Agency NAIP Imagery 2008, 2009, 2010, 2013, 2015 - 1 meter
• MN DNR Forest Stand Inventory
• MN DNR LiDAR Data - 3 meter
• MN DNR Spring Aerial Imagery 2013 and 2014 - 0.5 meter
• Alaska Satellite Facility - 10 meter
• Farm Service Agency NAIP Imagery 2008, 2009, 2010 - 1 meter
• Metropolitan Mosquito Control District Wetland Data
• MN DNR LiDAR Data - 3 meter
• MN DNR Spring Aerial Imagery 2010 - 0.5 meter
• MN DNR Spring Aerial Imagery 2010 and 2011 - 0.3 meter
• USGS National Elevation Data - 10 meter
• Farm Service Agency NAIP Imagery 2008, 2009, 2010, 2013 - 1 meter
• MN DNR Forest Stand Inventory
• MN DNR LiDAR Data - 3 meter
• MN DNR Spring Aerial Imagery 2009 - 0.5 meter
• MN DNR Spring Aerial Imagery 2009 - 0.3 meter
• Farm Service Agency NAIP Imagery 2010, 2013, 2015, and 2017 - 1 meter
• MNDNR LiDAR Data - 1 meter
• MNDNR Spring Aerial Imagery 2013 or 2014 - 0.5 meter
• Farm Service Agency NAIP Imagery 2008, 2009, 2010 - 1 meter
• MNDNR LiDAR Data - 1 meter
• MNDNR Spring Aerial Imagery 2011 - 0.5 meter

Step 1 – Field Data Acquisition and Field Verification
Field training data for photo interpretation projects was acquired to guide the interpretation. The field training data served three purposes:
1) Provided field experience for interpreters updating the Minnesota NWI in identification and classification of local wetlands
2) Gathered images for use in a guidebook for wetland photo interpretation
3) Provided quality assurance data for review

Field verification sites were visited on separate occasions throughout each project area. A determination of the proper code was made by a consensus of the image interpreters. In cases of confusion, the MN DNR and USFWS NWI coordinator were consulted.
Sites were selected using the following criteria:
1) Representative of the variety of wetland types and imagery signatures
2) Spatially distributed across the entire project area
3) Consideration of site accessibility

Step 2 - Image Segmentation and Data Preprocessing
For the Central, East-Central, and Northeast project areas, Ducks Unlimited used the eCognition segmentation process, the inputs being a tiff layer stack and raw spring aerial imagery. The DEM and the LiDAR DEM derivatives were processed for each of the buffered HUC10 watershed boundaries in the project areas. These products were then mosaicked and clipped to the project area boundaries, and became input rasters for the layerstack. The final tiff layer stack consisted of the following layers:
1) DEM
2) DEM No Data Mask
3) Slope
4) Topographic Position Index (TPI)
5) Compound Topographic Index (CTI)
6) Average Elevation of 1st Returns
7) Average Intensity of Bare Earth Returns

For the Northwest and South project areas, the MNDNR Resource Assessment Office generated several derivatives from the LiDAR and soils data as an aid in photo-interpretation. These data were provided to St. Mary's University along with MNDNR spring aerial imagery and FSA summer aerial imagery.
Data derivatives included:
1) Percent Slope
2) Topographic Position Index (TPI)
3) Compound Topographic Index (CTI)
4) SSURGO Hydric Percentage
5) Contours
6) Hillshade
7) 1st Return Vegetation
8) Percent Canopy Cover
9) HPI - Hydrographic Position Index
10) Hydro Break Lines in certain northwest counties

Step 3 - Random Forest Classification
For the Central, East-Central, and Northeast project areas, Ducks Unlimited took the output of the segmentation process consisting of two shapefiles; a polygon shapefile of the segments and a point shapefile of the centroids of the polygons. The polygons and points were related using a unique identification number (ID). The point file contained all of the descriptive information from the polygon segments and was used as the input into the random forest classification. Topology was built for the image segments and any issues were corrected. Additional fields (attribute, comments, field verified) were added to the image segments for the photo interpretation process. A random forest classification was run using the point file and training data. The random forest classification classified each point based on the training data and assigned a confidence value to each classification. The resulting classification was then joined to the polygons using the unique ID.

Step 4 - Photo Interpretation, Wetland Delineation, and Classification
For the Central, East-Central, and Northeast project areas, Ducks Unlimited photo interpreters viewed the segments over the spring imagery to identify wetland segments. The photo interpreters used the spring imagery, professional knowledge, photointerpretation guide, as well as the summer imagery to assign the NWI code. Additional data layers (e.g. USGS DRG, SSURGO soils, DEM and other LiDAR-derived products) were also available to assist with the NWI classification. Adjacent segments of the same class were merged. Segments that have multiple wetland classes or combine wetland and upland classes were cut into separate polygons to conform to the NWI class boundary. Each watershed of the USGS HUC 10 tiling scheme was interpreted systematically until the entire area had been completed.

For the Northwest and South project areas, St. Mary's University worked in county-level tiles, where photo-interpreters systematically panned through the imagery looking for identifiable wetland signatures. Wetland boundaries were digitized on-screen primarily using the spring aerial imagery and LiDAR data in ArcGIS. Each polygon was assigned a wetland classification using a variety of imagery and other ancillary data. Once a work area (county tile) was completed, the attribute for the simplified HGM class was added.

Step 5 - Initial Quality Control
For the Northwest and South project areas, photo-interpreters performed frequent self reviews, checking for errors. This was followed by review by a QAQC specialist. The data were checked to ensure only allowable wetland classes were assigned. Features smaller than the minimum mapping unit were deleted or merged. Adjacent polygons with the same class were merged. The entire data set was visually inspected at 1:10,000 scale to look for classification errors using a system of signature matching. Linework was reviewed at 1:5,000 scale. An automated topology check was performed to identify and fix errors such as gaps and overlaps.

Step 6 - Draft Data Review
The draft version of the NWI classification were posted via web-mapping service for the MNDNR Resource Assessment Office, MNIT at DNR, the USFWS Regional NWI Coordinator, and other project stakeholders to review. Review comments were forwarded to Ducks Unlimited or St. Mary's University to address. Additional field checks were performed for selected sites to provide additional feedback to the photo-interpreters.

Step 7 - Final Processing
Once the review was complete and any required edits made, the data from the county tiles were edge-matched to create a seamless coverage for the entire project area. The NWI verification tool from the USFWS was run on the data repeating many of the automated checks previously performed. All identified errors were fixed and then the final validation was performed. Final data were posted to the MN Geospatial Commons and to the USFWS Wetlands Mapper.

Section 3: Spatial Data Organization (not used in this metadata)

Section 4: Coordinate System

Horizontal Coordinate Scheme: Universal Transverse Mercator

UTM Zone Number: 15

Horizontal Datum: NAD83

Horizontal Units: meters

Vertical Datum:

Vertical Units:

Depth Datum:

Depth Units:

Cell Width: 0.000100

Cell Height: 0.000100

Section 5: Attributes


Detailed Citation:

Table Detail:
Statewide NWI - Reference: Cowardin et al. 1979 (U.S. Fish and Wildlife Service)
Field NameValid ValuesDefinitionDefinition Source
Internal feature numberESRI
Feature geometryESRI
Alphanumeric code identifying the wetland classification of the polygon using the system described by Cowardin et al. (1979)
General description of the wetland classification based on the Cowardin classification
Area of the polygon in acresCalculated in an UTM zone 15 projection using ESRI's geometry calculator.
Hydrogeomorphic classification code based on a simplified version of the system described by Tiner (2003)
Full text description of the hydrogeomorphic classification
SPCC_DESCenumeratedSimplified Plant Community Classification based on a simplified version of the system described by Eggers and Reed (2011). This field is used as symbology for the 'NWI Simplified Plant Community Classification' layer file.
Seasonally Flooded Basin
Wet Meadow
Shallow Marsh
Deep Marsh
Shallow Open Water Community
Non-Vegetated Aquatic Community
Coniferous Wetland
Hardwood Wetland
Shrub Wetland
Artificially Flooded
COW_CLASS1enumeratedWetland class from the Cowardin system without the water regime and special modifiers. This field is used as symbology for the 'NWI Cowardin Classification' layer file. In the layer file, AB and EM2 are combined in a single 'Aquatic Bed/Nonpersistent Emergent Vegetation' category.
ABAquatic Bed
EM1Persistent Emergent Vegetation
EM2Nonpersistent Emergent Vegetation
RBRock Bottom
RSRocky Shore
SBStreambed (Intermittent)
UBUnconsolidated Bottom (Open Water)
USUnconsolidated Shore (Banks & Sandbars)
CIRC39_CLASSenumeratedWetland classification based on the system describer by USFWS Circular 39. This field is used as symbology for the 'NWI USFWS Circular 39 Classification' layer file.
1Seasonally Flooded Basin or Flat
2Wet Meadow
3Shallow Marsh
4Deep Marsh
5Shallow Open Water
6Shrub Swamp
7Wooded Swamp
80Municipal and Industrial Activities
90Riverine Systems
HGM_SYMBOLenumeratedSimplified Landscape Position category based on the hydrogromorphic classification. This field is used as symbology for the 'NWI Simplified HGM Classification' layer file.MNDNR
Mineral Flat
Length of feature in internal unitsESRI
Area of feature in internal units squaredESRI
Project Boundary - Boundaries representing each phase of the NWI data development (MNDNR)
Field NameValid ValuesDefinitionDefinition Source
project_areaenumeratedShort code identifier for project area
ECMNEast central
Descriptive name for project area
Organization who completed the wetland mapping and provided the project area data
Date project data were delivered to MNDNR
Year(s) of spring imagery used for data interpretation

Section 6: Distribution

Publisher: Minnesota Department of Natural Resources (DNR)

Publication Date: 04/30/2015

Contact Person Information: Zeb Thomas, GIS Data Systems Coordinator
Minnesota DNR - MIS/GIS Unit
500 Lafayette Rd
Saint Paul, MN  55155
Phone: 651-259-5637

Distributor's Data Set Identifier: water_nat_wetlands_inv_2009_2014

Distribution Liability: The Minnesota Department of Natural Resources General Geographic Data License Agreement is online:

Ordering Instructions: Visit the web site noted in the online linkage section, or send an email to the Distribution Contact listed in this metadata record

Online Linkage: I AGREE to the notice in "Distribution Liability" above. Clicking to agree will either begin the download process, link to a service, or provide more instructions. See "Ordering Instructions" above for details.

Section 7: Metadata Reference

Metadata Date: 06/19/2019

Contact Person Information: Steve Kloiber, Wetland Monitoring Coordinator
Minnesota Department of Natural Resources
500 Lafayette Road
St. Paul, MN  55155
Phone: 651-259-5164

Metadata Standard Name: Minnesota Geographic Metadata Guidelines

Metadata Standard Version: 1.2

This page last updated: 06/19/2019
Go back to top

Created with MGMG stylesheet version 2017.04.25.1