Coastal Wetlands

Data Download, Preparation, and Filtering

Data was downloaded from the NWI (https://www.fws.gov/wetlands/Data/State-Downloads.html) for each state encompassing the coastal United States. Data for a total of 25 states were downloaded. Each File Geodatabase of raw data from NWI inherently contains 4 feature classes including all wetlands data for that state, historical mapping information, metadata documentation as spatial data describing data compilation information for different regions, or projects within a state, and a dataset extent boundary.

The NWI feature class containing wetlands data was extracted for each state and reviewed separately for each step of the data preparation process, until final data compilation.

A number of the raw datasets contained errors within the spatial index. The spatial index for every dataset was reset as an initial processing step.

Coastal wetlands included in this dataset were defined by the NWI classification codes of system Marine (M) and system Estuarine (E). All sub-classifications below these two systems were included. Other systems were excluded from the coastal wetlands dataset.

In order to accommodate XY tolerances in meters as part of the geometry review, filtered data was projected to North_America_Albers_Equal_Area_Conic, EPSG 102008. This projection also ensured adherence to the NOAA standard of using GRS_1980.

Prepared data was saved in a new File Geodatabase, as a feature dataset by state. Data was stored as such in order to build topology as part of the geometry review.

Geometry Review

Topology was built within ArcGIS Pro version 2.8.3 for each state feature dataset separately, in order to identify and repair errors.

A series of tests was conducted to determine the optimal topology XY resolution parameter. Generally, this parameter should be no less than 0.001 meters. It was determined that resolutions tested that were greater than 0.1 meters (0.5 meters, 1 meter, 4 meters) significantly altered the data when validated. A resolution of 0.1 meters was found to be optimal for these data and used consistently across all project topologies.

Appropriate topologic rules for evaluating a single polygon dataset were applied to all state datasets. These rules were a) features must not overlap, and b) features must not contain gaps. Errors were then evaluated and repaired.

All overlaps were removed across all states.

A large number of gaps were identified for each state, as any portion of a feature that did not touch another feature was identified as a gap. Non-adjacent features are inherently a valid part of NWI wetlands, and are permitted, however smaller gaps and slivers as errors do also exist. Assessment of gaps was conducted initially to mark the vast majority of the large gaps as exceptions. This allowed smaller, gap errors to be more visible. Gap errors were scanned at a scale of 1:12,000 and fixed where possible. Given the number of errors and the scope of this project, best efforts to resolve gap errors were made, but some remain in the final data compilation.

Self-intersecting polygons are another geometry error that was assessed. These are instances where a single polygon intersects itself.

Checks for these geometry errors was first conducted using the ESRI Check Geometry and Repair Geometry tools and the ESRI validation rule. The OGC validation rule was evaluated, but not used in all cases, as it requires that no non-linear features exist. Non-linear features do exist in NWI data, and eliminating these features resulted in drastic changes to the raw data, causing additional overlapping polygons, gaps, and slivers. It was determined that conducting Check Geometry with the OGC validation rule resulted in more errors than it resolved, and under the scope of this project additional clean-up to mitigate these created errors was not possible.

QGIS software was also used to locate self-intersecting polygons by checking validation. The GEOS option was set to on, and the option to Ignore Ring Self Intersections was not checked.

Errors were identified in QGIS, but resolved in ArcGIS Pro. Given the magnitude of the number of these errors best efforts were made to clean up the majority of errors, however some still remain.

Finally, the extent of the feature class was reset using the ArcGIS Pro function, RecalculateFeatureClassExtent

Attribution

Attribution was based on information in the NWI classification scheme. NWI classifications were also cross-walked with CMECS codes, and these coded domains and descriptions were also included in the attribution.

The raw NWI data contains a field titled ATTRIBUTE that defines the NWI classification code. These alphanumeric classification codes correspond to nomenclature, defined by NWI, that best describes a particular wetland habitat. NWI provides a lookup table for each code that defines each portion of the classification code.

Attribution was applied to the coastal wetlands dataset that broke out each classification code and described the system, subsystem, class, subclass, and any modifiers including water regime, special modifiers, water chemistry, and soil for each wetland feature. These definitions were derived directly from NWI. Wetland classifications were broadly reviewed, and in some cases suggested edits were provided by NOAA OCM to fix incorrect classifications.

A crosswalk of NWI classification codes and CMECS codes was provided by NOAA OCM. This crosswalk was used to define the CMECS biotic definition, biotic code, substrate definition, and substrate code for each wetland feature.

Data Compilation

Following geometry review and attribution, projects were merged together based on region. Regions include: the Atlantic coast, the Gulf coast, the Pacific coast, Alaska, and U.S. territories and islands. Data was compiled at these scales instead of the national scale to make viewing and using the data more manageable.

Adjacent NWI state datasets overlap each other. Overlapping state datasets share identical features except at the boundaries of the dataset, where features are cut-off at a hard boundary line. For this reason, adjacent state datasets could not simply be merged together, as merged products would contain duplicate and overlapping non-identical features. This overlapping needed to be handled first:

Using state dataset extents overlapping data between states was clipped out of neighboring datasets. State datasets were then merged together and it was confirmed that data along the created stitch line met perfectly without overlaps or gaps created. Data between the now adjacent, but not overlapping, state datasets were merged together to create a single polygon spanning both state datasets where applicable. In instances where NWI classifications differed between adjacent polygons, the differentiation between the polygons remained.

Data Integration and Tuning

All regions were projected to NAD83

All regions were combined using the ArcGIS Pro function MERGE

The estuarine and marine subtidal features were removed

Field size settings were reduced to approximate field size values

The ArcGIS Pro function REPAIR GEOMETRY with the OGC parameter was executed

The ArcGIS Pro functions RECALCULATE FEATURE CLASS EXTENT was executed