Geospatial data for the Vegetation Mapping Inventory Project of Lewis and Clark National Historical Park
The files linked to this reference are the geospatial data created as part of the completion of the baseline vegetation inventory project for the NPS park unit. Current format is ArcGIS file geodatabase but older formats may exist as shapefiles.
Several imagery sources were used to derive the vegetation map, primarily Light Imaging and Detection Radar (LiDAR) and aerial imagery from the National Agriculture Imagery Program. The eCognition software package and Berkley Imaging Segmentation was used to create the initial image segments and polygon map. The mean values of variable inputs were summarized for each of the training data polygons and were used to generate a predictive non-parametric model using RandomForest in the statistical program R. The model was then applied to all polygons. The resulting draft map was reviewed by experts familiar with the vegetation types of the area.
The final map includes 24 classes, representing 3 land cover types and 21 alliance-based map classes. A field-based, blind random sample accuracy assessment of the map was carried out in mid-August 2011 and December 2011. A total of 175 accuracy assessment points were collected. Based on the assessment, the total map accuracy was 88.9%, exceeding the program standard of 80%.
Complete Metadata
| accessLevel | public |
|---|---|
| bureauCode |
[
"010:24"
]
|
| contactPoint |
{
"fn": "NPS IRMA Help",
"@type": "vcard:Contact",
"hasEmail": "mailto:NRSS_DataStore@nps.gov"
}
|
| description | The files linked to this reference are the geospatial data created as part of the completion of the baseline vegetation inventory project for the NPS park unit. Current format is ArcGIS file geodatabase but older formats may exist as shapefiles. Several imagery sources were used to derive the vegetation map, primarily Light Imaging and Detection Radar (LiDAR) and aerial imagery from the National Agriculture Imagery Program. The eCognition software package and Berkley Imaging Segmentation was used to create the initial image segments and polygon map. The mean values of variable inputs were summarized for each of the training data polygons and were used to generate a predictive non-parametric model using RandomForest in the statistical program R. The model was then applied to all polygons. The resulting draft map was reviewed by experts familiar with the vegetation types of the area. The final map includes 24 classes, representing 3 land cover types and 21 alliance-based map classes. A field-based, blind random sample accuracy assessment of the map was carried out in mid-August 2011 and December 2011. A total of 175 accuracy assessment points were collected. Based on the assessment, the total map accuracy was 88.9%, exceeding the program standard of 80%. |
| distribution |
[
{
"@type": "dcat:Distribution",
"title": "lewigeodata.zip",
"format": "ZIP",
"mediaType": "application/zip",
"description": "ArcGIS geodatabase with metadata, .mxd and layer file",
"downloadURL": "https://irma.nps.gov/DataStore/DownloadFile/723798?Reference=2311155"
}
]
|
| identifier | http://datainventory.doi.gov/id/dataset/NPS_DataStore_2311155 |
| issued | 2025-05-06T00:00:00Z |
| keyword |
[
"Accuracy Assessment",
"Classification",
"Geodatabase",
"Habitat Types",
"Imagery",
"NVC",
"NVCS",
"National Vegetation Classification",
"Plots",
"Releve",
"Relevé",
"Remote Sensing",
"USNVCS",
"Vegetation",
"Vegetation Inventory",
"Vegetation Map",
"Vegetation Mapping",
"botanical survey"
]
|
| landingPage | https://irma.nps.gov/DataStore/Reference/Profile/2311155 |
| modified | 2025-05-06T00:00:00Z |
| programCode |
[
"010:118",
"010:119"
]
|
| publisher |
{
"name": "National Park Service",
"@type": "org:Organization"
}
|
| spatial | -124.761482,45.0340538,-119.5767,49.002388 |
| theme |
[
"Geospatial"
]
|
| title | Geospatial data for the Vegetation Mapping Inventory Project of Lewis and Clark National Historical Park |
