Layer: Barrens (ID: 17)

Parent Layer: Other

Name: Barrens

Display Field: SWAPClass

Type: Raster Layer

Geometry Type: null

Description: <DIV STYLE="text-align:Left;">This data layer serves as a classification of barrens ecosystems across Colorado. It is derived from a random forest land cover classification model and is made for inclusion on the 2025 State Wildlife Action Plan. Benefits* This layer is specific to Colorado, making it more accurate than some nation-wide datasets.The barrens layer is best used as a prompt for conversation. It:* Should be used to identify barren ecosystems at a regional scale* Should be bolstered with locally available data and knowledge* Should not be scrutinized at a fine scale* Should not be used as the single determining factor for future projectsProcessStep 1: Vegetation indices served as valuable predictors of barren ecosystems and were generated in Google Earth Engine using Harmonized Landsat Sentinel (HLSS30) data. NDVI, NDWI, and NDBI indices were exported after creation into ArcGIS Pro, along with RAP BGR and SSURGO-Cor datasets.Step 2: Training data for this model was generated in Google Earth Engine. A series of random points were sampled across two layers of training polygons (barrens and non-barrens), which were drawn by Jack Hagenbuch and based on three data sources. These points were then exported to ArcGIS Pro. Step 3: In ArcGIS Pro, a fishnet of points was generated to represent each cell in a LANDFIRE EVT dataset. These test points fell within the center of each cell. Once created, location data was sampled for each point in both training and testing dataset attribute tables. Following that, raster data from step 1 was sampled for each point in a similar manner. Once sampling had been performed, attribute tables for training and testing data were exported as CSV files. Step 4: CSV files containing point locations (x/y coordinates) and raster values (NDVI, NDBI, NDWI, RAP BGR, SSURGO-Cor) for both training and testing data were imported into R. Training data were used to generate a random forest model and test its accuracy. Once the model had been set up, training data were used to classify the statewide testing data. Outputs were then filtered to only select barrens classifications and transformed from points to 30m raster cells to mirror LANDFIRE. Step 5: A masking layer was generated using LANDFIRE EVT values, which were determined to contain no barrens; this layer was designed to remove false positive classifications. This layer was applied to the 30m raster output from step 4 to finalize the 2025 SWAP Barrens Ecosystem Layer.</DIV>

Copyright Text: CNHP, CPW, USGS, US DOI, USFWS, NWI, LANDFIRE, Esri, US EPA, NASA

Default Visibility: true

MaxRecordCount: 0

Supported Query Formats: JSON, geoJSON, PBF

Min Scale: 0

Max Scale: 0

Supports Advanced Queries: false

Supports Statistics: false

Has Labels: false

Can Modify Layer: false

Can Scale Symbols: false

Use Standardized Queries: true

Supports Datum Transformation: true

Extent:

Drawing Info:

N/A Advanced Query Capabilities:

HasZ: false

HasM: false

Has Attachments: false

HTML Popup Type: esriServerHTMLPopupTypeNone

Type ID Field: null

Fields:

OBJECTID ( type: esriFieldTypeOID, alias: OBJECTID )
Value ( type: esriFieldTypeInteger, alias: Value )
SWAPClass ( type: esriFieldTypeString, alias: SWAPClass, length: 254 )
GroupClass ( type: esriFieldTypeString, alias: GroupClass, length: 254 )
Legend ( type: esriFieldTypeString, alias: Legend, length: 254 )
PixelCount ( type: esriFieldTypeDouble, alias: PixelCount )
Count ( type: esriFieldTypeDouble, alias: Count )

Supported Operations: Query Query Attachments Generate Renderer Return Updates

Iteminfo Thumbnail Metadata