Skip to Page Content
 

Data quality

The following spatial data quality issues should be considered when evaluating how EFH and HAPC data can be effectively used in a geographic information system (GIS). The usage caveats derived from these data issues (included in the data inventory) should be carefully considered when interpreting any analyses based on EFH or HAPC data layers.

General data issues

data quality warning Georeferencing Errors
Errors in the processing of GIS data can sometimes create inaccuracies in the spatial representations of geographic features. While these types of errors can be insignificant when creating large-scale maps, they become more problematic when relying on data to provide accurate information at very specific locations.
Above: Spatial errors visible in a 1:2,000,000 scale polygon representing Treasure Island in the San Francisco Bay.

data quality warning Spatial Resolution
All GIS data is created with accuracy and detail appropriate for use at specific spatial scales. When low-resolution data is used in a high-resolution context it may not adequately describe the complexity of the features it is meant to represent.
Left: A island off the coast of Maine represented by a low-resolution (1:3,000,000 scale) coastline.
Right: The same island represented with a higher resolution (1:100,000 scale) coastline.

data quality warning Out-of-date information
Geographic and environmental features can change significantly over time, particularly in the marine and coastal environment. As ground conditions change, GIS data representing conditions at a specific point in time will becomes increasingly unreliable.
Above: Out-of-date spatial data representing barrier islands off the coast of Louisiana overlaid on more recent imagery.

data quality warning Inadequate Metadata
The metadata describing a GIS dataset is a critical component in evaluating how it can and should be used. If metadata does not exist or if it is incomplete, it can be very difficult to accurately interpret the information that the data provides.

data quality warning Overly Exclusive Data
Attempts to refine the spatial extents of EFH using high-resolution habitat or coastline data can result in a data set that has a high probability providing false negative EFH information (i.e. indicates that EFH is not present when it actually might be). In situations where a conservative (or precautionary) analysis is required, highly refined datasets may be too exclusive to appropriately account for the uncertainty involved in spatially defining marine habitat.

Coastline Artifacts
Marine GIS data layers that extend to the shoreline are generally cropped using one of the many available shoreline polygons. If the GIS data is going to be used to return location-specific information in near-shore or estuarine areas the following potential artifacts inherent in shoreline data should be considered:

data quality warning Spatial Resolution of Shoreline
If the resolution of the shoreline data used to define the boundaries of EFH is too low to conform to the complexity of actual geographic shoreline features, it will not be reliable in near-shore areas.
Left: An aerial image covering an approximately 40 sq mile area of the coast of South Carolina showing Port Royal Sound branching into the Broad, Beaufort, and Chechessee Rivers.
Right: A 1:3,000,000 scale coastline representing the same area.

data quality warning Inland Extents
GIS shorelines generally describe the interface between open water and intertidal land areas. EFH data clipped to a shoreline may limit the inland extent of aquatic species to the open water and falsely exclude what might actually be important habitat areas.
Left: An aerial image covering an approximately 7.5 sq mile area of coastal South Carolina showing a section of the Broad River with the land-water boundary defined by a standard 1:2,000,000 scale coastline
Right: The same area with the with an inland boundary defined by the limits of the estuarine system as mapped by the National Wetlands Inventory

data quality warning Upstream Extents
The upstream ranges of marine and anadromous species depend on many physical and geographic factors specific to the area of interest. Often the polygons that are used to represent species distributions were not created with any consideration of the factors that would determine the actual upstream extents of a species, but rather incorporate the arbitrary delineation between a river and a bay that are inherent in the coastline data used.
Left: A 5 square mile area of Coastal South Carolina showing imagery of the branches of the Tultifinny River with a 1:10,000 scale coastline overlaid in red. Note the arbitrary upstream cutoff points of the river branches.
Right: The same area with the upstream and inland extents defined by the limits of the estuarine system as mapped by the National Wetlands Inventory

data quality warning Land Overlap
In an attempt to maintain a conservatively inclusive boundary for EFH, data may not be clipped to a shoreline at all, but rather allowed to overlap land. Data of this type can be overly inclusive and may not be appropriate for analyses that require some more refined mapping of upstream or inland extents.
Above: EFH polygons for bonnethead shark shown overlapping land.

Data Aggregation Errors
Habitat mapping in the marine environment is rarely done on a large scale. It is therefore common that regional-scale EFH maps rely on the aggregation of numerous small-scale habitat datasets. GIS datasets created in this way should be used cautiously due to the following common issues:

data quality warning Inconsistent Methodologies
Marine habitat datasets from different sources rarely use the same methodologies or classification systems. The process of combining them, therefore, often must reduce the classification scheme to the lowest common denominator of the two datasets. In this process, important distinctions in habitat type, density, or patchiness can be reduced to simple presence absence. This process can also create edge matching artifacts where adjacent polygons from different data sets do not line up correctly due to variations in sampling methods or changes in the ground conditions at the different times.

data quality warning Loss of "no data" information
In the process of assembling an aggregated dataset, the information about the bounding areas of the individual datasets is often lost. When this happens, areas where there was no data collected become intermingled with the portions of the surveyed areas where there was no habitat found. If the absences of habitat and the absence of data are indistinguishable, the entire dataset becomes unreliable and tends to falsely imply absence in areas where, in fact, it contains no information.

Source data
Aggregated data set
Left: An illustration of source data maintaining the distinctions between areas of presence, absence, and no data.
Right: An illustration of how the distinction between absence and not data can be confused in an aggregated dataset.

data quality warning Loss of source metadata
In an aggregated dataset each polygon can potentially be derived from a different source data set. If there is no attribute associated with each polygon to indicate its source, this important information can be lost.