TITLE: Aggregated land cover characteristic data set for the Conterminous United States: Dominant land cover classes and associated fractional green vegetation cover. AUTHOR(S): Kevin Gallo, NOAA/NESDIS, Office of Research and Applications, mailing address: USGS EROS Data Center, 47914 252nd St., Sioux Falls, SD 57198 (email: Kevin.P.Gallo@noaa.gov) Timothy Owen, NOAA, National Climatic Data Center, Asheville, NC 28801 Garik Gutman, NASA Headquarters, Washington, D.C. 20546 Brad Reed, Raytheon Inc., USGS EROS Data Center, Sioux Falls, SD 57198 1.0 DATA SET OVERVIEW: The land cover classes developed under the coordination of the International Geosphere-Biosphere Programme Data and Information System (IGBP-DIS) have been analyzed for a study area that includes the Conterminous United States and portions of Mexico and Canada. The 1-km resolution data available from the IGBP-DIS global 1 km land cover data set have been analyzed to produce a 20-km resolution gridded data set that includes the three most dominant land cover classes within each 20-km grid cell. Additionally, the fractional area associated with each of the three most dominant land cover classes as well as the area covered by water, was computed for each 20-km grid cell. The fractional green vegetation cover (fgreen) associated with each of the three most dominant land cover classes per grid cell was computed on a monthly basis. The mean and standard deviation of fgreen were developed from monthly values of the normalized difference vegetation index from 5-years (1993, 1995-98) of data derived from 1-km resolution NOAA-AVHRR data. 2.0 INSTRUMENT DESCRIPTION: NOAA AVHRR (http://edc.usgs.gov/glis/hyper/guide/avhrr) 3.0 DATA COLLECTION AND PROCESSING: A detailed description of the development of this data set and can be found in Gallo et. al., (2001). The land cover information for this study was extracted from the IGBP-DIS global land cover data set (Loveland and Belward, 1997). The data were reprojected from the original Interrupted Goode Homolosine map projection to a Lambert Azimuthal Equal Area map projection. A. Land Cover The 17 IGBP classes were condensed into 14 classes (CIGBP) by combining several of the classes (Table 1). The classes were combined due to a concern that a grid cell comprised of many of the more specific classes (e.g., 10% aerial coverage of each of the five forest classes) and a single general class (e.g., 30% aerial coverage of grasslands) would be defined as predominantly grassland although a generic forest class (with 50% aerial coverage) realistically could be defined as the predominant class. IGBP classes 4 (Deciduous Broadleaf Forest) and 5 (Mixed Forest) were combined with class 1 (Evergreen Needleleaf Forest). Additionally class 14 (Cropland/Natural Vegetation Mosaic) was combined with class 12 (Croplands). The combined classes displayed similar seasonal characteristics in a subjective analysis of an annual time series of the normalized difference vegetation index at monthly intervals over the Conterminous USA. Comparisons within specific regions indicated greater similarity. The three most dominant classes, excluding water bodies, were determined for each 20 by 20 km grid cell in the study area. The proportion of the grid cell area occupied by each of the three most dominant classes, and by water bodies, was also determined. Thus, from the available proportional areas, the area occupied by any one of the three most dominant classes in a grid cell can be derived relative to the total area (including water bodies). B. Fractional green vegetation cover The fraction of green vegetation per grid cell (fgreen, Equation 1), fgreen = (NDVI - NDVIo)/(NDVImax - NDVIo) [1], is defined as the fraction of horizontal area associated with the photosynthetically active green vegetation that occupies a model grid cell. Fgreen as presented in [1], and described by Gutman and Ignatov (1998), has been demonstrated useful in coupled land-atmosphere models (Betts et al., 1997 ). Fgreen requires maximum and minimum thresholds of NDVI. Selection of the maximum (NDVImax) and minimum (NDVIo) thresholds of NDVI required for computation of fgreen were based on NDVI values of the conterminous USA 1-km AVHRR data set (Eidenshink, 1992) in combination with the IGBP land cover classes (Loveland and Belward, 1997) developed from a global 1-km AVHRR data set (Eidenshink and Faundeen, 1994). An interpolated mid-month NDVI data set was prepared from the biweekly Conterminous USA data set (Eidenshink, 1992). All biweekly data available from 1993 and 1995 through 1998 were interpolated on a daily basis and temporally processed to remove fluctuations in the NDVI signal due to cloud or other contamination (Swets, et al., 1999). The mid-month NDVI value was retained from the interpolated daily values. Data of 1994 were excluded from the analysis due to low solar elevation angles associated with the AVHRR acquisitions during this year. NDVIo and NDVImax values of 0.09 and 0.69, respectively, were developed as described in Gallo et al. (2001). The mid-month NDVI values were averaged on a monthly basis over the 5 available years of data. Monthly values of fgreen were computed [equation 1] from the 5-year monthly NDVI values with NDVIo = 0.09 and NDVImax = 0.69. When observed NDVI values were less than 0.09 fgreen was assigned a value of 0.0, and when observed values were greater than 0.69 fgreen was assigned a value of 1.0. 4.0 DATA FORMAT: Data Set Definition and Access A. Land Cover and Fractional Green Vegetation The data set includes binary files mapped in a Lambert Azimuthal Equal Area map projection. The data set includes files (Table 2) with the three most dominant land cover types for each 20-km grid cell and the proportion of the grid cell area (excluding water) associated with each of the three most dominant classes. An additional file includes the proportion of the grid cell area associated with water. The data set includes fgreen (and standard deviation) files (Table 2) for each month and each of the three most dominant land cover classes. B. Product Definition There are full and reduced resolution data sets provided. The full resolution data set includes 1-km resolution data to enable more accurate remapping of the data. Thus, for this full resolution data set the 1-km pixels associated with each 20-km grid cell all have redundant values. The reduced resolution data set has one pixel represent each 20-km grid cell. The fgreen values are scaled such that a value of 0 represents non-valid data or water. A value of 100 represents a value of 0% fgreen and a value of 200 represents a value of 100% fgreen. Each one unit increment between 100 and 200 corresponds to an increase of 1% fgreen. Thus, a value of 125 corresponds to 25% fgreen and 175 corresponds to a value of 75% fgreen. There are no values between 0 and 100. The values of the standard deviation of fgreen are similarly scaled such that a data value of 100 represents a standard deviation of fgreen of 0% and a value of 200 represents a standard deviation of fgreen of 100%. File size and projection specifics are provided in Table 3a and 3b for the full and reduced resolution data sets, respectively. The files are binary files of byte data. ___________________________________________________________________________ Table 1. Current IGBP and condensed IGBP (CIGBP) classes. Numbers associated with condensed IGBP classes are those included in the digital data set files. ___________________________________________________________________________ IGBP classes Condensed IGBP classes ------------ ------------------------------------- 1 Evergreen Needleleaf Forest 0 Water Bodies/No Class Values 2 Evergreen Broadleaf Forest 1 Forests 3 Deciduous Needleleaf Forest 2 Evergreen Broadleaf Forest 4 Deciduous Broadleaf Forest 3 Deciduous Needleleaf Forest 5 Mixed Forests 6 Closed Shrublands 6 Closed Shrublands 7 Open Shrublands 7 Open Shrublands 8 Woody Savannas 8 Woody Savannas 9 Savannas 9 Savannas 10 Grasslands 10 Grasslands 11 Permanent Wetlands 11 Permanent Wetlands 12 Croplands/Natural Vegetation Mosaic 12 Croplands 13 Urban and Built-Up 13 Urban and Built-Up 15 Snow and Ice 14 Cropland/Natural Vegetation Mosaic 16 Barren or Sparsely Vegetated 15 Snow and Ice 16 Barren or Sparsely Vegetated 17 Water Bodies ____________________________________________________________________________ __________________________________________________________________________ Table 2. File names associated with the aggregated land cover characteristic and fgreen data set, file contents, and range of grid cell values. __________________________________________________________________________ File Name File Contents Range of Values --------- --------------------------------------------- --------------- igbpc1.img most dominant class per 20 km grid cell 0 to 16 igbpc2.img second most dominant class per 20 km grid cell 0 to 16 igbpc3.img third most dominant class per 20 km grid cell 0 to 16 igbpp1.img percent area per 20 km grid cell of most dominant class 0 to 100 igbpp2.img percent area per 20 km grid cell of second most dominant class 0 to 100 igbpp3.img percent area per 20 km grid cell of third most dominant class 0 to 100 igbppw.img percent area per 20 km grid cell defined as water 0 to 100 fg011.img fgreen values, month 01, for most dominant class 0 to 200 fg012.img fgreen values, month 01, for second most dominant class 0 to 200 fg013.img fgreen values, month 01, for third most dominant class 0 to 200 fg021.img fgreen values, month 02, for most dominant class 0 to 200 : : : : : : : : : : : : : : : fg123.img fgreen values, month 12, for third most dominant class 0 to 200 A similar set of files exists with "r" added to name (e.g., igbpcr1.img, fgr011.img) at the reduced resolution of 20 km per grid cell. Standard deviation files are designated similar to the fgreen files with the addition of an "s" in the name, e.g., fgs001.img for the full resolution and fgsr001.img for the reduced resolution image. __________________________________________________________________________ ________________________________________________________________________ Table 3a. Lambert Azimuthal Equal Area (LAZEA) Projection Parameters for the 1 km resolution image files (full data set). ________________________________________________________________________ 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. Center of pixel (1,1) ( -2050000, 752000 ) (48.40051, -128.52118) Number of lines (rows) 2889 Number of samples (columns) 4587 LAZEA corner coordinates, center of pixel: (meters and degrees, x y) Upper Left ( -2050000, 752000 ) Lower Left ( -2050000, -2136000 ) Upper Right ( 2536000, 752000 ) Lower Right ( 2536000, -2136000 ) Upper Left ( 48.401, -128.521 ) Lower Left (23.589, -119.968 ) Upper Right ( 46.705, -65.404 ) Lower Right ( 22.485, -75.420 ) ________________________________________________________________________ ________________________________________________________________________ Table 3b. Lambert Azimuthal Equal Area (LAZEA) Projection Parameters for the 20 km resolution image files (reduced data set). ________________________________________________________________________ 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 20000 meters For the Conterminous U.S. Center of pixel (1,1) ( -2040000, 742000 ) (48.34938, -128.34383) Number of lines (rows) 145 Number of samples (columns) 230 LAZEA corner coordinates, center of pixel:(meters and degrees, x y) Upper Left ( -2050000, 752000 ) Lower Left ( -2050500, -2128000 ) Upper Right ( 2530000, 752000 ) Lower Right ( 2530000, -2128000 ) Upper Left ( 48.401, -128.521 ) Lower Left ( 23.660, -119.982 ) Upper Right ( 46.726, -65.476 ) Lower Right ( 22.570, -75.460 ) ________________________________________________________________________ 5.0 DATA REMARKS: none 6.0 REFERENCES: Betts, A., F. Chen, K. Mitchell, and Z. Janji. 1997. Assessment of Land Surface and Boundary Layer Models in Two Operational Versions of the NCEP Eta Model Using FIFE Data. Mon. Wea. Rev. 125:2896-2915. Eidenshink, J.E., 1992. The 1990 Conterminous U.S. AVHRR Data Set. Photogram. Engr. & Rem. Sensing. 58: 809-813. Eidenshink, J.E., and J.L. Faundeen, 1994. The 1 km AVHRR global land data set: first stages in implementation. Intl. J. Rem. Sens. 15:3443-3462. Gallo, K., D. Tarpley, K. Mitchell, I. Csiszar, T. Owen, and B. Reed. 2001. Monthly fractional green vegetation cover associated with land cover classes of the conterminous USA. Geophys. Res. Lett. 28:2089-2092. Gutman, G. and A. Ignatov. 1998. Derivation of Green Vegetation Fraction from NOAA/AVHRR for use in Numerical Weather Prediction Models. Intl. J. Remote Sensing. 19:1533-1543. Loveland, T.R. and A.S. Belward, 1997. The IGBP-DIS global 1 km land cover data set, DISCover: first results. Intl. J. Remote Sensing. 18: 3289-3295, 1997. Swets, D.L., B.C. Reed, J.D. Rowland and S.E. Marko. 1999. A Weighted Least-Squares Approach to Temporal NDVI Smoothing. Proceedings Amr. Soc. Photogram. Rem. Sens. 17-21 May, Portland OR., ASPRS, Washington, D.C. pp. 526-536.