Environmental Engineering Reference
In-Depth Information
5.3 Results and Analysis
5.3.1 Validation of Spectral Mapping
Since AIRS and MODIS are on the same satellite, i.e., NASA Aqua, it provides a
unique opportunity to evaluate the accuracy of spectral mapping by comparing
AIRS-simulated MODIS measurements and spatially aggregated MODIS
measurements at AIRS field of views (FOVs). Spatially, the AIRS-simulated
AVHRR and MODIS measurements are for AIRS footprints, not for the native
AVHRR and MODIS pixels. To evaluate the accuracy of spectral simulation,
MODIS L1B measurements at 1-km pixels were aggregated to collocated AIRS
footprints through spatial matchup. AIRS has a field of view of
49.5
in cross
track. Each AIRS scan line contains 90 IR footprints, with a spatial resolution of
13.5 km at nadir and 41
21.4 km at the scan extremes from nominal 705.3-km
orbit. MODIS has a field of view of
55
in cross track. Each scan MODIS line has
1,354 footprints, with a spatial resolution of 1 km at nadir. AIRS has a 1.1
circular
IFOV; thus, as demonstrated in Fig.
5.4a
, for a given AIRS pixel, we can find the
MODIS pixels within the AIRS footprint by calculating the deviation of scan angle
between the AIRS pixel and the MODIS pixel; if the deviation is less than 0.55
,
then the MODIS pixel is inside the AIRS pixel. Figure
5.4b
illustrated an example
of collocated AIRS and MODIS pixels. Average MODIS 1-km data to colocated
AIRS pixels will produce “truth” MODIS data at AIRS FOVs and can be used to
compare with simulated MODIS.
Based on the global MODIS and AIRS L1 data of the selected 8 days, i.e.,
135*90*240*8
23,328,000 pixels, the difference between simulated MODIS
andaggregatedMODISradiance is quite small. For the 11-
¼
m band, the bright-
ness temperature difference between simulated MODIS and aggregated MODIS
is
μ
0.0039 K, with a standard deviation of 0.6031 K, as shown in Fig.
5.5a, b
.
The brightness temperature differenceislessthan1Kfor93.01%ofthetotal
23,328,000 pixels (Table
5.1
). And the related difference is less than 1%
for 99.15% of all the pixels (Table
5.2
). For the 12-
mband,thebrightness
temperature difference between simulated MODIS and aggregated MODIS is
μ
0.0535 K, with a standard deviation of 0.5461 K, as shown in Fig.
5.5c, d
.
The brightness temperature differenceislessthan1Kfor94.12%ofthetotal
23,328,000 pixels (Table
5.1
). And the related difference is less than 1% for
99.36% of all the pixels.
From these statistics of brightness temperature differences over global datasets
of the selected 8 days, the simulatedMODIS with AIRS hyper-spectral measurements
are quite close to the “true” MODIS measurements, i.e., aggregated MODIS
measurements at AIRS FOVs. So, it should be fine to use AIRS-simulated proxy
data to derive the relationships between different sensors.
Search WWH ::
Custom Search