ND
nce Vegetation
conducted over
pace borne IRS
ments of field-
odels based on
vo sites (Indore
yr validation of
SS-III data and
0.33 for Indore
| assigning land
ped to generate
ling LAUFPAR
m EROS Data
-DAAC) for
)D15 LAI and
nd 8-day basis.
assessment of
s by analytical
ed to represent
ry and has been
(Privette ef al.,
e over India. A
VE) aiming at
cific vegetation
Al product was
wheat-growing
omponents, (1)
eric properties
neration of fine
and field data;
1eir comparison
DS
dia, Indore and
1ese sites repre-
/. Both the sites
major crop with
pea. Indore has
sricultural prac-
ata used in the
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B7. Istanbul 2004
Table 1: LAI sites and details of satellite data acquisitions
Description Site: Indore Site: Bhopal
Site location 75°57 E, 27 28 L.
Long., lat. 22153 N 23° 10° N
Date of IRS- 02 Dec 01 24 Dec 01
1D LISS-III* 27 Dec 01 18 Jan 02 *
acquisitions 21 Jan 02 * 12 Feb 02
IRS-1D 96/56 97/55
Path/Row
Date of 3Dec.-10Dec. 2001 19Dec.-26Dec. 2001
MODIS LAI 27Dec.2001-3Jan.02
8-day product - 10Feb.-17Feb. 2002
MODIS Tile 24H 6V 25H 6V
Number
/
* LISS-ITI has green, red, nir bands (spatial resolution: 23 m),
swir band (spatial resolution: 70 m)
# Overcast conditions, hence not used in analysis
2.2 Satellite data used
2.» IRS LISS-III
The data of Linear Imaging Self Scanning sensor-IIl. (LISS-IIT)
onboard Indian Remote Sensing satellite IRS-1D was used in
the study. The LISS-III sensor has multispectral bands in green,
red, near infrared (NIR) with 23 meter and short wave infrared
(SWIR) with 70 meter spatial resolution.
2.4 The MODIS LAI product/algorithm
The MODIS LAI/FPAR product is produced at 1 km spatial
resolution daily (MODI15A1) and composited over an 8-day
period, where the selected value in a compositing period is that
with the highest corresponding fraction of . absorbed
photosynthetically active radiation (FPAR). The 8-day product
(MODI5A2) is distributed from the EROS Data Center. The
products are projected on the Integerized Sinusoidal (IS) 100
grid, where the globe is tiled into 36 tiles along the east-west
axis, and 18 tiles along the north-south axis, each
approximately 1200X1200 km.
A brief summary of LAI algorithm is provided by Myneni ef al.,
(2002). The algorithm is based on rigorous three-dimensional
radiative transfer (RT) theory (Myneni ef al., 1990). A look-up
table (LUT) method is used to achieve inversion of the three-
dimensional RT problem. The 250 and/or 500 m resolution
bands are aggregated into normalized 1 km resolution grid cells
prior to ingest (Wolfe ef al., 1998).
The algorithm also employs a 1 km land cover map stratified by
six major world biomes (grasses/cereal crops, shrubs, broadleaf
crops, savannas, broadleaf forests and needleleaf forests). Look-
up tables are then generated for each biome by running the
model for various combinations of LAI and fractional cover.
During algorithm execution, the algorithm compares the
modeled and observed reflectances for a suite of canopy
structures and soil patterns that represent the range of expected
natural conditions. All canopy/soil patterns for which modeled
and the observed reflectances are considered acceptable
solutions. A scale-independent test of energy conservation is
also applied. The mean LAI for this solution set is reported as
the MODIS LAI product values. When this method fails to
provide a solution, a backup algorithm based on relations
145
between the vegetation difference vegetation index (NDVI) and .
LAI (Knyazikhin ef al., 1998) is employed along with a biome
classification map.
2.5 Experimental measurements of LAI and atmespheric
parameters
The objective was to make LAI measurements and to generate
site-specific LAI map. Therefore a suitable site of 30 km X 30
km representing the region was focused at two locations. The
sites had adequate variability in terms of sowing date and
variability in LAI. Optical methods were used in this study to
acquire a large number of data points. LAI-2000 Plant Canopy
Analyzer (LI-COR Inc.) was used to measure LAI in the fields.
It is based on “fisheye” measurement of diffuse radiation
interception by measuring gap fraction. The LAI-2000 measures
attenuation of diffuse sky radiation at five zenith angles
simultaneously (approximately 0-139, 16-28^, 32-435; 47,53",
61-74"). The measured gap fraction data are inverted to obtain
the effective LAI under the assumption of a random spatial
distribution of leaves. All the measurements were taken by
holding the sensors opposite to the direction of the sun. A 90?
mask was used to prevent interference caused by the operator's
presence. A 270" mask was used in some fields of Bhopal
because of heterogeneous distribution of trees around the fields.
The LAI measurements were collected at six to eight locations
within a field (with each observation being based on six point
measurements) in order to obtain representative field LAI
values. A total of 75 fields were sampled at various growth
stages (monthly once, for three months at two sites) of the
crops. The LAI measurements were carried out mainly for
wheat crop with few observations on gram and pea. The
locations of fields were marked on FCC paper prints and also
determined with Global Positioning System (GPS).
The atmospheric measurements of aerosol optical thickness
(AOT) and water vapor content were carried out concurrently
with the LAI measurements on the date of IRS-1D satellite
acquisitions using handheld Microtops-II Sunphotometer with
five optical collimators working at 500, 675, 870, 936 and 1020
nm and a full field of view of 2.5".
2.6 LAI map validation procedure
The ground plots, in which LAI was measured, were generally
25-75m in size. Because of the surface heterogeneity (cover
type and density changes), it was necessary to use fine-
resolution images, in which ground plots can be located
accurately to validate low-resolution products. The procedures
for LAI map validation were:
1) Selection of representative areas in the Central India
regions and identification .of IRS LISS-III scenes
covering these areas;
2) Collection of LAI data in multiple (30-40) plots
within each LISS-III scene using the same types of
instruments and following the same measurement
protocols;
3) Identification of ground plots in the scenes and
extraction of the remote sensing data for each of the
plots;
4) Development of non-linear LAI-NDVI model for
different sites using satellite and field data.
5) Generation of LAI maps for each site using the model
developed.
