The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part Bl. Beijing 2008
951
2000 and 2003.
2) Simulated datasets by computer simulation model
Because we don’t have the experiment measured data for
grass and forest, in this study, we use computer simulation
model-Radiosity model, which can simulate the 3D realistic
scenes of vegetation (such as crops and forest) in the whole
growing period, and get their radiation regime(Song Jinling
2007(a); Song Jinling 2007(b)). We can product the look up
table of the vegetation structure parameters, sun direction, view
geometry and the corresponding simulated spectrum. And then
the datasets of grass and forest are built by using the simulated
data.
2.1.2 Satellite Data Used in This Study
1) Beijing-1 Microsatellite Multispectral images:
Beijing-l(BJ-l) microsatellite is an applied earth observing
microsatellite of China. It combines SSTL’s standard Disaster
Monitoring Constellation (DMC) multi-spectral camera with a
high resolution panchromatic imager, which can achieve the
32m spatial resolution multi-spectral images and the 4m
resolution panchromatic images synchronously. Beijing-1
images have the same characteristics and capabilities as TM
image at 2> 3^ 4 bands. In addition, BJ-1 remote sending data
can also give us the good data of short cycle time and wider
coverage. So it is necessary to generate the quantitative product
of BJ-1 remote sensing data.
2) MODIS LAI product: The eight-day composites of the
LAI/fAPAR products(MOD15A2) version 4 pertaining to study
sites were downloaded from the Land Processes-Distributed
Achieve Center(LP-DAAC) Internet site. The product was
composited over an eight-day period, where the selected value
in a compositing period is that with the highest corresponding
Fapar. The products are projected on the integerized sinusoidal
and sinusoidal lOo grid, respectively.
3) MODIS land cover product: the MODIS land cover product
(MOD12Q1) version 4 was used (Table 1.) to investigate the
biome assigned to the region of the study sites. The latest
available MOD12Q1 product for our study period was 2004
and the product was used with the assumption that the biome
distribution does not change with a year.
2,2 METHOD
2.2.1 Beijing-1 LAI Map Generation
In order to get the LAI distribution map at high spatial and high
temporal resolution of Beijing-1 image, works were done
shown in the flow chart (fig. 1) :l)the spatial matchment of the
Beijing image and MODIS products; 2) Generation of
Beijing-1 LAI maps; 3) MODIS Products Processing; 4) The
generation of time serials LAI of the whole year at Beijing-1
spatial resolution.
1) The image registration of Beijing image and MODIS
products
Because MODIS products project is SIN project, which is
different form BJ-l’s projection, in order to match with BJ-1
image, MOIS products (MOD15A2 and MOD12Q1) are
reprojected to the same projection with BJ-1 image, UTM
projection. In this paper, we use the HDF-EOS and GCTP
library in the actual dataset, which can provide coordinate
transformation function, so can complete the conversion
between the different projections easily.
Based on the coverage of the scene of Beijing-1 image
containing Beijing experiment sites, we select 4 tiles MODIS
products (h26v04, h26v05, h27v04 and h27v05), accomplish
the images Mosaic and then reproject them to UTM projection.
Fig.2 is the reprojection image of MOD12Q1 product.
According to MODIS PFT classification standard, the map is
classified to the following classes: evergreen needleleaf trees,
evergreen broadleaf trees, deciduous needleleaf trees,
deciduous broadleaf trees, Shrub, Grass, Cereal crop, Broadleaf
crop, Urban and built up and water. In this study, evergreen
needleleaf trees, evergreen broadleaf trees, deciduous
needleleaf trees and deciduous broadleaf trees are classified to
one class, tree in the BJ-1 classification map(shown in fig.3), so
in MOD12Q1 image, trees class represents evergreen
needleleaf trees, evergreen broadleaf trees, deciduous
needleleaf trees and deciduous broadleaf trees. Fig.3 is the
Beijing-1 classification map, which is classified to six classes:
broadleaf crop, cereal crop, forest, grass, urban and water.
Fig. IThe flow chart of fusing MODIS and Beijing-1 image