U.S. Land Cover Characteristics 1990 Prototype (LANDCOV)
Table of Contents
The Conterminous U.S. Land Cover Characteristics Data Set 1990 Prototype is a classification of seasonal land cover types (each with a distinct onset, peak, and duration of greenness) at 1-km resolution, developed from a combination of multitemporal satellite data with a variety of earth science data sets, including climate, elevation, and ecoregions. The data set was developed by the U.S. Geological Survey (USGS) EROS Data Center (EDC) and the Center for Advanced Land Management Information Technologies University of Nebraska-Lincoln.
The coverage area is the conterminous United States.
Upper Left 48.40051 N 128.52118 W Lower Left 23.58922 N 119.96836 W Upper Right 46.70283 N 65.40354 W Lower Right 22.48502 N 75.42005 W
The base image of the conterminous U.S. used in this data set was produced by transformation of the (USGS) 1:2,000,000 scale Digital Line Graph (DLG) data to the Lambert Azimuthal Equal Area map projection.
The original source data used for classification of seasonally distinct land cover regions were National Oceanic and Atmospheric Administration (NOAA) Advanced Very High Resolution Radiometer (AVHRR) High Resolution Picture Transmission (HRPT) satellite imagery. Multitemporal AVHRR imagery is especially suitable as a base for land cover analyses at continental and global scales because of its moderately coarse spatial resolution (1-km), combined with a relatively fine temporal resolution. Observations of any point on the Earth's surface are collected daily by the NOAA polar orbiting satellites across a wide swath (2,700 km).
This allows multiple-day compositing of AVHRR images which minimizes effects of cloud contamination without significant loss of information on short-term and seasonal processes such as vegetative growth and development, deforestation, crop harvesting cycles, or regional drought.
The preliminary classification of land cover characteristics in the 1990 prototype study was derived from a time-series of eight AVHRR 28-day maximum Normalized Difference Vegetation Index (NDVI) composite images spanning the 1990 growing season (March 16 - October 25). When working with the very wide diversity of climate, soils, and terrain found at global or continental scales, however, cluster confusion occurs with AVHRR data alone, and additional sources of information are necessary to distinguish and characterize land cover regions. The 70 spectral-temporal classes of the preliminary classification were subsequently refined into 159 seasonally distinct regions or classes using ancillary data sets and a variety of statistical and visualization tools. Tables of descriptive and quantitative attributes were developed for the 159 final land-cover classes as part of this process.
Preliminary Image Classification
The AVHRR time-series of eight 28-day maximum NDVI composites for March-October 1990 were initially stratified into vegetated and non-vegetated regions using a maximum NDVI image representing this 8-month period. Biweekly AVHRR composites were also used in the process of land cover class refinement (for more information on the AVHRR time-series data program see Eidenshink, 1992, or the Global Land Information System (GLIS) Guides AVHRR and USAVHRR Next, an unsupervised classification using a clustering algorithm (ISOCLASS) and minimum-distance-to-mean classifier was used to define 70 spectral-temporal (seasonally distinct) classes within the vegetated stratum.
Class Labeling and Description
Evaluation and characterization of the initial 70 classes utilized a variety of statistical and visual tools and techniques. Each of the 70 classes was evaluated for its spatial distribution, specific phenological characteristic and association with ancillary map data, images, and statistics. Each spectral-temporal class was correlated with ecoregions, major land resource areas (MLRAs), and USGS Land Use and Land Cover (LULC) regions for initial labeling and description.
Post Classification Refinement
In about 75 percent of the preliminary classes, spectral-temporal classes were not uniquely associated with land-cover types. For example, irrigated agriculture has a spectral-temporal signature similar to deciduous forest. Similarly, warm-season desert grasslands are difficult to distinguish from alpine meadows on the basis of spectral-temporal characteristics alone. In these cases, ancillary data on ecosystems, topography, and climate were used to sort and identify specific cover types, thus refining the 70 initial classes into 159 seasonally distinct land cover regions.
The final characterization of the 159 land cover regions was completed by development of descriptive and quantitative attribute tables, correlating each of the 159 final classes with source data set attributes and summary statistics (for more information on this process, see Loveland and others, 1991 and Brown and others, 1993).
Several derivative thematic data sets were also developed as part of this process, and are included in the Conterminous U.S. Land Cover Characteristics Data Set 1990 Prototype, along with the major source data sets, preliminary and final classifications, and attribute tables.
The Conterminous U.S. Land Cover Characteristics Data Set 1990 Prototype CD-ROM contains both raster image data files and standard ASCII text files. Raster data files are binary image files in a flat headerless format. Data dimensions are 2,889 lines (rows) by 4,587 samples (columns). Data type may be 8-bit or 16-bit depending on the file. File size is approximately 13 megabytes for the 8-bit images and over 26 megabytes for 16-bit images.
All raster images are registered to the Lambert Azimuthal Equal-Area map projection.
Text files are provided in two standard ASCII output formats (comma-delimited and text report). The ASCII records available as comma-delimited files have character fields enclosed in double quotes and column entries separated by commas. These files are designed for import into spreadsheet or database management programs.
Text report files, like the README files given at each directory level, are standard ASCII text suitable for printout or display.
This data set is derived from AVHRR 1-km resolution time-series data. Ancillary data sets utilized in the characterization process vary from 1:250,000 to 1:7,500,000 in original map scale, but all resultant data sets preserve 1-km spatial resolution.
The preliminary land cover classification was based upon a time-series of eight 28-day maximum NDVI composites spanning the 1990 growing season. Dates for the eight compositing periods were:
Mar 16 - Apr 12 Apr 13 - May 10 May 11 - Jun 07 Jun 08 - Jul 05 Jul 06 - Aug 02 Aug 03 - Aug 30 Aug 31 - Sep 27 Sep 28 - Oct 25
To place orders and obtain additional information regarding technical details and pricing schedules, contact:
Customer Services, EROS Data Center
Online requests for these data can be placed via the USGS Global Land Information System (GLIS) interactive query system. The GLIS system contains metadata and online samples of Earth science data. With GLIS, you may review metadata, determine product availability, and place online requests for products.
The Conterminous U.S. Land Cover Characteristics Data Set 1990 Prototype is available from the EROS Data Center on CD-ROM. The CD-ROM contains source data sets, the initial and final classifications of land cover regions, tables of descriptive and quantitative attributes for each of the final land cover classes, and several derived thematic data sets.
Land cover data utilizing multi-source data at continental and global scales can be used as input to atmospheric mesoscale and general circulation models, hydrologic and ecological models, and studies of broad-scale land surface processes such as crop development, deforestation, and regional drought or desertification.
This data set is intended to complement, rather than replace, other land cover classifications, including those derived from higher-resolution satellite imagery such as Landsat or SPOT. The concept of the seasonally distinct land cover region as used in this data set is fundamentally different than that used in most spectral classifications of remotely sensed data. Independently derived classifications of land cover may differ in number and nature of classes, their spatial or temporal scale, and in the presence or treatment of mixtures and classes. Other classifications of land cover and vegetation may have been developed using single season observations and/or many years of data, and probably employ quite different data analysis strategies.
Several of the source data sets used in this study are available from EDC as separate products. Specifically, the data on this CD-ROM complements the USAVHRR CD-ROMs (biweekly composites 1989 to present) and the AVHRR Companion Disc produced by the USGS. In addition, there are a number of regional data sets based on AVHRR data. For more information, see the following GLIS Guides: AVHRR, USAVHRR, 30ASDCWDEM, MLRA, ECOREGIONS, 1_250_LULC, NGP88, GLEND, and AKAVHRR.
Longitude of central meridian 100 00 00 W Latitude of origin 45 00 00 N False Easting 0 False Northing 0 Units of measure Meters Pixel size 1000 meters For the Conterminous U.S. data set: X Y Center of pixel (1,1) ( -2050000, 752000 ) Number of lines 2889 Number of samples 4587 LAZEA minimum bounding rectangle: Lower Left ( -2050500, -2136500 ) Upper Left ( -2050500, 752500 ) Upper Right ( 2536500, 752500 ) Lower Right ( 2536500, -2136500 ) Lower Left ( -119.9722899, 23.5837576 ) Upper Left ( -128.5300591 48.4030555 ) Upper Right ( -65.3946489 46.7048989 ) Lower Right ( -75.4163527 22.4793919 ) Lower Left ( -119 58 20 23 35 02 ) Upper Left ( -128 31 48 48 24 11 ) Upper Right ( -65 23 41 46 42 18 ) Lower Right ( -75 24 59 22 28 46 )
Advanced Very High Resolution Radiometer (AVHRR) High Resolution Picture Transmission (HRPT) data are received daily from the NOAA polar orbiting satellites at the U.S. Geological Survey's EROS Data Center (EDC), for the entire conterminous United States, southern Canada, and northern Mexico. Daily observations in 5 channels are calibrated to reflectance, scaled to byte data, composited and registered to a common map projection.
These data may be used to compute an index of vegetative growth or greenness, the normalized difference vegetation index (NDVI). NDVI is defined as the difference of near-infrared (AVHRR Channel 2) and visible (AVHRR Channel 1) reflectance values, divided by total reflectance:
IR(Band 2) - Visible(Band 1) NDVI = ------------------------------ IR(Band 2) + Visible(Band 1)
This equation produces values in the range of -1.0 to 1.0, where negative values generally represent non-vegetated surfaces. Computed values are rescaled to an 8-bit data range of 0 - 200, such that computed -1.0 equals 0, computed 0 equals 100, and computed 1.0 equals 200. Rescaled NDVI values less than 100 now represent clouds, snow, water, and other non-vegetated surfaces.
Maximum NDVI composite images are constructed by registering a set of daily observations to a common cartographic projection, then retaining the highest NDVI value reached at each pixel over a specified compositing period. This method reduces the number of cloud contaminated pixels, because clouds and cloud shadows generally have low values (less than 100, in byte scaled data) while clear day observations of vegetated surfaces have values greater than 100 (in the byte scaled data). The resulting composite image is a near cloud free record of vegetative greenness for the compositing time period.
Length of compositing period determines the temporal resolution of these observations. The most commonly used compositing periods are 10-day, 14-day (biweekly) and 28-day (monthly), but periods from 5 days to a year in length are used for studies of vegetation, land cover, and environmental change.
For more information on the AVHRR time-series data program, see Eidenshink, 1992, or the following GLIS Guides: AVHRR, NGP88, USAVHRR, GLEND, and AKAVHRR.
Process_Step: Process_Description:IMAGEGRID LCCALBERS_2.TIF
Process_Step: Imported grid into ArcInfo
Process_Step: In ArcInfo (Tables), added an item to the .vat
Process_Step: Filled in COVTYPE values for each record
Process_Step: Projected grid into new projection