Replaced EMDI Class A NPP 81 observational data with new revision from ORNL DAAC
The original EMDI Class A NPP 81 data provided by the ORNL DAAC contained
error as follows:
from Cook, Robert B. <cookrb@ornl.gov>
to jranders@uci.edu,
Forrest Hoffman <fmhoffman@gmail.com>,
"Hoffman, Forrest M." <hoffmanfm@ornl.gov>
cc "Hook, Leslie A." <hookla@ornl.gov>,
uso@daac.ornl.gov
date Fri, Oct 31, 2008 at 11:48 AM
subject revised EMDI data: please use this version
mailed-by ornl.gov
10/31/08 [NPP_EMDI_R1_20081028.html,EMDI_ClassA_NPP_81_R1.csv]
Hi Jim and Forrest,
We recently found out that the data we sent to you two years ago for
use in your model-data intercomparison activity has some problems.
One of our users discovered problems with the TNPP data for Class
A site tropical forest biomes. For a description of the problem
and what we've done to correct it are in the attached html file.
The attached csv file has the correct values. Please remove the
old file / data and use the new file.
We apologize for the inconvenience this may cause. We understand
that you have a paper in press that uses these data. Sorry about
this.
I am sending this note to you before we formally make the change
on our Web site, because you have a paper in press. When we place
the revision on our Web site, we will update the data set citation
and DOI.
Olson, R. J., J. M. O. Scurlock, S. D. Prince, D. L. Zheng,
and K. R. Johnson (eds.). 2008. NPP Multi-Biome: NPP and Driver
Data for Ecosystem Model-Data Intercomparison, R[evision]1. Data
set. Available on-line [http://www.daac.ornl.gov] from the Oak Ridge
National Laboratory Distributed Active Archive Center, Oak Ridge,
Tennessee, U.S.A. doi:10.3334/ORNLDAAC/xxx
Please let me know if you have any questions.
Thanks Bob
On Jan 26, 2009, I moved the original data.81 file to data.81.rev0.bad and
data.81.nc to data.81.rev0.bad.nc. Then I copied the file provided by the
ORNL DAAC, called EMDI_ClassA_NPP_81_R1.csv, to data.81 and re-generated
data.81.nc using the NCL script clamp/npp/01.read_ascii_81.ncl
Forrest Hoffman
Mon Jan 26 23:23:46 EST 2009
2 ##############################################################################
3 # University of Maryland Global Land Cover Classifications #
4 # at 0.25, 0.5 and 1 degree (1992-1993) #
5 ##############################################################################
7 The files in this directory were provided by Drs. Ruth DeFries and Matt Hansen
8 from the Department of Geography at the University of Maryland, College Park.
9 These are global, coarse resolution (1/4, 1/2 and 1 degree spatial resolutions)
10 land cover classifications (dominant type and fractional cover) generated using
11 data from the Advanced Very High Resolution Radiometer (AVHRR) (see Hansen et
14 The objective of this study was to create a land cover map derived from 1
15 kilometer AVHRR data using all available bands and derived Normalized Difference
16 Vegetation Index (NDVI) and a full year of data (April 1992-March 1993). This
17 thematic map was resampled to 0.25, 0.5 and 1.0 degree grids for the
18 International Satellite Land Surface Climatology Project (ISLSCP) data
19 initiative II. During this re-processing, the original UMD land cover type and
20 fraction maps were adjusted to match the water/land fraction of the ISLSCP II
21 land/water mask. These maps were generated for use by modelers of global
22 biogeochemical cycles and others in need of an internally consistent,
23 global depiction of land cover. This 1km map was also one of the MODIS (Moderate
24 Resolution Imaging Spectroradiometer) at-launch land cover maps.
26 The data set describes the geographic distributions of 13 classes of vegetation
27 cover (plus water and unclassified classes) based on a modified International
28 Geosphere-Biosphere Programme (IGBP) legend (Rasool 1992). The data set also
29 provides the fraction of each of the 15 classes within the coarser resolution
30 cells. The data set is provided at three spatial resolutions of 0.25, 0.5 and 1
31 degrees in latitude and longitude. We refer the reader to the reference below
32 for more information on data processing of the original 1km data set.
34 Please acknowledge Drs. DeFries and Hansen and the ISLSCP2 data collection when
35 these data are used. Please cite the following publications when these data are
38 Hansen, M. C., DeFries, R.S., Townshend, J.R.G., and Sohlberg, R., 2000, Global
39 land cover classification at 1km spatial resolution using a classification tree
40 approach, International Journal of Remote Sensing, 21, 1331-1364.
42 Any problems encountered with the data set should be reported to Eric Brown de
43 Colstoun at ericbdc@ltpmail.gsfc.nasa.gov
45 ##############################################################################
49 The data sets in this directory are provided at three spatial resolutions of
50 0.25, 0.5 and 1 degrees lat./long. For each spatial resolution there is a land
51 cover type classification layer (with numbers from 0 to 14) and 15 associated
52 layers that provide the fraction, from 0 to 100, of each land cover type per
53 cell. The land cover type files are named umd_landcover_class_XX.asc, where XX
54 is qd, hd, or 1d, denoting a spatial resolution of 1/4, 1/2 or 1 degrees,
55 respectively. The fractional files are called umd_landcover_XX_cZZ.asc, where XX
56 is the same as above, and ZZ is a number from 00 to 14 which represents the land
57 cover type code as described below. As an example, the file named
58 umd_landcover_qd_c02.asc is the fraction of Evergreen Broadleaf Forest at a
59 quarter degree spatial resolution. This file is associated with the
60 umd_landcover_class_qd.asc land cover type file.
62 ##############################################################################
66 All of the files in the ISLSCP Initiative II data collection are in the ASCII,
67 or text format. The file format consists of numerical fields of varying length,
68 which are delimited by a single space and arranged in columns and rows. The
69 values in the land cover class type files are written as integers from 0 to 14.
70 All values in the land cover fraction files are written as real numbers.
72 The files at different spatial resolutions each contain the following numbers of
74 One degree: 360 columns by 180 rows
75 1/2 degree: 720 columns by 360 rows
76 1/4 degree: 1440 columns by 720 rows
78 All files are gridded to a common equal-angle lat/long grid, where the
79 coordinates of the upper left corner of the files are located at 180oW, 90oN and
80 the lower right corner coordinates are located at 180oE, 90oS. Data in the files
81 are ordered from North to South and from West to East beginning at 180 degrees
82 West and 90 degrees North.
84 The data files are PKZip compressed. On UNIX, they can be decompressed using the
85 "unzip -a" command. On Windows, they can be decompressed using WinZip or other
86 PKZip software. On the Macintosh, they can be decompressed using Stuffit
89 ################################################################################
91 Land Cover Type Codes and Definitions
92 -------------------------------------
93 The following 15 classes are used. Note that here are no cells with a value of
94 13 (Urban and Builtup) in the 1degree land cover type file.
96 0 = Water bodies: oceans, seas, lakes, reservoirs, and rivers. Can be either
97 fresh or salt water. This class is derived from a land/water mask.
99 1 = Evergreen Needleleaf Forests: lands dominated by trees with a percent canopy
100 cover >60% and height exceeding 5m. Almost all trees remain green all year.
101 Canopy is never without green foliage.
103 2 = Evergreen Broadleaf Forests: lands dominated by trees with a percent canopy
104 cover > 60% and height exceeding 5m. Almost all trees remain green all year.
105 Canopy is never without green foliage.
107 3 = Deciduous Needleleaf Forest: lands dominated by trees with a percent canopy
108 cover >60% and height exceeding 5m. Trees shed their leaves simultaneously in
109 response to cold seasons.
111 4 = Deciduous Broadleaf Forests: lands dominated by trees with a percent canopy
112 cover >60% and height exceeding 5m. Trees shed their leaves simultaneously in
113 response to dry or cold seasons.
115 5 = Mixed Forests: lands dominated by trees with a percent canopy cover >60% and
116 height exceeding 5m. Consists of tree communities with interspersed mixtures or
117 mosaics of needleleaf and deciduous forest types. Neither type has <25% or >75%
120 6 = Woodlands: lands with herbaceous or woody understories and tree canopy cover
121 of >40% and <60%. Trees exceed 5m in height and can be either evergreen or
124 7 = Wooded Grasslands/Shrublands: lands with herbaceous or woody understories and
125 tree canopy cover of >10% and <40%. Trees exceed 5m in height and can be either
126 evergreen or deciduous.
128 8 = Closed Bushlands or Shrublands: lands dominated by bushes or shrubs. Bush and
129 shrub percent canopy cover is >40%. Bushes do not exceed 5m in height. Shrubs
130 or bushes can be either evergreen or deciduous. Tree canopy cover is <10%. The
131 remaining cover is either barren or herbaceous.
133 9 = Open Shrubland: lands dominated by shrubs. Shrub canopy cover is >10% and
134 <40%. Shrubs do not exceed 2m in height and can be either evergreen or
135 deciduous. The remaining cover is either barren or of annual herbaceous type.
137 10 = Grasslands: lands with continuous herbaceous cover and <10% tree or shrub
140 11 = Croplands: lands with >80% of the landscape covered in crop-producing fields.
141 Note that perennial woody crops will be classified as the appropriate forest or
142 shrubs land cover type.
144 12 = Barren: lands of exposed soil, sand, rocks, snow or ice which never have more
145 than 10% vegetated cover during any time of year.
147 13 = Urban and Builtup: land covered by buildings and other man-made structures.
148 Note that this class is not mapped from the AVHRR imagery but is developed from
149 the populated places layer that is part of the Digital Chart of the World (Danko
152 14 = Unclassified: Points where the ISLSCP2 land/sea mask was labeled as land and
153 the UMD original data had sea and which could not be filled in from a 3 by 3
154 average of surrounding cells.