Lineage:
Source_Information:
Source_Citation:
Citation_Information:
Originator:
Publication_Date: unpublished material
Title: Digital Elevation Model (DEM)
Geospatial_Data_Presentation_Form: raster digital data
Type_of_Source_Media: None
Source_Citation_Abbreviation: DEM
Source_Contribution: elevation data
Source_Information:
Source_Citation:
Citation_Information:
Originator:
Publication_Date: unpublished material
Title: Photo-interpreted Canopy Cover (FIA)
Geospatial_Data_Presentation_Form: vector digital data
Type_of_Source_Media: None
Source_Citation_Abbreviation: FIACC
Source_Contribution: canopy cover estimate (training/validation)
Source_Information:
Source_Citation:
Citation_Information:
Originator:
Publication_Date: unpublished material
Title: X-Coordinate
Geospatial_Data_Presentation_Form: raster digital data
Type_of_Source_Media: None
Source_Citation_Abbreviation: XCoord
Source_Contribution: east-west location (relative to 96.0 W longitude)
Source_Information:
Source_Citation:
Citation_Information:
Originator:
Publication_Date: 20110101
Title: NLCD 2006 Land Cover
Geospatial_Data_Presentation_Form: raster digital data
Publication_Information:
Publication_Place:
Publisher: U.S. Geological Survey
Type_of_Source_Media: None
Source_Citation_Abbreviation: NLCD06LC
Source_Contribution: land cover information
Source_Information:
Source_Citation:
Citation_Information:
Originator:
Publication_Date: 20030101
Title: NLCD2001 Percent Tree Canopy Cover
Geospatial_Data_Presentation_Form: raster digital data
Publication_Information:
Publication_Place:
Publisher: U.S. Geological Survey
Type_of_Source_Media: None
Source_Citation_Abbreviation: NLCD01TC
Source_Contribution: percent tree canopy cover
Source_Information:
Source_Citation:
Citation_Information:
Originator:
Publication_Date: 19950101
Title: Bailey's Ecoregion Sections
Geospatial_Data_Presentation_Form: raster digital data
Publication_Information:
Publication_Place:
Publisher: USDA Forest Service Rocky Mountain Research Station
Type_of_Source_Media: None
Source_Citation_Abbreviation: Bailey
Source_Contribution: ecosystem geography
Source_Information:
Source_Citation:
Citation_Information:
Originator:
Publication_Date: unknown
Title: Landsat 5 Multispectral Thematic Mapper Imagery
Geospatial_Data_Presentation_Form: raster digital data
Publication_Information:
Publication_Place:
Publisher: U.S. Geological Survey
Type_of_Source_Media: None
Source_Citation_Abbreviation: L5
Source_Contribution: spectral information
Source_Information:
Source_Citation:
Citation_Information:
Originator:
Publication_Date: unpublished material
Title: Y-Coordinate
Geospatial_Data_Presentation_Form: raster digital data
Type_of_Source_Media: None
Source_Citation_Abbreviation: YCoord
Source_Contribution: north-south location (relative to 23.0 N latitude)
Source_Information:
Source_Citation:
Citation_Information:
Originator:
Publication_Date: unpublished material
Title: Landsat 5 Multispectral Composite Imagery
Geospatial_Data_Presentation_Form: raster digital data
Type_of_Source_Media: None
Source_Citation_Abbreviation: L5Comp
Source_Contribution: spectral information
Process_Step:
Process_Description: Creation of Landsat derivatives. Spectral derivative images were calculated from
the Landsat composite image for each WRS-2 path/row. NDMI (normalized difference moisture
index), NDVI (normalized difference vegetation index), and the 6-band tasseled cap
transformation (Crist and Kauth1986) were calculated following industry standards. Focal
standard deviations (FSD) were also calculated for the composite image (all bands), NDMI,
NDVI, and the tasseled cap derivative (all bands).
Source_Used_Citation_Abbreviation: L5Comp
Process_Date: 20130101
Source_Produced_Citation_Abbreviation: NDMI
Source_Produced_Citation_Abbreviation: NDMIFSD
Source_Produced_Citation_Abbreviation: TasCapFSD
Source_Produced_Citation_Abbreviation: NDVIFSD
Source_Produced_Citation_Abbreviation: L5FSD
Source_Produced_Citation_Abbreviation: TasCap
Source_Produced_Citation_Abbreviation: NDVI
Process_Step:
Process_Description: Creation of NLCD 2001 percent tree canopy cover derivative. Focal standard
deviations (FSD) were calculated for the NLCD 2001 tree canopy data. The dataset was then
subset to individual WRS-2 path/row boundaries for use in subsequent processes.
Source_Used_Citation_Abbreviation: NLCD01TC
Process_Date: 20130101
Source_Produced_Citation_Abbreviation: NLCD01TCFSD
Process_Step:
Process_Description: Creation of DEM derivatives. A CONUS-wide 30-m DEM, ortho-rectified to the 2001
NLCD percent tree canopy cover dataset, was provided by USGS for this project. Slope,
aspect, and the sine and cosine of aspect were calculated for each pixel following
industry standards. Focal standard deviations (FSD) were also calculated for the DEM and
each derivative dataset. Each of these data layers was subset to individual WRS-2 path/row
boundaries for use in subsequent processes.
Source_Used_Citation_Abbreviation: DEM
Process_Date: 20130101
Source_Produced_Citation_Abbreviation: AspSin
Source_Produced_Citation_Abbreviation: AspSinFSD
Source_Produced_Citation_Abbreviation: Slope
Source_Produced_Citation_Abbreviation: AspectFSD
Source_Produced_Citation_Abbreviation: AspCosFSD
Source_Produced_Citation_Abbreviation: DEMFSD
Source_Produced_Citation_Abbreviation: Aspect
Source_Produced_Citation_Abbreviation: SlopeFSD
Source_Produced_Citation_Abbreviation: AspCos
Process_Step:
Process_Description: Creation of Landsat composite. Fifteen Landsat 5 scenes were selected and
processed for each WRS-2 path/row. Selected scenes were acquired between 2007 and 2011,
with the majority evenly distributed between 2009, 2010 and 2011. The selection process
favored scenes with minimal cloud cover and with MODIS-based NDVI values near the annual
peak for the dominant forest cover type. Remaining clouds were removed using the Fmask
tool (Zhu and Woodcock 2012). Six spectral bands (TM bands 1-5 and 7) within each scene
were atmospherically corrected with dark object subtraction and transformed to surface
reflectance (Chander et al. 2009; Chavez 1988). Each set of 15 6-band scenes was then
combined to form a cloud-free composite image for the given path/row.
Source_Used_Citation_Abbreviation: L5
Process_Date: 20130101
Source_Produced_Citation_Abbreviation: L5Comp
Process_Step:
Process_Description: Creation of percent tree canopy cover dataset (main process). For practical
reasons, the NLCD 2011 percent tree canopy cover (TCC 2011) CONUS dataset was created
piecewise in 68 zones, called “mapping areas”. Each mapping area was based on the zones
defined in Homer and Gallant (2001), but extended to include the footprint of all WRS-2
path/rows that intersected the given zone. Each mapping area included between 9 and 27
WRS-2 path/rows. The final dataset is a mosaic of TCC 2011 for all CONUS WRS-2 path/rows.
Five major steps were employed to map tree canopy cover: collection of reference data,
acquisition and/or creation of predictor layers, calibration of Random Forests™ regression
models for each mapping area using reference data and predictor layers, application of
those models to predict per-pixel tree canopy cover across the entire mapping area, and
creation of the CONUS-wide mosaic. The methodology is described further below and in
Coulston et al. (2012). Reference data, consisting of estimated tree canopy cover at each
of 63,008 FIA plot locations, were generated via photographic interpretation of high
spatial resolution images acquired by the National Agricultural Inventory Program (NAIP).
The reference data were collected and supplied by the USDA Forest Service Forest Inventory
and Analysis (FIA) program. The spatial distribution of the sample points follows the FIA
systematic grid (Brand et al. 2000). Predictor layers included Landsat 5 Thematic Mapper
composite imagery and spectral derivatives thereof (NDMI, NDVI, and tasseled cap);
elevation data and spatial derivatives thereof (slope and aspect, as well as sine and
cosine of aspect); NLCD 2001 percent tree canopy cover; NLCD 2006 land cover; Bailey's
eco-regions; and the focal standard deviation of each of the preceding layers except for
land cover and eco-regions. Explicit location data were also included as predictor layers.
The processes for creating the derived layers are described separately (see related
Process Steps). Modeling was carried out using the Random Forests™ (Breiman 2001) data
mining technique as outlined in the Attribute Accuracy Report above. To minimize the root
mean square error, variable importance information obtained from the Random Forests™
models was used to reduce the number of predictor variables. These reduced datasets were
used for the final Random Forests™ modeling. The models developed for each mapping area
were applied to individual WRS-2 path/rows within the mapping area, producing a 2-layered
image. The first layer was the RandomForests™ estimate of tree canopy cover and the second
layer was the standard error, which is the per-pixel square root of the variance of the
Random Forests™ estimates from the individual trees. Since models were applied to each
mapping area independently, there were multiple estimates for pixels in overlapping areas.
For these pixels, the estimate with the lowest standard error was carried into the
CONUS-wide mosaic.
Source_Used_Citation_Abbreviation: FIACC, L5Comp, L5FSD, NDMI, NDMIFSD, NDVI, NDVIFSD, TasCap, TasCapFSD, DEM, DEMFSD,
Aspect, AspectFSD, AspCos , AspCosFSD, AspSin, AspSinFSD, Slope, SlopeFSD, Bailey,
NLCD06LC, NLCD01TC, NLCD01TCFSD, XCoord, YCoord
Process_Date: 20140101
Process_Step:
Process_Description: CUGIR staff clipped data to New York state boundary.
Process_Date: 20180430
