This dataset is the finished product of the NPS Vegetation Mapping Project at Fire Island National Seashore. This dataset depicts the association-level vegetation map for the entire length of Fire Island and the William Floyd Estate. The park islands in the Great South Bay can be found in a separate file included on the disk. These vegetation polygons were interpreted and delineated from 1:1200-scale true-color aerial photographs taken in April 1997. They are attributed with NVCS associations as well as height, pattern, and density information.
The purpose of this data is to provide the managers and researchers on Fire Island with an accurate spatially referenced dataset to assist in their efforts.
ground condition as of 1997
These data were designed to identify and map vegetation polygons only. They are not appropriate for other uses such as legal property boundary identification, tax assessment, etc.
203 West Roanoke Street
The accuracy of classification to association was completed with both a traditional and a fuzzy-set assessment. This assessment only determined accuracy for polygons at or above the MMU. The overall accuracy of the map was determined to be .58 (Kappa). The subsequent Level 5,4, and 3 level fuzzy accuracy assessment produced valuesof .64, .77 and .87 respectively. The Level 4 accuracy assessment value from the fuzzy assessment is provided for each class. Accuracy assessments were also performed for polygons below the MMU. These can be found in the final report.
Producers:88.9% Users: 85.7% This type was very similar to Maritime Deciduous Scrub Forest and was considered the same at Level 4.
Producers:100% Users: 100% The type was easily identified and limited to two large stands on the Floyd Estate.
Producers: 0% Users: 0% Only one stand was identified as this type. That stand was classified as Coastal Oak-Heath Forest during accuracy assessment.
Producers: 100% Users: 88.9% This type was most often confused with Pitch pine - Oak Forest which differs only in amount of Pinus rigida in the canopy.
Producers: 75% Users: 78.6% This type was most often confused with Pitch Pine Dune Woodland. The two types were considered the same at Level 4.
Producers: 100% Users: 85.7% The photointerpreters felt that some polygons on the Floyd Estate were more appropriately assigned to this class rather than the Coastal Oak-Heath Forest. This type was very similar to the Coastal Oak -Heath type and was considered to be the same thing at Level 4. This type may actually be a product of a more oak dominated canopy with dense Smilax spp. beneath.
Prodcuers: 81.3% Users: 83.3% This type was often found in small, linear, polygons, which may account for confusion with non-coniferous associations. Considered the same as Japanese Black Pine Forest for Level 4.
Producers: 57.1% Users: 76.9% There is no clear pattern of confusion beyond other shrubs, although some confusion (with herbaceous types) is due to complex polygon interspersion and small polygons.
Prodcuers: 68.0% Users: 64.7% Nearly all of the confusion with this type occurs with other shrub associations such as Highbush Blueberry Shrub Forest. This type was considered the same as Maritime Holly Forest at Level 4.
Producers: 25.0% Users: 25.0% This type is difficult to identify both from photography and in the field. It is closely associated with Northern Dune Shrubland and is often confused with it. It is also a rare type on Fire Island.
Prodcuers: 50.0% Users: 20.0% This type is frequently confused with the other, more common wetland shrub types Maritime Deciduous Scrub Forest and Northern Salt Shrub.
Prodcuers: 60.0% Users: 57.9% This type is frequently confused with other wetland types such as Highbush Blueberry Shrub Forest. Phragmites is found frequently within these stands as well.
Producers: 81.5% Users: 86.4% This type exists both as an association and with Northern Beach Grass Dune in mosaic. Errors are thought to occur in smaller polygons juxtaposed with Northern Beach Grass Dune and Northern Dune Shrubland.
Producers: 100% Users: 40% No single polygon of this association exists at the 0.25 MMU. These accuracy estimates are from small polygons. This type may be over-predicted on the landscape because it is easily confused with small herbaceous wetlands that are also filled with water at the time of photo acquisition.
Producers: 87.5% Users: 76.9% This is the most prevalent association on Fire Island. It is part of a mosaic with Beach Heather Dune and most observed confusion is likely due to smaller polygons interspersed with that and Northern Dune Shrubland.
Producers: 0% Users: 0% This type was very limited in distribution on Fire Island. Although several polygons were labeled as this type, none were identified as such in the field. This type is easily confused with Beach Heather Dune or Northern Beach Grass Dune which further confounds mapping efforts.
Producers: 66.7% Users: 50.0% This wetland type was often delineated with adjacent Reedgrass Marsh. It is also a rare type on Fire Island existing in polygons at or below the 0.25 ha MMU.
Producers: 100% Users: 20.0% Although this type occurs all over Fire Island, it is found in narrow bands or small polygons often associated with Reedgrass Marsh. Only 1 field assessment point was located within this type.
Producers: 58.8% Users: 64.7% Low accuracy is likely due to small sample size in the accuracy assessment set. This type was almost exclusively confused with other wetland types. Variable coverage density of Phragmites in other types may lead to confusion.
Producers: 97.3% Users: 100% This type was most often confused with High Salt Marsh. Photointerpretation was mostly determined by presence of water in the photographs which is highly variable (tides, season). This type was considered the same as High Salt Marsh for Level 4.
Producers: 81.3% Users: 100% This type is found in close proximity and intermingled with the Low Salt Marsh type. These two were considered the same for the Level 4 assessment.
Producers: 0% Users: 0% This type was only mapped in a single small polygon and was identified elsewhere during accuracy assessment. It is a very rare type and likely exists in very few small patches. Further confounding this type is its similarity to Northern Dune Shrubland.
Producers: 100% Users: 100% This type is easily identified on the Floyd Estate.
Producers: 100% Users: 89.5%This mosaic was considered correct if identified as either Northern Beach Grass Dune or Beach Heather Dune. There is likely much more of this type on Fire Island, but the sub-0.25 ha polygons make its appearance in the map more rare.
All linework was cleaned before building topology. All reulting polygons are attributed with the appropriate information.
The minimum mapping unit for this project is 0.25-ha. There are, however, several polygons in the map that are smaller than this mapping area. We decided to include these small polygons for several reasons. The most important of these is that many of the recognizable vegetation and man-made features on Fire Island are smaller than the 0.25 MMU planned in the contract. Inclusion of smaller mapping units allowed us to capture these features and improve the utility of the vegetation map. It also allowed us to avoid conglomerating obviously homogeneous blocks of vegetation into "mosaics" or "complexes" at the 0.25-ha unit level.
We collected a total 89 map points on Fire Island and the William Floyd Estate. Of these, 21 points were removed because the standard deviation of the differentially corrected location was higher than 10 m. The resulting 68 points were then overlaid onto the vegetation map and georeferenced aerial photography to assess spatial accuracy of the map. Each point was examined, and the difference (in meters) between the GPS location and map location was measured using the masuring tool in Arc View. This value was entered into the appropriate field in the database. If the exact location of the map point could not be determined or discerned from the aerial photograph, the point was eliminated from consideration. A total of 21 points were removed because they could not be accurately measured. This left 47 points for assessing the spatial accuracy of the vegetation map. The mean error distance was found to be 4.42 m (± 4.94 m). Errors distances ranged from 0.00 m - 30.0 m. When the single 30 m error point was removed (assumed to be an outlier), the mean error distance was 3.86 m ((± 3.18 m) with a range of 0.00 m - 14.09 m.
This dataset was used for photointerpretation (with steroscopes).
These are the scanned and georeferenced versions of the true color photography. These were used to delineate polygon boundaries with on-screen digitizing.
This vegetation map was created from aerial photographs. The photos were scanned at 600 dpi and georeferenced using control points gathered from georeferenced digital orthophoto quads from the USGS. ONce the digital photos were georeferenced, they were used to delineate map polygons interpreted from the stero-paired aerial photographs. Once the delineation of polygons was complete, each polygon was attributed with it's appropriate map class (or vegetation association), height class, pattern class, and density value. The final layer was converted to an ARCINFO coverage and topology was built.
Internal feature number.
Area of feature in internal units squared.
Perimeter of feature in internal units.
numeric code corresponding to map class
The distributor maintains no liability for the use and application of these data beyond it's intended use as a depiction of vegetation.
203 West Roanoke Street