Initial Analysis of GOES-13 Imager Solar Contamination During Eclipse
X. Wu; Drafted Nov. 20, 2006; Revised Dec. 13, 2006
Introduction
The newly launched Geostationary Operational Environmental Satellite (GOES13) is the first in a series (GOES-N/O/P) that is equipped with improved battery to enable
normal operation throughout the semiannual eclipse that occur around the solstices.
During the Post Launch Tests (PLT), however, it was recognized that measurements
should not be taken too close to the Sun. For the Imager, proximity of the filed of view
(FOV) to the Sun is often associated with serious stray light contamination that
compromises the value of the data for remote sensing applications.
Figure 1: Visible (top panels) and near infrared (3.9 μm, bottom panels) channel images from GOES13 Imager around 0600 UTC (satellite midnight) on August 31, 2006. Horizontal lines are
caused by noise in data reception not related to sensor performance.
The problem was first identified around 0600 UTC on August 31, 2006, when
GOES-13 was stationed above the equator at 90ºW. The time coincides with the local
midnight at the satellite nadir point. On this first day of the autumnal eclipse season for
GOES, the Sun was at the top of the GOES-13 field of regard (FOR) before and after
0600 UTC, tangent to the Earth at the North Pole at 0600 UTC. (The Sun’s position
around 0600 UTC would progressively move to the South Pole in the next 48 days.)
Figure 1 shows that the visible channel detected strong signals where they are not
expected, in the space views and Earth views that are fairly far away from the Sun. The
infrared (IR) channels (only the 3.9 μm channel images were shown in Fig. 1) indicate
contaminated data in those regions.
These contaminated data, while not a threat to the instrument safety and health,
are useless nevertheless and represent a waste of valuable resources for observing other
parts of the world. It was therefore decided to establish a “Keep Out Zone” (KOZ) such
that the Imagers on GOES-N/O/P series will not acquire contaminated data once
commissioned. This analysis summarizes the initial attempt to characterize the GOES-13
Imager KOZ. The objective is to answer the following questions:
1.
2.
3.
4.
5.
Which channels are affected? By how much?
How to quantify the KOZ?
What is the KOZ in terms of time and space?
Does the KOZ depend on channel?
Does the KOZ depend on the position of the Sun in the GOES FOR?
Sounder
Compared to the Imager, Sounder has longer dwell time for each pixel in normal
operation, which caused part of the instrument temperature exceeding the safe limit on
August 31, 2006 when exposed to direct sunlight. Sounder has since been excluded from
KOZ experiment. Occasionally, Sounder data quality was compromised when the Sun is
nearby, as suggested in Fig. 2. In general, it is expected that the KOZ derived from the
Imager data analysis is also applicable to the Sounder, although that has not been verified
and need to be carefully monitored.
Band=15
(4.4um)
06:00 UTC
Figure 2: Sounder data contamination at 0600 UTC on October 6, 2006, when the Sun came from
behind the Earth.
Data
In collaboration with the GOES-13 PLT team, Space Science and Engineering
Center at the University of Wisconsin (UW/SSEC), and Office of Satellite Data
Processing and Distribution (OSDPD), three special sectors were designed for GOES-13
Imager during the KOZ experiment from September 12 to October 17, 2006. These
sectors are full width in the East-West direction but approximately one third of the full
disk in the North-South direction (from slightly above the North Pole to approximately
13ºN, between approximately 15ºN and 15ºS, and from approximately 13ºS to slightly
below the South Pole). During the KOZ experiment, GOES-13 Imager was commanded
to scan one of these sectors every 15 minutes from 0455 UTC to 0655 UTC. Two of these
images, at 0555 UTC and 0655 UTC, were away from the Sun; the purpose was to
confirm the absence of contamination in the “safe zone”. The rest of the images were
following the position of the Sun on the day; the purpose was to monitor the evolution of
the KOZ.
Figure 3 is an example of the solar contamination in one of the three sectors
around 0510 UTC on Sept. 12, 2006. In the visible channel image (left), white indicates
bright or contamination since no light is expected except at the left edge of the GOES
FOR. In the IR image (right), black indicates hot, so the feature resembling a quarter of a
ring to the left of the image is caused by contamination.
Figure 3: GOES-13 Imager Channel 1 (visible, left panel) and Channel 2 (3.9 μm, right panel) images
at 0510 UTC on Sept. 12, 2006, the first images collected for GOES-13 Imager KOZ study.
Analysis
KOZ Identification
The KOZ was identified through comparison with a reference image. For the
visible channel, since no reflected sunlight is expected at the time of observation, all
signals above noise level are due to contamination. For the infrared channels, KOZ
causes large and consistent difference between the contaminated and reference images
that cannot be explained by change in scene temperature between the observing times, as
shown in Fig. 4. This analysis was applied to all images. It was found that images taken
around the satellite midnight but are sufficiently far away from the Sun (in the sector
from approximately 13ºS to slightly below the South Pole, scanned at 0555 UTC and
0655 UTC) are not subject to the stray light contamination.
Figure 4: Brightness temperature difference between 0510 UTC (contaminated) and 0525 UTC (not
contaminated) on Sept. 12, 2006 for the four IR channels of GOES-13 Imager (the Channel 2
image in the upper left panel is that of Tb instead of δTb). The impact is severe on Channel 2
(3.9 μm), up to 75°K, and progressively smaller for longer wavelength but still traceable
even for Channel 6 (13.3 μm). (T. Schmit)
KOZ Features
As can be seen from previous figures, the solar contamination is not uniformly
decreasing away from the Sun. Rather, there seems to be a “ring feature” when the Sun is
to the west, and this feature seems similar for both channels. Other features also exist, for
example the “rays from the center” feature when the Sun is to the east. Examination of all
contaminated images collected during the KOZ experiment suggests that these features
are persistent from day to day. Some of these features are collected in Fig. 5. It seems that
the shape of contamination, especially for Channel 2, is similar to that of Imager’s
Secondary Mirror, with some variations at different time and day. It also seems that the
contaminated area in the visible channel, though lacking many fine structures, is
otherwise similar to that in the IR channels.
Figure 5: Selected Channel 2 images to show features in the contaminated data. Upper right: 0625
UTC on September 12, 2006. Upper left: 0510 UTC on October 12, 2006. Lower right: 0610
UTC on October 12, 2006. Lower left: 0525 UTC on October 17, 2006.
KOZ Quantification
If the Imager records solar radiance when it looks away from (though close to) the
Sun, it is contamination. When the Imager looks sufficiently further away from the Sun,
the contamination disappears. The KOZ can therefore be quantified by the distance, in
terms of the angle between the Sun and the pixel when viewed from the satellite, that
separate the contaminated and un-contaminated pixels.
For an image, the mean sensor response can be computed and plotted as a
function of the angular distance. Figure 6 is such a plot for the 0510 UTC image on
September 12, 2006, by GOES-13 Imager visible channel. There is a general trend of
decreasing contamination intensity away from the center of the Sun, with a secondary
peak around 4° that corresponding to the ring feature in Fig. 3, followed by a sharp
decline after 5°. The mean signal is never reduced to zero, most likely because of the few
bright lines in Fig. 2 (some invisible in the small image) that were caused by ground
receiving.
Figure 6: Mean sensor response (raw count – space count) for GOES-13 Imager visible channel at
0510 UTC on Sept. 12, 2006, plotted as a function of angular distance between the Sun and a
pixel with respect to the Imager. 0° means the Imager is looking toward the center of the
Sun.
It might be reasonable to state that the KOZ is 5° or 6° based on Fig. 6, however it
would be premature to characterize the KOZ that way for all images. Figure 7 shows that
the KOZ can be quite different, depending on the position of the Sun. Similar plot should
be analyzed for other UTC times, especially when the Sun is at the other side of the
Earth, to gain a more comprehensive view. Current results suggest that the KOZ is
normally around 5° but can be up to 10° when the Sun is near the pole.
Figure 7: As Fig. 6, but for all 0510 UTC images during the investigation. Color is used to indicate
the date.
Conclusions:
1. Stray light from the Sun during eclipse can affect all channels of the Imager. The
impact is large (several hundred of counts) for Channel 1 & 2, smaller for IR
channels with longer wavelength, but still noticeable even for Channel 6.
2. Images obtained around the satellite midnight but sufficiently far away from the
Sun are not affected.
3. A Keep Out Zone (KOZ) is recommended such that the Imager should not scan
closer to the Sun than the KOZ to avoid solar contamination.
4. The KOZ does not seem to depend on channel.
5. The KOZ seems to be smaller (~5°) when the Sun is near the equator and larger
(up to 10°) when the Sun is near the poles.
6. Recommend to repeat the data collection and analysis during spring equinox
a. Confirm the effect of yaw-flip (or the lack thereof)
b. Monitor the KOZ before/after the Sun approaches the poles