Vermont_Mount_Mansfield_Alpine_Tundra_2013

Metadata also available as

Metadata:

Identification_Information:
Citation:
Citation_Information:
Originator:
Cale Kochenour, The Pennsylvania State University, Graduate Student
Publication_Date: 20181222
Title: Vermont_Mount_Mansfield_Alpine_Tundra_2013
Geospatial_Data_Presentation_Form: vector digital data
Description:
Abstract:
Mapping of the alpine tundra features on Mount Mansfield, Vermont. For this shapefile, alpine tundra is defined by three (3) features: Bare Rock, Alpine Vegetation (non-evergreen), and Subalpine Krummholz (evergreen). The coverage of the alpine tundra spans the Sunset Ridge Trail, the Maple Ridge Trail, and the portion of the Long Trail from the Mount Mansfield Forehead to the Mount Mansfield Adam's Apple. The shapefile was created through an object-based image analysis. Data sources include high-resolution 4-band imagery (Red, Green, Blue, Infrared) from the Vermont Imagery Program (0.2 meters, 0.5 meters) collected in 2013 during leaf-off conditions and the National Agriculture Imagery Program (1 meter) collected in 2014 during leaf-on conditions, and high-resolution elevation data (normalized digital surface model, digital elevation model) from the Vermont Lidar Program (0.7 meters) collected in 2014 during leaf-off conditions.
Purpose:
This shapefile was created to establish a baseline land cover map for the alpine tundra on Mount Mansfield, Vermont. The delivery of the shapefile to the Vermont Center for Geographic Information is to make the data available for download, viewing, and use in future GIS, remote sensing, and interdisciplinary studies.
Status:
Progress: Complete
Maintenance_and_Update_Frequency: None planned
Spatial_Domain:
Bounding_Coordinates:
West_Bounding_Coordinate: -72.840115
East_Bounding_Coordinate: -72.808369
North_Bounding_Coordinate: 44.548701
South_Bounding_Coordinate: 44.515278
Keywords:
Theme:
Theme_Keyword_Thesaurus: None
Theme_Keyword: Remote Sensing
Theme_Keyword: Object-Based Image Analysis
Theme_Keyword: Feature Extraction
Theme_Keyword: Land Cover
Theme_Keyword: Alpine Tundra
Theme:
Theme_Keyword_Thesaurus: ISO 19115 Topic Categories
Theme_Keyword: environment
Theme_Keyword: imageryBaseMapsEarthCover
Theme_Keyword: biota
Place:
Place_Keyword_Thesaurus: Geographic Names Information System
Place_Keyword: Mount Mansfield
Place_Keyword: Chittenden County
Place_Keyword: Lamoille County
Place_Keyword: Green Mountains
Place_Keyword: Vermont
Place_Keyword: United States
Access_Constraints:
No access or use constraints. The originator makes no representations of any kind, including but not limited to the warranties of merchantability or fitness for a particular use, nor are any such warranties to be implied with respect to the data. VCGI and the State of VT make no representations of any kind, including but not limited to the warranties of merchantability or fitness for a particular use, nor are any such warranties to be implied with respect to the data.
Use_Constraints:
No use limitations. Users of this shapefile should be aware that changes in the land cover may have occurred since the time period when the imagery and elevation data used for feature extraction was collected. Some parts of the shapefile may no longer represent actual surface conditions. Users of this shapefile should be aware of this limitation prior to using it for additional analyses or studies.
Point_of_Contact:
Contact_Information:
Contact_Organization_Primary:
Contact_Organization: The Pennsylvania State University
Contact_Person: Cale Kochenour
Contact_Position: Graduate Student
Contact_Address:
Address_Type: mailing and physical
Address: 317 Katahdin Dr
City: Lexington
State_or_Province: MA
Postal_Code: 02421
Country: US
Contact_Voice_Telephone: 802-282-3526
Contact_Electronic_Mail_Address: calekochenour@gmail.com
Hours_of_Service: 8:00 AM - 8:00 PM, Eastern Time, Monday - Friday
Contact_Instructions:
Email and phone number provided are permanent contact information.
Data_Set_Credit:
The shapefile and metadata were produced by Cale Kochenour, a graduate student enrolled in the Master of Geographic Information Systems (MGIS) program at The Pennsylvania State University, for his Master's degree capstone project.
Security_Information:
Security_Classification_System: N/A
Security_Classification: Unclassified
Security_Handling_Description: N/A
Native_Data_Set_Environment: Version 6.2 (Build 9200) ; Esri ArcGIS 10.6.1.9273
Data_Quality_Information:
Lineage:
Source_Information:
Source_Citation:
Citation_Information:
Publication_Date: 20131209
Title: VTORTHO_0_2M_CLRIR_2013
Edition: 2013A
Geospatial_Data_Presentation_Form: remote-sensing image
Online_Linkage:
https://maps.vcgi.vermont.gov/gisdata/vcgi/imagery/VTORTHO/0_2M/CLRIR/2013/COMP/
Source_Scale_Denominator: 1600
Type_of_Source_Media: onLine
Source_Citation_Abbreviation: 2013 - Color & Infrared - Eastern Chittenden Co. (20 cm)
Source_Contribution:
From VTORTHO_0_2M_CLRIR_2013 metadata: “Natural color and color infrared digital ground orthoimagery covering a portion of Chittenden County, Vermont. The digital orthophotography and color infrared digital orthos are composed of 24 bit natural color digital orthos at a scale of 1 to 1,600 with a 20 centimeter pixel resolution.” Metadata link: https://maps.vcgi.vermont.gov/gisdata/metadata/VTORTHO_0_2M_CLRIR_2013.htm. The acquisition date for the imagery used in this analysis is 20130501.
Source_Information:
Source_Citation:
Citation_Information:
Publication_Date: 20131204
Title: VTORTHO_0_5M_CLRIR_2013
Edition: 2013A
Geospatial_Data_Presentation_Form: remote-sensing image
Online_Linkage:
https://maps.vcgi.vermont.gov/gisdata/vcgi/imagery/VTORTHO/0_5M/CLRIR/2013/COMP/
Source_Scale_Denominator: 4000
Type_of_Source_Media: onLine
Source_Citation_Abbreviation: 2013 - Color & Infrared - Northwestern Vermont (0.5m)
Source_Contribution:
From VTORTHO_0_5M_CLRIR_2013 metadata: “Natural color and color infrared digital ground orthoimagery covering a portion of Vermont State. The digital orthophotography and color infrared digital orthos are composed of 24 bit natural color digital orthos at a scale of 1 to 4,000 with a 0.5 meter pixel resolution.” Metadata link: https://maps.vcgi.vermont.gov/gisdata/metadata/VTORTHO_0_5M_CLRIR_2013.htm. The acquisition date for the imagery used in this analysis is 20130421.
Source_Information:
Source_Citation:
Citation_Information:
Publication_Date: 20150306
Title: NAIP_1M_CLRIR_2014
Edition: 2014A
Geospatial_Data_Presentation_Form: remote-sensing image
Online_Linkage:
https://maps.vcgi.vermont.gov/gisdata/vcgi/imagery/NAIP/1M/CLRIR/2014/COMP/
Source_Scale_Denominator: 12000
Type_of_Source_Media: onLine
Source_Citation_Abbreviation: 2014 - Color & Infrared (4 band) - Statewide NAIP (1m)
Source_Contribution:
From NAIP_1M_CLRIR_2014 metadata: “The NAIP_1M_CLRIR_2014 dataset is a (1 meter) truecolor and infrared (4 band) NAIP imagery product acquired during the summer of 2014 by the USDA-FSA-APFO NAIP program, then reprojected to VT State Plane Meters and cropped to the USGS quarter quad boundary by VCGI.” Metadata link: https://maps.vcgi.vermont.gov/gisdata/metadata/NAIP_1M_CLRIR_2014.htm. The acquisition date for the imagery used in this analysis is 20140811.
Source_Information:
Source_Citation:
Citation_Information:
Publication_Date: 20180209
Title: ElevationOther_nDSM0p7M2014
Edition: 2018A
Geospatial_Data_Presentation_Form: raster digital data
Series_Information:
Series_Name:
Normalized Digital Surface Model (nDSM) generated from DEM & DSM data in the VCGI Lidar Program archive of the same resolution, i.e., resolution class 'RESCLASS'
Issue_Identification: 2018A
Online_Linkage:
https://maps.vcgi.vermont.gov/gisdata/vcgi/lidar/0_7M/2014/nDSM/
Type_of_Source_Media: onLine
Source_Citation_Abbreviation:
VT Data - Lidar nDSM (0.7m) 2014, Chittenden Co., Lamoille Co., Orleans Co., Washington Co.
Source_Contribution:
From ElevationOther_nDSM0p7M2014 metadata: “Eastern VT 2014 0.7m and related 'normalized' Digital Surface Model (nDSM). Created nDSM using the ArcGIS 'MINUS' command where each pixel represents the height above ground (not above sea level) and is calculated by subtracting the DEM from the DSM, with vertical units in meters.” Metadata link: https://maps.vcgi.vermont.gov/gisdata/metadata/ElevationOther_nDSM0p7M2014.htm.
Source_Information:
Source_Citation:
Citation_Information:
Publication_Date: 20170720
Title: ElevationDEM_DEMHE0p7M2014
Edition: 2017
Geospatial_Data_Presentation_Form: raster digital data
Series_Information:
Series_Name:
Quality Level 2 Lidar Hydro Enforced Digital Elevation Model (DEMHE) data from the 3D Elevation Program (3DEP)
Issue_Identification: 2017A
Online_Linkage:
https://maps.vcgi.vermont.gov/gisdata/vcgi/lidar/0_7M/2014/DEMHE/
Type_of_Source_Media: onLine
Source_Citation_Abbreviation:
VT Data - Lidar Hydro-Enforced DEM (0.7m) 2014, Chittenden Co., Lamoille Co., Orleans Co., Washington Co.
Source_Contribution:
From ElevationDEM_DEMHE0p7M2014 metadata: “Eastern VT 2014 0.7m and Hydro Enforced Digital Elevation Model (DEMHE) dataset.” Metadata link: https://maps.vcgi.vermont.gov/gisdata/metadata/ElevationDEM_DEMHE0p7M2014.htm.
Process_Step:
Process_Description:
Step 1: Data Acquisition. The analysis obtained imagery and elevation dataset tiles that covered the extent of Mount Mansfield (Sunset Ridge Trail, Maple Ridge Trail, Long Trail from the Mount Mansfield Forehead to the Mount Mansfield Adam’s Apple) from the Vermont Open Geodata Portal. Tiles sizes varied by data source. Note that the VTORTHO_0_2M_CLRIR_2013 dataset (primary imagery data source) did not cover the entire study area. This data set captured imagery covering Chittenden County, but not Lamoille County, which contains the Mount Mansfield Forehead. The analysis used this dataset in lieu of the missing coverage because it contained the highest resolution (0.2 meters) and the acquisition date (early May 2013) occurred in leaf-off conditions. Both characteristics proved ideal for alpine tundra feature extraction. The missing coverage over the Mount Mansfield Forehead created the need to use a second leaf-off data set, and VTORTHO_0_5M_CLRIR_2013 filled that role. In addition, the analysis used leaf-on imagery from the National Agriculture Imagery Program. The analysis used the 2014 NAIP imagery (1 meter) instead of the 2016 NAIP imagery (0.6 meters) due to spatial alignment issues with the 2016 imagery over the changing elevation and terrain on Mount Mansfield. A visual comparison between the NAIP imagery showed that the 2014 NAIP dataset better aligned with the Vermont Imagery Program datasets. Both the normalized digital surface model and digital elevation model datasets covered the full study area.
Process_Step:
Process_Description:
Step 2: Data Preprocessing. The analysis loaded the individual raster tiles for all five datasets into ArcGIS Desktop and used the Mosaic to New Raster tool to create five mosaic raster files, one for each data set. The analysis used the Project Raster tool with a cubic convolution resampling method to project the imagery datasets into NAD 1983(2011) State Plane Vermont FIPS 4400 (EPSG: 6589) to match the native coordinate system of the elevation datasets. Note that the delivered shapefile contains NAD 1983 State Plane Vermont FIPS 4400 (EPSG: 32145) as the spatial reference. The analysis projected the output shapefile to NAD 1983 State Plane Vermont FIPS 4400 (EPSG: 32145) for delivery, but conducted all data processing in NAD 1983(2011) State Plane Vermont FIPS 4400 (EPSG: 6589) for reasons discussed below. The analysis intended to conduct all data processing within spatial reference NAD 1983 State Plane Vermont FIPS 4400 (EPSG: 32145), but errors occurred within eCognition Developer. When the analysis attempted to load all input datasets with a spatial reference NAD 1983 State Plane Vermont FIPS 4400 (EPSG: 32145) into eCognition Developer, the program crashed, and the data processing stalled before it began. Troubleshooting identified this error occurred when eCognition Developer attempted to interpret the text string that defined the spatial reference for each dataset. eCognition Developer crashed when the analysis attempted to load datasets with different spatial references into the same eCognition project. Even though all datasets contained the same coordinate reference system from a human-readable point of view (and the same text string in ArcGIS), the act of projecting the elevation data sets (publish dates of 2017-2018) to match the imagery data sets (publish dates of 2013-2014) altered the text string that eCognition Developers used to interpret the spatial reference in such a way that it did not match the data that natively contained the NAD 1983 State Plane Vermont FIPS 4400 (EPSG: 32145) spatial reference. The difference (only a few characters) proved enough to cause errors in eCognition Developer. The error occurred unidirectionally, when the analysis projected the newer (by publish date) datasets (elevation) to match the older (by publish date) datasets (imagery). The error in eCognition did not occur when the analysis projected the older (by publish date) datasets (imagery) to match the newer (by publish date) datasets (elevation). For this reason, the analysis projected all datasets into NAD 1983(2011) State Plane Vermont FIPS 4400 (EPSG: 6589) for the purposes of data processing, and then projected the output shapefile to NAD 1983 State Plane Vermont FIPS 4400 (EPSG: 32145) for the accuracy assessment and delivery. The analysis used the Slope function in ArcGIS Pro to create a slope layer from the digital elevation model. The analysis used the ArcGIS Clip tool to clip each dataset (to include the slope layer), bounded by the following extent in NAD 1983(2011) State Plane Vermont FIPS 4400 (EPSG: 6589): Top – 227,654.937674 meters, Bottom – 223,950.937674 meters, Left – 472,974.254334 meters, Right – 475,482.854334 meters. The analysis loaded the clipped data from all six datasets (including slope) into eCognition Developer for processing.
Process_Step:
Process_Description:
Step 3: Data Processing. All data processing occurred in eCognition Developer. The analysis loaded all clipped imagery and elevation layers generated in Step 2 (with spatial reference NAD 1983(2011) State Plane Vermont FIPS 4400 (EPSG: 6589)) into eCognition Developer. The eCognition Developer rule set extracted the land cover features of interest (Bare Rock, Alpine Vegetation, Subalpine Krummholz) and classified all other features (Building, Radio/TV Tower, Auto Road/Parking Lot, Gravel Construction Road, Coniferous Tree, Deciduous Tree, Ski Trail, Car) into a single category labeled Other. Steps 3A-3F define the methods the eCognition Developer rule set used to extract features.
Process_Step:
Process_Description:
Step 3A: Separation of Input Data into Parts. The rule set separated the input data into two (2) different areas that required separate segmentation and classification: 1) Area 1 (full coverage of Mount Mansfield, excluding the Forehead) and 2) Area 2 (only the Mount Mansfield Forehead). The need to separate the study area into parts and process them separately occurred because the VTORTHO_0_2M_CLRIR_2013 dataset only covers Chittenden County, but not Lamoille County (which contains the Mount Mansfield Forehead). The VTORTHO_0_5M_CLRIR_2013 dataset covers the full study area, to include the Mount Mansfield Forehead. The Vermont Imagery Program datasets have different acquisition dates and lighting conditions, which required the use of different attribute threshold values and ranges in the rule set to extract the features of interest. The rule set segmented the entire study area based on the VTORTHO_0_2M_CLRIR_2013 imagery and used the brightness attribute to identify and distinguish the Part 1 (brightness >= 2, 8-bit image) and Part 2 (brightness < 2, 8-bit image) areas. The rule set assigned the two areas to different classes in order to proceed with separate segmentation and classification.
Process_Step:
Process_Description:
Step 3B: Identification of Alpine Tundra Candidates in Area 1. The rule set used the normalized digital surface model (ElevationOther_nDSM0p7M2014) to identify alpine tundra candidates within Area 1. The rule set segmented the ElevationOther_nDSM0p7M2014 data and classified image objects with heights (normalized digital surface model values) less than or equal to 2 meters as alpine tundra candidates. The analysis used the demarcation value of 2 meters based on assumptions and characteristics of the features of interest: 1) Bare Rock features exist as bare ground and have heights close to or at 0 meters. 2) Alpine Vegetation (non-evergreen) features must grow low to the ground in order to survive the harsh climate on the ridges of Mount Mansfield and have heights close to or at 0 meters. 3) Subalpine Krummholz (stunted evergreen vegetation) must grow low to the ground relative to non-stunted evergreen vegetation on Mount Mansfield and have heights “from a few inches to a few feet” (Thompson et al., 2005). The analysis interpreted “a few feet” as 3.281 feet (1 meter). 4) The rapid variation of the Mount Mansfield terrain combined with the (assumed) interpolation in the digital surface model and digital elevation model used to create the normalized digital surface model caused some features of interest to register with heights greater than 1 meter. Example occurrences of this include the Mount Mansfield Chin and other areas with steep and rapid elevation change. 5) The transition from coniferous (but not subalpine krummholz) trees to Subalpine Krummholz is not definitive from aerial imagery. 6) The analysis decided to err on the side of inclusion (commission errors) instead of exclusion (omission errors) for the Subalpine Krummholz feature heights. Due to the ambiguity in the term “few”, the nature of the ElevationOther_nDSM0p7M2014 values over the varied and complex terrain, and the decision to err on the side of inclusion, the rule set chose 2 meters as the demarcation line for the height of alpine tundra candidates. The 2-meter height demarcation line created a buffer that followed the three bounding ridgelines (Sunset Ridge, Maple Ridge, Long Trail). This buffer left some gaps, and based on the context (adjacency to alpine tundra candidates), the rule set classified the gaps as alpine tundra candidates. The rule set also rectified “islands” – objects that existed far away in distance from the bounding ridgelines and that the rule set had incorrectly classified as alpine tundra candidates – and classified these objects as non-tundra. The rule set merged all alpine tundra candidate objects. This left a semi-continuous (more than 1 image object, but no gaps within image objects) area of alpine tundra candidates ready for segmentation and classification.
Process_Step:
Process_Description:
Step 3C: Segmentation and Object Classification for Alpine Tundra Candidates in Area 1. The rule set segmented the alpine tundra candidates in Area 1 with a multiresolution segmentation followed by a spectral difference segmentation. Both segmentation methods created image objects based on the VTORTHO_0_2M_CLRIR_2013 imagery. The initial segmentation used a scale parameter of 50, and included the following spectral band layers weights: Red (1), Green (1), Blue (1), Infrared (2). The rule set assigned image objects into seven initial classes: Snow, Shadow, Bare Rock, Alpine Vegetation, Subalpine Krummholz, Mixed Class – Bare Rock/Alpine Vegetation, and Mixed Class – Alpine Vegetation/Subalpine Krummholz. The rule set identified Snow objects with brightness values >= 220 (8-bit image). The rule set identified Shadow with infrared band values <= 50 (8-bit image). The rule set used normalized difference vegetation index (NDVI) ranges to assign classes to the remaining unclassified alpine tundra candidate image objects. The rule set assigned image objects with NDVI values <= -0.125 to Bare Rock. The rule set assigned image objects with -0.05 < NDVI < 0.05 to Alpine Vegetation. The VTORTHO_0_2M_CLRIR_2013 was acquired during leaf-off conditions (May 1, 2013), where the NDVI values remain lower than where they would register in leaf-on conditions. The leaf-off acquisition date provides the opportunity to distinguish the Alpine Vegetation (non-evergreen) from the Subalpine Krummholz (evergreen) features. The rule set assigned image objects with NDVI >= 0.2 to Subalpine Krummholz. The rule set assigned image objects with -0.125 < NDVI < -0.05 to Mixed Class – Bare Rock/Alpine Vegetation. The rule set assigned image objects with 0.05 < NDVI < 0.2 to Mixed Class – Alpine Vegetation/Subalpine Krummholz. The need for the mixed classes within the initial class assignment arose because the analysis could not definitively label the objects in these classes at the scale parameter of 50. The rule set merged each individual class and re-segmented the classes with a smaller scale parameter (10). The rule set classified the re-segmented image objects based on different source layers to rectify commission errors from the initial classification, properly assign the mixed class objects, and classify the land cover underneath the snow and shadow. The rule set merged and re-segmented the Bare Rock objects with a scale parameter of 10 and used NDVI ranges to assign any misclassified Bare Rock features to the appropriate class (i.e. the rule set assigned any Alpine Vegetation, Subalpine Krummholz, Mixed Class – Bare Rock/Alpine Vegetation, or Mixed Class – Alpine Vegetation/Subalpine Krummholz objects misclassified as Bare Rock to the appropriate class). The rule set rectified commission errors in the Alpine Vegetation and Subalpine Krummholz classes through the same method at a scale parameter of 10. The rule set re-segmented the Mixed Class – Bare Rock/Alpine Vegetation objects with a scale parameter of 10 and assigned classes based on NDVI ranges and object brightness (>= 150 for Bare Rock, < 150 for Alpine Vegetation, 8-bit image). The rule set re-segmented the Mixed Class – Alpine Vegetation/Subalpine Krummholz class with a scale parameter of 10 and assigned classes based on NDVI ranges and object context (e.g. relative border to). For context, the rule set assigned the remaining objects to the class to which the image object shared the majority (> 50%) of its border with. The rule set rectified shadow objects by merging and re-segmenting based on the other imagery layers, VTORTHO_0_5M_CLRIR_2013 and NAIP_1M_CLRIR_2014. The Vermont Imagery Program datasets (VTORTHO_0_2M_CLRIR_2013, VTORTHO_0_5M_CLRIR_2013) provided imagery with different and almost opposite sun azimuth angles. The VTORTHO_0_2M_CLRIR_2013 imagery has an approximate (estimated visually, not defined in the metadata) sun azimuth angle of 90 degrees (sun in the East). The VTORTHO_0_5M_CLRIR_2013 imagery has an approximate (estimated visually, not defined in the metadata) sun azimuth angle of 270 degrees (sun in the West). The rule set merged the shadow objects and re-segmented the objects based on the VTORTHO_0_5M_CLRIR_2013 dataset. The rule set assigned shadow objects with VTORTHO_0_5M_CLRIR_2013 infrared band values < 60 (8-bit image) to the overlapping shadow class, defined as shadows in both Vermont Imagery Program data sources that required further rectification. The rule set assigned shadow objects with brightness values > 220 (8-bit image) to snow. The rule set assigned shadow objects with NDVI < 0 to Subalpine Krummholz. The rule set assigned shadow objects with NDVI values >= 0 to Bare Rock. The rule set merged overlapping shadow objects and re-segmented the objects based on the NAIP_1M_CLRIR_2014 dataset with a scale parameter of 25. The NAIP_1M_CLRIR_2014 imagery has an approximate (estimated visually, not defined in the metadata) sun azimuth angle of 90 degrees (sun in the East). The rule set assigned overlapping shadow objects with NDVI values < 0 to Subalpine Krummholz. The rule set assigned overlapping shadow objects with NDVI values >= 0 to Bare Rock. The rule set merged snow objects and re-segmented the snow objects based on NAIP_1M_CLRIR_2014 dataset (leaf-on conditions, no snow present). The rule set assigned the snow objects with NDVI < 0 and brightness > 150 (8-bit image) to Bare Rock. The rule set used object context (relative border to) to classify the remaining snow objects. The rule set assigned the snow objects to the class which the image object shared the majority (> 50%) of its border with, either Bare Rock or Subalpine Krummholz. Similar to the overlapping shadows, the rule set assigned objects to either Bare Rock or Subalpine Krummholz (and not Alpine Vegetation) because it was difficult to use tone to delineate between Alpine Vegetation and Subalpine Krummholz in leaf-on conditions. The rule set removed ski trail, gravel road, auto road, and parking lot features that had been incorrectly assigned to one of the alpine tundra classes. The rule set identified misclassified features with lower slope values and larger relative border to the outer edge of the alpine tundra. The rule set assigned the misclassified features to the Other class. The rule set assigned all non-tundra image objects in Area 1 to the Other class. The rule set separately merged the image objects in each class (Bare Rock, Alpine Tundra, Subalpine Krummholz, Other).
Process_Step:
Process_Description:
Step 3D: Identification of Alpine Tundra Candidates in Area 2. The rule set merged the objects in Area 2 (the part of the study area containing the Mount Mansfield Forehead) and re-segmented the area based on the VTORTHO_0_5M_CLRIR_2013 imagery. The rule set applied the same method and rationale defined in Step 3B, using the normalized digital surface model (ElevationOther_nDSM0p7M2014), to identify alpine tundra candidates in Area 2. The rule set segmented the ElevationOther_nDSM0p7M2014 data and classified image objects with heights (normalized digital surface model values) less than or equal to 2 meters as alpine tundra candidates. The analysis used the demarcation value of 2 meters based on assumptions and characteristics of the features of interest: 1) Bare Rock features exist as bare ground and have heights close to or at 0 meters. 2) Alpine Vegetation (non-evergreen) features must grow low to the ground in order to survive the harsh climate on the ridges of Mount Mansfield and have heights close to or at 0 meters. 3) Subalpine Krummholz (stunted evergreen vegetation) must grow low to the ground relative to non-stunted evergreen vegetation on Mount Mansfield and have heights “from a few inches to a few feet” (Thompson et al., 2005). The analysis interpreted “a few feet” as 3.281 feet (1 meter). 4) The rapid variation of the Mount Mansfield terrain combined with the (assumed) interpolation in the digital surface model and digital elevation model used to create the normalized digital surface model caused some features of interest to register with heights greater than 1 meter. 5) The transition from coniferous (but not subalpine krummholz) trees to Subalpine Krummholz is not definitive from aerial imagery. 6) The analysis decided to err on the side of inclusion (commission errors) instead of exclusion (omission errors) for the Subalpine Krummholz feature heights. Due to the ambiguity in the term “few”, the nature of the ElevationOther_nDSM0p7M2014 values over the varied and complex terrain, and the decision to err on the side of inclusion, the rule set chose 2 meters as the demarcation line for the height of alpine tundra candidates. The 2-meter height demarcation line created a buffer that surrounded the Mount Mansfield Forehead and part of the Maple Ridge Trail. This buffer left some gaps, and based on the context (adjacency to alpine tundra candidates), the rule set classified the gaps as alpine tundra candidates. The rule set also rectified “islands” – objects that existed far away in distance from the bounding ridgelines and that the rule set had incorrectly classified as alpine tundra candidates – and classified these objects as non-tundra. The rule set merged all alpine tundra candidate objects. This left a semi-continuous (more than 1 image object, but no gaps within image objects) area of alpine tundra candidates ready for segmentation and classification.
Process_Step:
Process_Description:
Step 3E: Segmentation and Object Classification for Alpine Tundra Candidates in Area 2. The rule set segmented the alpine tundra candidates in Area 2 with a multiresolution segmentation followed by a spectral difference segmentation. Both segmentation methods created image objects based on the VTORTHO_0_5M_CLRIR_2013 imagery. The initial segmentation used a scale parameter of 50, and included the following spectral band layers weights: Red (1), Green (1), Blue (1), Infrared (2). The rule set assigned image objects into seven initial classes: Snow, Shadow, Bare Rock, Alpine Vegetation, Subalpine Krummholz, Mixed Class – Bare Rock/Alpine Vegetation, and Mixed Class – Alpine Vegetation/Subalpine Krummholz. The rule set identified snow objects with brightness values >= 220 (8-bit image) and NDVI values < -0.25. The rule set identified shadows with low infrared band values (< 60, 8-bit image). The rule set used NDVI ranges to assign classes to the remaining unclassified alpine tundra candidate image objects. The rule set assigned image objects with NDVI values <= -0.2 to Bare Rock. The rule set assigned image objects with -0.125 < NDVI < 0.125 to Alpine Vegetation. The VTORTHO_0_5M_CLRIR_2013 was acquired during leaf-off conditions (April 21, 2013), where the NDVI values remain lower than where they would register in leaf-on conditions. The leaf-off acquisition date provides the opportunity to distinguish the Alpine Vegetation (non-evergreen) from the Subalpine Krummholz (evergreen) features. The rule set assigned image objects with NDVI >= 0.2 to Subalpine Krummholz. The rule set assigned image objects with -0.2 < NDVI < -0.125 to Mixed Class – Bare Rock/Alpine Vegetation. The rule set assigned image objects with 0.125 < NDVI < 0.2 to Mixed Class – Alpine Vegetation/Subalpine Krummholz. The need for the mixed classes within the initial class assignment arose because the analysis could not definitively label the objects in these classes at the scale parameter of 50. The rule set merged each individual class and re-segmented the classes with a smaller scale parameter (25). The rule set classified the re-segmented image objects based on the NAIP_1M_CLRIR_2014 imagery to rectify commission errors from the initial classification, properly assign the mixed class objects, and classify the land cover underneath the snow and shadow. The rule set merged and re-segmented the Alpine Vegetation class with a scale parameter of 25. The rule set used NDVI values < -0.2 to assign misclassified Alpine Vegetation objects to Bare Rock and used brightness values > 200 (8-bit image) to assign misclassified Alpine Vegetation to Snow. The rule set merged and re-segmented the Subalpine Krummholz class with scale parameter of 25 and used NDVI values (-0.125 < NDVI < 0.125) to assign any misclassified Subalpine Krummholz as Alpine Vegetation. The rule set captured the Bare Rock sufficiently with the initial class assignment, and no re-segmentation or re-classification was required. The rule set re-segmented the Mixed Class – Bare Rock/Alpine Vegetation objects with a scale parameter of 25 and assigned classes based on NDVI value (< -0.2 for Bare Rock, -0.125 < NDVI < 0.125 for Alpine Vegetation). The rule set assigned the remaining Mixed Class – Bare Rock/Alpine Vegetation objects by brightness values (>= 120 for Bare Rock, < 120 for Alpine Vegetation, 8-bit image). The rule set re-segmented the Mixed Class – Alpine Vegetation/Subalpine Krummholz class with a scale parameter of 25 and assigned classes based on NDVI ranges and object context (relative border to). The rule set used -0.125 < NDVI < 0.125 to assign objects to the Alpine Vegetation class. The rule set assigned the remaining image objects with context. The rule set assigned objects to the class to which the image object shared the majority (> 50%) of its border with, either Aline Vegetation or Subalpine Krummholz. The rule set merged snow objects and re-segmented the snow objects based on NAIP_1M_CLRIR_2014 dataset (leaf-on conditions, no snow present). The rule set assigned the snow objects with NDVI < 0 and brightness > 150 (8-bit image) to Bare Rock. The rule set assigned snow objects with NDVI > 0 to Subalpine Krummholz. The rule set used object context (relative border to) to classify the remaining snow objects. The rule set assigned the snow objects to the class which the image object shared the majority (> 50%) of its border with, either Bare Rock or Subalpine Krummholz. The rule set assigned objects to either Bare Rock or Subalpine Krummholz (and not Alpine Vegetation) because it was difficult to use tone to delineate between Alpine Vegetation and Subalpine Krummholz in leaf-on conditions. The rule set assigned all non-tundra image objects in Area 2 to the Other class. The rule set separately merged the image objects in each class (Bare Rock, Alpine Tundra, Subalpine Krummholz, Other).
Process_Step:
Process_Description:
Step 3F: Shapefile Export and Delivery Preparation. Steps 3A-3E assigned all image objects in each area (Area 1 and Area 2) to one of four classes: Bare Rock, Alpine Vegetation, Subalpine Krummholz, Other. The rule set merged the image objects from Area 1 and Area 2 for each individual class. The rule set exported a shapefile with all four classes (Bare Rock, Alpine Vegetation, Subalpine Krummholz, Other) for use in the accuracy assessment. The rule set exported the shapefile with a coordinate system of NAD 1983(2011) State Plane Vermont FIPS 4400 (EPSG: 6589). The analysis projected the shapefile to NAD 1983 State Plane Vermont FIPS 4400 (EPSG: 32145) for the purpose of the accuracy assessment. The rule set exported a separate shapefile with the three features of interest (Bare Rock, Alpine Vegetation, Subalpine Krummholz) for the purpose of data delivery. The rule set exported the shapefile with a coordinate system of NAD 1983(2011) State Plane Vermont FIPS 4400 (EPSG: 6589). The analysis projected the shapefile to NAD 1983 State Plane Vermont FIPS 4400 (EPSG: 32145) for the purpose of the data delivery. The analysis used the ArcGIS Pro Check Geometry tool (with Esri Validation) to check for geometry problems in the shapefile set for data delivery. The analysis used the ArcGIS Pro Repair Geometry tool (with Esri Validation) to repair problems identified by the Check Geometry tool. After repairing geometry, the analysis used the Check Geometry tool (with Esri Validation) again to ensure geometry problems had been fixed. The only warning that appeared showed “WARNING 000442: could not find spatial index at -1 in Vermont_Mount_Mansfield_Alpine_Tundra_2013”. The analysis deemed this warning acceptable since the spatial index for the first record starts at 0, and the attribute table does not contain a record with a spatial index of -1.
Process_Step:
Process_Description:
Step 4: Accuracy Assessment. The analysis used the multinomial distribution defined by Congalton and Green (2009) to determine that 624 accuracy assessment points provided a sufficient sample size for the accuracy assessment. This calculation included the following considerations: 4 classes, 95th percentile confidence, 5% precision, and an assumption that one class consisted of at least 50% of the total study area. The analysis performed the accuracy assessment in ArcGIS Pro and used the following workflow: 1) Used the Create Accuracy Assessment Points tool with the Equalized Stratified Random sampling option to create a shapefile with 624 accuracy assessment points (156 for each mapped class). 2) Populated the reference class for each record in the shapefile attribute table through visual interpretation of reference imagery and manual assignment of the reference class. Reference layers for visual interpretation included Vermont Center for Geographic Information Web Services (MAP_VCGI_ALLIMAGERYCLR_SP_NOCACHE_v1, MAP_VCGI_ALLIMAGERYCIR_SP_NOCACHE_v1) and project data sources (VTORTHO_0_5M_CLRIR_2013, NAIP_1M_CLRIR_2014). Note that the Web Services provided the same imagery as the highest resolution imagery data sources used in the feature extraction, VTORTHO_0_2M_CLRIR_2013 and VTORTHO_0_5M_CLRIR_2013. The analysis used the Web Services to ensure continuous coverage of the study area during the accuracy assessment (the VTORTHO_0_2M_CLRIR_2013 imagery does not span the entire study area). The analysis hid the mapped class attribute for each record in order to remove the potential for bias while determining the reference class for each accuracy assessment point. 3) Used the Compute Confusion Matrix tool to create the error matrix. 4) Used the Table to Excel tool to save the error matrix in XLS format for viewing and formatting in Microsoft Excel. 5) Reformatted the error matrix in Microsoft Excel to contain the names of the features in the study and to improve the error matrix readability. Factors that could have affected the quantitative accuracy assessment include the minimum mapping unit (0.2 cm), varied and rapidly changing terrain on Mount Mansfield, variation in the size and shape of the features of interest, the general challenge of visual interpretation of features on Mount Mansfield, and/or the use of the same data sources for both data processing and accuracy assessment.
Spatial_Data_Organization_Information:
Direct_Spatial_Reference_Method: Vector
Point_and_Vector_Object_Information:
SDTS_Terms_Description:
SDTS_Point_and_Vector_Object_Type: GT-polygon composed of chains
Point_and_Vector_Object_Count: 321735
Raster_Object_Information:
Raster_Object_Type: Grid Cell
Spatial_Reference_Information:
Horizontal_Coordinate_System_Definition:
Planar:
Map_Projection:
Map_Projection_Name: NAD 1983 StatePlane Vermont FIPS 4400
Transverse_Mercator:
Scale_Factor_at_Central_Meridian: 0.9999642857142857
Longitude_of_Central_Meridian: -72.5
Latitude_of_Projection_Origin: 42.5
False_Easting: 500000.0
False_Northing: 0.0
Planar_Coordinate_Information:
Planar_Coordinate_Encoding_Method: coordinate pair
Coordinate_Representation:
Abscissa_Resolution: 0.0000000022208457295391785
Ordinate_Resolution: 0.0000000022208457295391785
Planar_Distance_Units: meter
Geodetic_Model:
Horizontal_Datum_Name: D North American 1983
Ellipsoid_Name: GRS 1980
Semi-major_Axis: 6378137.0
Denominator_of_Flattening_Ratio: 298.257222101
Entity_and_Attribute_Information:
Detailed_Description:
Entity_Type:
Entity_Type_Label: Land_MTMANSFIELDTUNDRA_poly
Entity_Type_Definition:
This table defines the land cover for alpine tundra on Mount Mansfield, Vermont
Entity_Type_Definition_Source: Esri
Attribute:
Attribute_Label: FID
Attribute_Definition: Internal feature number
Attribute_Definition_Source: Esri
Attribute_Domain_Values:
Unrepresentable_Domain:
Sequential unique whole numbers that are automatically generated
Attribute:
Attribute_Label: Shape
Attribute_Definition: Feature geometry
Attribute_Definition_Source: Esri
Attribute_Domain_Values:
Unrepresentable_Domain: Coordinates defining the features
Attribute:
Attribute_Label: Land_Cover
Attribute_Definition: Thematic classification
Attribute_Definition_Source: Esri
Attribute_Domain_Values:
Unrepresentable_Domain: String identifying the feature's land cover class
Distribution_Information:
Distributor:
Contact_Information:
Contact_Organization_Primary:
Contact_Organization: VT Center for Geographic Information
Contact_Address:
Address_Type: Mailing and Physical Address
Address: 1 National Life Dr, Dewey Building, 2nd Floor
City: Montpelier
State_or_Province: VT
Postal_Code: 05620-2001
Country: USA
Contact_Voice_Telephone: (802) 585-0820
Contact_TDD/TTY_Telephone: None
Contact_Facsimile_Telephone:
Contact_Electronic_Mail_Address: vcgi@vermont.gov
Hours_of_Service: 9am - 5pm, M-F
Resource_Description: LandLandcov_MTMANSFIELDTUNDRA
Distribution_Liability:
VCGI and the State of Vermont make no representations of any kind, including but not limited to the warranties of merchantability or fitness for a particular use, nor are any such warranties to be implied with respect to the data.
Standard_Order_Process:
Digital_Form:
Digital_Transfer_Information:
Format_Name: SHP
Format_Specification: ESRI Shapefile format
Format_Information_Content: geospatial data
File_Decompression_Technique: ZIP extraction tool
Digital_Transfer_Option:
Online_Option:
Computer_Contact_Information:
Network_Address:
Network_Resource_Name:
Access_Instructions: Download from web site.
Offline_Option:
Offline_Media: USB drive
Fees:
No charge when downloading from the internet, and when no custom processing is required.
Ordering_Instructions: Send an email describing data request to vcgi@vermont.gov
Turnaround: About 5 days.
Custom_Order_Process: Send an email describing data request to vcgi@vermont.gov
Technical_Prerequisites: GIS Software
Available_Time_Period:
Time_Period_Information:
Single_Date/Time:
Calendar_Date: 20190801
Time_of_Day: Unknown
Metadata_Reference_Information:
Metadata_Date: 20190801
Metadata_Contact:
Metadata_Contact:
Contact_Information:
Contact_Organization_Primary:
Contact_Organization: VT Center for Geographic Information
Contact_Person: GIS Database Administrator
Contact_Address:
Address_Type: Mailing and Physical Address
Address: 1 National Life Dr, Dewey Building, 2nd Floor
City: Montpelier
State_or_Province: VT
Postal_Code: 05620-2001
Country: USA
Contact_Voice_Telephone: (802) 585-0820
Contact_TDD/TTY_Telephone: None
Contact_Facsimile_Telephone:
Metadata_Standard_Name: FGDC Content Standard for Digital Geospatial Metadata
Metadata_Standard_Version: FGDC-STD-001-1998
Metadata_Time_Convention: local time
Metadata_Access_Constraints:
No access or use constraints. The originator makes no representations of any kind, including but not limited to the warranties of merchantability or fitness for a particular use, nor are any such warranties to be implied with respect to the metadata.
Metadata_Use_Constraints:
No use limitations. Users of this metadata should be aware that changes in the land cover may have occurred since the time period when the imagery and elevation data used for feature extraction was collected. Some parts of the data may no longer represent actual surface conditions. Users of this metadata should be aware of this limitation prior to using the data for additional analyses or studies.
Metadata_Security_Information:
Metadata_Security_Classification_System: N/A
Metadata_Security_Classification: Unclassified
Metadata_Security_Handling_Description: N/A
Generated by mp version 2.9.22 on Thu Aug 25 20:52:20 2022