onomy,
9, with
ER has
ASTER
han 1.5
1 of the
licated
veather
g track
ount of
ER
STER
cience
Every
ng the
/Terra
of all
ion (
S data
| to be
yortant
ümize
Team.
ster)
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B2. Istanbul 2004
One Scene=60km*60km
VNIR
Visible Near Infrared Radiometer
Wave Length: [F 3 Bands {Backward
@ @ Q9 052» 0.86 um
Spatia! Resolution ;F15 m
Pointing Angle; F — 1249
@@ @ @@o8t@kiDirectionr]
SWIR
Short Wave Infrared Radiometer
Wave Length _F 6 Bands
Q@ @ @ @1a6G- 2.43 pm
Spatial Resolution | F30 m
Pointing Angle; F x8.55°
QUQ Q '@@0@&t@kiDirection. j
TIR
Thermal Infrared Radiometer
Wave Length : .F5 Bands
@'@ Q 9825 -11.65 um
Spatial Resolution :.F90 m
Pointing Angle LF +8.55°
@r @r @"@iC@ss-truck Direction!>j
Fig. 2 Major Parameters of ASTER Sensors
NASA GSFC ( Goddard Space Flight Center ) under the form
of the * activity". NASA uploads the total schedule to Terra
under the form of command every day.
The data acquisition request is classified to ETR (Instrument
Team Request ), STAR ( Science Team Request ), and other
individual DAR, totally called xAR, and they are stored in the
xAR Data Base, which is referenced by the “Scheduler” when
ODS is generated. For the detail of the xAR, please refer to
Yamaguchi, et.al, 1995.
Resulting scene number of daily data acquisition is shown in
the figure 3, which shows that the data acquisition of ASTER is
very stable and can be approximated as 600 scenes per day in
most of the operation, but is slightly decreased to about 500
scenes per day, recently. The expected rate of data acquisition
derived from the duty cycle was 780 scenes per day in design
phase. But, considering the land area of the world, these
statistics seem to be reasonable. The reason of the slight
decrease of the data acquisition rate from 600 scenes per day to
500 is partly because of the damage in on board SSR ( Solid
State Recorder ), and also partly because the target area has
become scattered, resulting in the inefficiency of the data
acquisition.
One challenging operation flow of the ASTER mission
Acquired Scenes
> Cal. {every 33 days)
Periodical
À Calibration
2al (every 17 days
N
03.6.6
©
*
eo A
S o
02.6.6
e
I i | i
| MM A
E HE
ii Leonid ' i
i Show |
{ li Shower |
200 [ex i Leon” mi Deep C Án
maneuver Shower Terra/Bafehold Spaceldal ATC Anomaly
ASTER/Safe Mode > onal ie i Terra
100 E / T \ Safehol
SSR Anamaly | Icos Anomaly \ SSR Anomaly MÔDIS 1COS Anomaly | Ve &
0 X NJ Yaw Maneuvel d \
© © o © ae © ©
eo © = © e
- v
o S
0186 P^
01.106
01.126
02.26
02 106
12.126
01.46
o
0.
00.126 +
Fig3 Number of acquired ASTER scenes per day
operation is that the xAR is scheduled, referring to the
successful ( cloud-free ) data acquisition, i.c. if the acquired
data is cloud-free, ASTER will not try for this area, but if not,
ASTER will try again for this area. For this detail, please refer
to Miura, A. 2002.
Another challenging operation flow is to reflect weather
forecast information to make the ODS. By taking into account
weather forecast information, the cloud coverage is decreased
from the 40 - 45 % to 35 %. For this detail, please refer to the
same literature, Miura, A.,2002.
2.2 Data Processing of ASTER Data
As shown in the section 2.1, the data acquisition rate per day is
about 600 scenes per day. And the data volume of one Level 0
scene is about 100 MB. On the other hand, nominal life time of
ASTER is 5 year. Therefore, the estimated total scene number
is about 1.1 million, and the total volume for 5 years is about
110 TB. Since Level 1A product is produced
a gp nigga ;f
vri ra Eon T A my
qm m he 4 m um 7 — é i a
W790 2 3 0 30 4 iC HF DO BO UO YOGYy1T0 SOU 130 40 153 V6 1 ATR TAL TRQ LAC VE 1 361 20 10 VO 0 80 70 (ERSBACTOES |
* im Sal beta tuo vo vm 2 CRF SE mo ont etl UNS in
Fig.4 Coverage map of cloud free ASTER Data
for all the acquired data, estimated total number of LIA is 1.1
Million scenes and 110 TB. As for the Level 1B product,
corresponding estimated total number is about 275 thousand
scenes and 27.5TB, assuming the scenes to be requested to be
processed up to LIB is 25%! and one LIB product has 120
MB. Actual rule for the selection of the automatic LIB
generation is that LIB product will be generated if the cloud
cover of a scene is below 20 96, judged by the automatic cloud
evaluation algorithm. In addition, if there is a request to
generate the LIB scenes, L1B data will be generated regardless
with the cloudiness.
Since this data volume is very large, and for the easy access to
the archived data, automatic data storage system was
introduced. At the first stage, a robot system called "Powder
Horn" using 50 GB D3 cassettes was installed. But, by the
reasons described below, it was replaced to “Peta Site System”
using 200 GB DTF2 tape.
On the other hand, ASTER has been operated for more than 4
years, there is a statistics of the data acquisition, LIA/LIB
production. The cumulative sum of L1A and LIB data are 640
thousand ad 170 thousand scenes. The statistical values are
slightly smaller than the expected values for both LIA and
LIB. The major reasons for statistics for LIA are a) the tape
read errors b) bit flip. As for the tape read errors, relatively high
error rate had been experienced by the time of the media change
from D3 cassettes to DTF2 tapes ( Transition from D3 cassettes
to DTF2 were performed during 2002 October and 2004 April).
Change of the error rate was statistically evaluated. As for the
685
bit flip, a automatic correction algorithm has been applied if bit
flip occurs in a very critical portion of the data such as time tag.
This counter measure was applied in 2002 October. The LIA
processing error has been occurred less frequently. However,
bit flip could occur in the science data. In such case, processing
