Environmental Engineering Reference
In-Depth Information
the USGS EROS to produce the next generation Landsat data product. The Global
Land Survey Facility at the University of Maryland used LEDAPS approach and
produced global Landsat surface reflectance from Global Land Survey (GLS) data.
The extensive comparisons of surface reflectance between Landsat and MODIS show
general good agreements with overall discrepancies (root-mean-squared deviation
(RMSD)) between 1.0 and 2.5% reflectance for Landsat-7 ETM+ and between 1.6
and 3.2% reflectance for Landsat-5 TM (Feng et al.
2012
, under review). The Web-
enabled Landsat Data (WELD) project, a joint effort between South Dakota State
University and the USGS EROS, uses same atmosphere correction approach (6S) but
different ancillary data from MODIS atmosphere data products to generate 30-m
composites of Landsat mosaics at weekly, monthly, seasonal, and annual periods for
the conterminous United States (CONUS) and Alaska (Roy et al.
2010
;
www6
).
Surface reflectance is the basis for generating high-level biophysical products
such as Leaf Area Index (LAI). Using Landsat surface reflectance, NASA AMES
has prototyped Landsat LAI product by adopting MODIS LAI algorithm (Ganguly
et al.
2012
).
16.3 Data Fusion Approach
The data fusion solution integrates the spatial resolution of Landsat with the
temporal frequency of coarse-resolution MODIS sensor and thus produces fused
data products for applications that require high resolution in both time and space
(Gao et al.
2006
; Hansen et al.
2008
).
Traditional image fusion methods such as intensity-hue-saturation (IHS) trans-
formation, principal component substitution (PCS), and wavelet decomposition
focus on producing new multispectral images that combine high-resolution pan-
chromatic data with multispectral observations acquired simultaneously at coarser
resolution. They are useful for generating pan-sharpened images. However, they are
not effective in fusing spatial resolution and temporal coverage when input data
sources are acquired from different dates which may be affected by large
geolocation errors, high ratio of coarse-to-fine resolution, and land surface changes.
16.3.1 STARFM Approach
The Spatial and Temporal Adaptive Reflectance Fusion Model (STARFM) was
developed to combine high spatial Landsat and high temporal MODIS data (Gao
et al.
2006
). This approach requires input data pairs to be consistent. Observations
from different platforms first need to be calibrated and atmospherically corrected to
surface reflectance so that they are comparable spatially and temporally. Landsat
data are calibrated and atmospherically corrected using the LEDAPS approach.
Although MODIS and Landsat surface reflectance data are very consistent.
Search WWH ::
Custom Search