Appo 24.0 Faye 31.9 Pete 61.6 Bayb 21.5 Flor 11.8 Plym 27.0 Capf 0.4 Gree 30.1 Rale 68.4 Caph 1.8 Hens 74.2 Rich 2.7 Chap 35.2 Kins 28.6 Roan 85.0 Curr 4.1 Laur 6.3 Rock 37.4 Danv 6.7 Mant 6.0 Sbos 35.1 Dill 4.0 More 17.7 Spin 12.4 Elic 35.5 Myrt 5.3 Virg 1.8 Elit 29.8 Norf 47.0 Whit 27.9 Empo 53.7 Nriv 23.6 Wilb 31.1 Wilm 8.0 Landsat TM Scenes (Path and Row, Date Image Captured, Image id number): Path 14: Row 35; November 3, 1988; id# Y5170815114XO Path 14: Row 36; December 5, 1988; id# Y5174015115XO Path 15: Row 34; January 8, 1987; id# Y5104315062XO Path 15: Row 35; November 24, 1987: id# Y5136315152XO Path 15: Row 36; December 15, 1989; id# 89349 Path 16: Row 35; December 3, 1988; id# Y5173815235XO Certain classifications were more accurate than others, and a priority to the mosaicking was established. The following is a listing of which classified raster scenes were overlayed onto others to maximize accuracy during the mosaicking process. Scene B overlayed Scenes A, C, and D Scene C overlayed Scenes A and D Scene E overlayed Scene C Scene F overlayed Scenes A, C, and ESee full documentation for this data layer in A/P report "Albemarle-Pamlico Estuarine Study, mapping and GIS development of land use and land cover categories for the Albemarle-Pamlico drainage basin, Report No. 91-08," March 1992, NC Department of Environment, Health, and Natural Resources.
All the "raw" data (raster, unfiltered, classified, ERDAS-.GIS and .TRL files, divided by scene) are backed up on tape at NCCGIA.
Full coverage: all counties in the APES area west to Wake County, plus Person, Orange, Durham, and Cumberland. Partial coverage: Caswell, Alamance, Chatham, Lee, Harnett, Hoke, and Robeson Coverage also includes the APES area in Virginia. NOTE: Data is stored by 1:100,000-scale tile.
system filename: /lulc/liblulc/tiles/<quadabbrev>/lulc87
Revisions and updates to this layer include:
2.) In March 1998 this data was projected to 83, state plane meters using the ARC/INFO project command. No other changes were made. 1.) The May 1994 update added new areas and superceded existing areas of coverage. Scene F had a standard 5 x 5 pixel filter performed on the classified raster data (the same as the other scenes) and was subdivided into 1:100,000-scale tiles. It was then converted to vector ARC/INFO coverages. The following 100k quads were updated with this addition: Southern Pines, Fayetteville, Kinston, and New River. The following 100k quads were new additions to the data: Laurinburg, Elizabethtown, Florence, Whiteville, Wilmington, Myrtle Beach, and Cape Fear. NOTE: The Florence coverage actually extends farther south into South Carolina than its 100k neatline border normally permits.
Computer Graphics Center Director, Siamak Khorram Computer Graphics Center Staff, Heather Cheshire North Carolina State University Raleigh, Norht Carolina EOSAT President, Arturo Silvestrini EOSAT Staff, Brenda S. Burroughs Lanham, Maryland Database Administrator, Zsolt Nagy Project Manager, David Giordano North Carolina Center for Geographic Information and Analysis Governor's Office Office of State Planning Street 301 North Wilmington Street, Suite 700 Raleigh, North Carolina 27601-2825 Albemarle-Pamlico Estuarine Study APES Coordinator, Guy Stefanski APES Staff, Joan Giordano Dept. of Environment and Natural Resources; Div. of Environmental Management; Water Quality Section's Planning Branch Raleigh, North Carolina 27611
Data was converted from LAS to ERDAS format at NCSU's Computer Graphics Center prior to delivery to NCCGIA. Data was mosaicked and filtered using ERDAS version 7.3. It was then vectorized to ESRI's Arc/Info version 5.0 GIS software where it was clipped by USGS 1:100,000 quadrangle boundaries. Map units are feet,single precision, fuzzy tolerance is 1.00 and dangle tolerance is 0.0. This data is presently stored in ESRI's ARC/INFO, version 7.0.3 GIS software, on a UNIX, SUN workstation platform.
CGC converted the raw data from Space Oblique Mercator projections to Lambert Conformal Conic, North Carolina State Plane coordinates. CGC used the Land Analysis System (LAS) image processing software to classify the raw data. Ancillary resources such as NHAP color infrared aerial photographs, USGS 1:24,000 topographic and orthophoto maps, and NC DOT highway maps were used to assist in the classification. Geographic areas which had pixel values that shared a unique ground reflectance were identified and grouped together in what is known as training sites. These training sites of a minimum mapping unit of 5 to 10 acres were universally applied in a supervised classification across the entire raw data subset to identify every pixel.
After the classification was complete, a "ground truthing" procedure was initiated using the NHAP photography and error matrices or confusion tables were constructed. The results of these findings are in A/P report "Albemarle-Pamlico Estuarine Study, mapping and GIS development of land use and land cover categories for the Albemarle-Pamlico drainage basin, Report No. 91-08," March 1992, NC Department of Environment, Health and Natural Resources.
The conclusion of CGC's role for the project was to provide NCCGIA with magnetic tapes of the original raw data, and with the classified raster data which had been converted from LAS software to ERDAS software.
Because of differing error amounts by scene during classification, certain scenes intentionally "overwrote" others during this process to maximize accuracy. For example, in the Bayboro 100k there were areas of cloud masking from one scene overwritten by a "better" scene without masking. Determination of which scenes should overlay others are found in the explanatory notes in the beginning of this metadata report under 1.2.3 Supplemental_Information.
Once the subsetted 100k areas had been established, a 5x5 majority pixel filter was run to reduce the "spectral noise". This amount was decided based on test plots showing a focus area at 1:24,000-scale and raw data filtered at 3x3, 5x5, 7x7, and 9x9. The 3x3 filter showed too much spectral noise still existed. The 7x7 and 9x9 filters smoothed out too much information. It was decided the 5x5 filter was an adequate amount of noise reduction for purposes of the project.
After the filtering of the subsetted 100k areas was complete, the raster data was then converted over to vector data using ESRI's Arc/Info "ERDASSVF" and "GRIDPOLY" commands. The resulting coverages contained squared-off polygons that were originally pixels of like-value. Each polygon contained a label point that indicated the land use/land cover classification value.
The last step to completion was to clip the subsetted coverage with the actual USGS 1:100,000-scale neatline boundary. This eliminated all extraneous polygons so that the resulting coverages were properly tiled together. The item "gridpoly" found in the .PAT of each coverage was altered in INFO to become "lu#".
LANDUSE.PAT Polygon Attribute Table COLUMN ITEM NAME WIDTH OUTPUT TYPE DEC DESCRIPTION 1 AREA 4 12 F 3 Total area in feet 5 PERIMETER 4 12 F 3 Total perimeter in feet 9 LULC87# 4 5 B - Poly internal id number 13 LULC87-ID 4 5 B - Poly user id number 17 LU# 2 5 B - Classification number, relates to classification name 19 CO# 2 5 B - County Federal Information Processing Standards code 21 SB# 2 3 B - Subbasin code (8-digit)