"eol
Biosphere
stem(DIS) :
for global
ther hand,
À griculture
nework of
y deciding
se of land
1996). In
int role in
nce 1981.
\VHRR as
. Set from
ly covered
usable for
itation due
obal 1km
292. This
lapping in
effect by
some new
uch as
JIS(1998),
ite data for
id change
; introduce
land cover
cessing of
detection.
referred for
1. GLOBAL LAND COVER DATA SET
(4 ARC MINUTE GRID)
1.1 LCWG/AARS
The Land Cover Working Group(LCWG) of the Asian
Association on Remote Sensing(AARS) started its
activity in 1993 and released the first product, "AARS
Global 4-minute Land Cover Data Set with Ground Truth
Information" by a CD-ROM in 1997(Tateishi 1997).
The CD-ROM is available from the authors. The basic
idea of the LCWG/AARS is to develop land cover data
set of the whole Asia by Asian scientists and to
contribute to the development of global data set by
providing data set and information of Asia.
1.2 Land Cover Classification System
The land cover classification system is shown in Table 1
which was developed through discussion with members
of the LCWG/AARS.
The land cover classification system was developed based
on the following concepts.
The LCWG/AARS proposes general land Cover
classification system which meets both scientific and
social needs. For scientific needs, the proposed
classification system has similar land cover classes to
IGBP-DIS land cover classification system. Key land
cover class for social needs is cropland. In the proposed
LCWG/AARS classification system, cropland is
basically divided into three types such as tree crops, shrub
crops, and grass crops in order to match the classification
system for scientific needs.
The classification system of this CD-ROM consists of
60 classes including 48 classes for vegetation, 8 classes
for non vegetation, and 4 classes for water. Addition of
new classes up to 255 is possible. Class code is
recorded in one byte.
1.3 Ground truth collection
Ground truth data in this study means geographically
specified regions which are identified one of classes in the
land cover classification system by class code.
Collection of good ground truth data is a key issue for
reliable land cover mapping.
Ground truth data were collected mainly from existing
maps by the cooperation of the working group members.
Few ground truth were collected by field survey in
Central Asia.
Ground truth data of 35 land cover classes were collected.
Geographical regions of each ground truth data and the
information sources for each ground truth data are recorded
in the distributed CD-ROM.
1.4 AVHRR data and their preprocessing
Global 8 km 10 days composite AVHRR NDVI data
from January 1, 1990 to December 31, 1990 were used
for the development of land cover data set. The cloud flag
data for each pixel were also used for composite
processing. These AVHRR data are a part of the
NOAA/NASA Pathfinder AVHRR Land Data Set (Agbu
1994).
The map projection of the NOAA/NASA Pathfinder
AVHRR Land Data Set is based on the Interrupted Goode
Homolosine projection. The data were transformed to
latitude/longitude projection(Plate Carree projection) with
four minutes resolution and resampled to the Digital
Chart of the World(DCW) for the geometric registration .
The accuracy after geometric registration is 0.6 pixels
(one pixel : 4 minute grid) when compared with DCW at
several points in global land area between 75 degrees
north latitude and 55 degrees south latitude.
In order to produce a cloud free NDVI image and to
remove noises in an image, a monthly composite data
was generated from 10-days composite data. Monthly
composite is carried out pixel by pixel using three 10-
days composite NDVI data in a month and corresponding
three cloud flag data in a month. Cloud flag data gives
one of three cloud status: clear, mixed, or cloudy.
1.5 Classification
Land cover classification was performed basically by
phenological information from monthly NDVI data.
In this study, a clustering, k-means method, was applied
for monthly composite NDVI data of 1990 independently
both for global land area and for Asian and Oceanian land
area. Since AVHRR data in Pathfinder Data Set in the
northern high latitude region over about 55 degree north
are not available in winter season because of low sun
elevation angle, a clustering was also applied for monthly
data from March to October. That is, the following three
clusterings were carried out.
(1) 12 monthly NDVI data from January to December,
1990 for global land area
(2) 12 monthly NDVI data from January to December,
1990 for Asia and Oceania
International Archives of Photogrammetry and Remote Sensing. Vol. XXXII, Part 7, Budapest, 1998 491
