nasa modis characterization support team (mcst) geometric ...€¦ · scan-to-scan underlap for...
Post on 27-Jul-2020
1 Views
Preview:
TRANSCRIPT
(Terra, Aqua) MODIS Geolocation Status
NASA MODIS Characterization Support Team (MCST) Geometric Calibration Group
Guoqing (Gary) Lin, SSAI/GSFC Code 619
Robert E. Wolfe, NASA/GSFC Code 619
James C. Tilton, NASA/GSFC Code 606
Ping Zhang, Bin Tan, SSAI/GSFC Code 619
John Dellomo, GST/GSFC Code 619
NASA MODIS-VIIRS Science Team MeetingCalibration Workshop
Silver Spring, MD10/18/2018
Outline
Lin et al., 18 Oct 2018 2
• Geolocation performance for MODIS on Terra and Aqua
– Overall performance
– Trends & details
• Added Info -- scan-to-scan underlaps & on-orbit focal length
• Future work
• Conclusions
Overall Geolocation performance
Residuals Terra C6 Aqua C6 Terra C6.1 Aqua C6.1 NPP VIIRS C1
Track mean 1m 3 m 0 m 1 m 12 m
Scan mean 0 m 1 m 0 m 0 m 4 m
Track RMSE 43 m 47 m 43 m 46 m 58 m
Scan RMSE 44 m 54 m 44 m 53 m 52 m
Data-days 6739 (18.4 yrs) 5917 (16.2 yrs) 6725 (18.4 yrs) 5910 (16.2 yrs) 2447 (6.7 yrs)
Missing days 61 10 59 10 1
Daily matched
GCPs w/ B1/I1
258 190 258 223 204
• Nadir equivalent accuracy (RMSE = Root Mean Square Error)
– Mostly within 20% band B1 HSI (250 m) = 50 m @ nadir (75 m for VIIRS I1);
– Within 10 % for HKM bands and 5% for KM bands
• Band-to-band mis-registration to other bands adds bias to RMSE :
• Other features for MODIS geolocation
– Aqua uses definitive ephemeris data 27 hour latency (Terra uses TDRSS-based on-board ephemeris)
– Aqua C6.1 corrected pointing variations (most of them) caused by AMSR_E stop -go slow – full stop activities
Lin et al., 18 Oct 2018
22 RMSE
3
Terra trend and update details
-75
-50
-25
0
25
50
75
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19
Tra
ck (
ad
j.)
res.
(m)
Years since Jan. 1, 2000
Daily 16-day Global 16-day Southern Hemisphere 16-day Northern Hemisphere
-75
-50
-25
0
25
50
75
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19
Scan
(ad
j.)
res.
(m)
Years since Jan. 1, 2000
Lin et al., 18 Oct 2018 5
Terra C6.1 long-term trend (uncorrected)
RMSE with no correction: Track: 49 m (+6 m vs C6) Scan: 48 m (+4 m vs C6)
-75
-50
-25
0
25
50
75
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19
Tra
ck (
ad
j.)
res.
(m)
Years since Jan. 1, 2000
Daily 16-day Global 16-day Southern Hemisphere 16-day Northern Hemisphere
-75
-50
-25
0
25
50
75
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19
Scan
(ad
j.)
res.
(m)
Years since Jan. 1, 2000
Lin et al., 18 Oct 2018 6
Actual Terra C6.1 residuals
C6.1 RMSE Track: 43 m Scan: 44 m, nadir equivalent
±20% 250m-band pixel size
-75
-50
-25
0
25
50
75
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19
Tra
ck (
ad
j.)
res.
(m)
Years since Jan. 1, 2000
Daily 16-day Global 16-day Southern Hemisphere 16-day Northern Hemisphere
-75
-50
-25
0
25
50
75
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19
Scan
(ad
j.)
res.
(m)
Years since Jan. 1, 2000
Lin et al., 18 Oct 2018 7
Actual Terra C6 residuals
C6 RMSE Track: 43 m Scan: 44 m, nadir equivalent
±20% 250m-band pixel size
Aqua trend and update details
-50
-25
0
25
50
75
100
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19
Tra
ck (
ad
j.)
res.
(m)
Years since Jan. 1, 2000
Daily 16-day Global 16-day Southern Hemisphere 16-day Northern Hemisphere
-75
-50
-25
0
25
50
75
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19
Scan
(ad
j.)
res.
(m)
Years since Jan. 1, 2000
Lin et al., 18 Oct 2018 9
Aqua C6.1 Long-term Trend (unorrected)
RMSE with no correction: Track: 71 m (+25 m vs C6.1) Scan: 55 m (+2 m vs C6.1)
AMSR-E, 2->0 rpm
2015-12-4T5:34zAMSR-E, 40->0 rpm
2011-10-4T7:27z
AMSR-E, 0->2 rpm
2013-12-5, 16:50 UTC
-75
-50
-25
0
25
50
75
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19
Tra
ck (
ad
j.)
res.
(m)
Years since Jan. 1, 2000
Daily 16-day Global 16-day Southern Hemisphere 16-day Northern Hemisphere
-75
-50
-25
0
25
50
75
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19
Scan
(ad
j.)
res.
(m)
.
Years since Jan. 1, 2000
Lin et al., 18 Oct 2018 10
Actual Aqua C6.1 residuals
C6.1 RMSE Track: 46 m, Scan: 53 m, nadir equivalent
±20% 250m-band pixel size
-75
-50
-25
0
25
50
75
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19
Tra
ck (
ad
j.)
res.
(m)
Years since Jan. 1, 2000
Daily 16-day Global 16-day Southern Hemisphere 16-day Northern Hemisphere
-75
-50
-25
0
25
50
75
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19
Scan
(ad
j.)
res.
(m)
Years since Jan. 1, 2000
Lin et al., 18 Oct 2018 11
Actual Aqua C6 residuals
C6 RMSE Track: 46 m, Scan: 54 m, nadir equivalent
AMSR-E, 2->0 rpm
2015-12-4AMSR-E, 40->0 rpm
2011-10-4±20% 250m-band pixel sizeAMSR-E, 0->2 rpm
2013-12-5
Leap-second handling
Lin et al., 18 Oct 2018 12https://en.wikipedia.org/wiki/Leap_second
year June30
Dec 31
Terra C6
Terra C6.1
Aqua C6
Aqua C6.1
2005 +1 good good good good
2008 +1 good good good good
2012 +1 good good Note 1 Note 2
2015 +1 good good good good
2016 +1 good good good good
Notes 1 & 2:
A error was recently found and corrected in the Aqua spacecraft ephemeris and attitude
dataset for the first 12 hours from 00z to 12 z on 7/1/2012 after leap-second insertion at the
start of the day (2012-06-30T23:59:60z). Data products have been replaced in Sept. 2018
Caution: geolocation errors may still be large after leap-second in the first 5-min granule:
2006-01-01T00:00-00:05
2009-01-01T00:00-00:05
2012-07-01T00:00-00:05
2015-07-01T00:00-00:05
2017-01-01T00:00-00:05
We may “hide” the granules
upon further examination
Scan-to-scan underlap
Lin et al., 18 Oct 2018 13
where
n = # detectors,
p = detector “pitch” interval in the track direction,
F = effective focal length (EFL),
T = scan period, tied to F by BBR requirements,
i = inclination angle, > 90o for MODIS and VIIRS,
VECI = spacecraft ground speed in the inertial frame,
h = range from instrument to earth terrain surface
Vearth0 = surface speed of earth rotation at equator,
Overlap < 0 underlap.
𝑂𝑣𝑒𝑟𝑙𝑎𝑝 = 𝑛𝑝
𝐹𝒉 − [𝑽𝑬𝑪𝑰 − 𝑉𝑒𝑎𝑟𝑡ℎ0 𝐜𝐨𝐬 𝒊 ]𝑇
This term is added for
JPSS-3&4 VIIRS
Instrument properties
Earth property
Satellite properties
Scan-to-scan underlap for Terra MODIS
Lin et al., 18 Oct 2018 14
-90
-75
-60
-45
-30
-15
0
15
30
45
60
75
90
600
700
800
900
1000
1100
1200
39830 44830 49830 54830 59830 64830 69830 74830 79830
Lati
tud
e (
de
g)
Alo
ng
Tra
ck G
SD
(m
)
Scan line #
MODIS Terra 2016-06-30 (one orbit)
Within Scan
From Odd scan to Even Scan
From Even scan to Odd scan
Naidr Latitude
underlap
overlap
• Terra MODIS has underlap of ~ 90 m around 15oN in all neighboring scans
• High terrain worsens the underlap where it occurs
• The altitude variations, minimal near equator and higher near poles, can be seen from the variations of within scan ground sampling distance (GSD)
Terrain effects
Terrain effects
Scan-to-scan underlap for Aqua MODIS
• Aqua MODIS has more serious underlap in alternating scans around 15oN
• Orbit asymmetry due to drifts makes underlap more over descending side (~ 170m) than the ascending side (~ 150 m) in this case, may reverse in another caseLin et al., 18 Oct 2018 15
-90
-75
-60
-45
-30
-15
0
15
30
45
60
75
90
600
700
800
900
1000
1100
1200
531 5531 10531 15531 20531 25531 30531 35531 40531
Lati
tud
e (d
eg)
Alo
ng
Tra
ck G
SD (
m)
Scan line #
MODIS aqua 2016-06-30 (one orbit)
Within Scan
From Odd scan to Even Scan
From Even scan to Odd scan
Latitude
overlap
underlap underlap
Terrain effects
Terrain effects
Aqua MODIS underlap vs scan sample
• Maximum underlap, if exists, exists at nadir and around 15oN
• Due to off-nadir bow-tie effects, scan-to-scan underlap closes off-nadir ~ 122 km for the large one and ~ 60 km for the smaller one in this case.
Lin et al., 18 Oct 2018 16
900
950
1000
1050
1100
1150
1200
-150 -100 -50 0 50 100 150
Gro
un
d S
am
ple
Dis
tan
ce (m
)
Sample # off-nadir
Aqua MODIS underlap off-nadir
Within scan
From odd scan to even scan
From even scan to odd scan
overlap
underlapoverlap
Underlap, max 170 m
-1
-0.5
0
0.5
1
B4
B4
B4
B4
B4
B4
B12
B12
B12
B12
B12
B2
B2
B2
B2
B2
B2
B7
B7
B7
B7
B7
B7
B32
B32
B32
B32
B32
B32
EFL
dev
iati
on
(% N
om
ina
l)
Band Name
EFL on-orbit measurements for Terra MODIS
-1
-0.5
0
0.5
1
B4
B4
B4
B4
B4
B4
B4
B4
B12
B12
B12
B12
B12
B12
B12
B12
B2
B2
B2
B2
B2
B2
B2
B2
B7
B7
B7
B7
B7
B7
B7
B7
B32
B32
B32
B32
B32
B32
B32
B32
EFL
dev
iati
on
(% N
om
ina
l)
Band Name
EFL on-orbit measurements for Aqua MODIS
On-orbit focal length measurements
• On-orbit measured effective focal length (EFL) varies among builds and bands.
• A +0.1% EFL change means +10 m change in scan-to-scan underlap where it exits.– Band B7 has less underlaps for both Aqua and Terra MODIS
– Band B12 has more underlaps for Terra MODIS
Lin et al., 18 Oct 2018 17
Vis FPA NIR S/MWIR LWIR
LWIR
NIR S/MWIR
Vis FPA
2002 2018
2002 2018
Future work (C7.0 soon?)
Lin et al., 18 Oct 2018 18
1) Routine monitor and LUTs update as needed
2) Refresh ground control point chip library
1) Chips are extended from 24x24 to 36x36 km
2) Error measurements are extended from ± 45o to ± 55o)
3) Automate GEO LUT updates
4) Update DEM, LWM (year by year?)
5) Add 250m offsets(?) to geolocation files (currently 1000m with 500m offsets)
6) Create Level-1 geolocation web (needed?)
1) similar to L1B
Anything Else?
Any change in priority order above
Concluding Remarks
Lin et al., 18 Oct 2018 19
• Geolocation performance for MODIS on Terra and Aqua is
good
– mean errors for band B1 near 0 and uncertainties are ~ 50 m at nadir for
the missions, statistically
– C6.1 corrected artifacts in C6 caused by LUT updates
– AMSR_E stop-go-stop activities on Aqua induced MODIS geolocation
errors, which were corrected in C6.1
– Scan-to-scan underlaps exist around nadir in the equator regions
• Maximal underlaps are at nadir near 15oN, at ~ 150 m for Aqua MODIS B2
Backup slides: sun angle dependent Residuals
Lin et al., 18 Oct 2018 20
y = -0.0003x2 - 0.0076x + 4.4306R² = 0.2968
y = -0.0025x2 + 0.5511x - 22.682R² = 0.7863, C5 used for C6, C6.1
-100
-75
-50
-25
0
25
50
75
100
-20 0 20 40 60 80 100 120 140 160 180
Sh
ift (m
ete
rs)
Sun angle (degrees)
Track
y = 0.0002x2 - 0.0574x + 3.4522R² = 0.0981
y = -0.0031x2 + 0.7185x - 34.362R² = 0.8647, C5 used for C6, C6.1
-100
-75
-50
-25
0
25
50
75
100
-20 0 20 40 60 80 100 120 140 160 180
Sh
ift (m
ete
rs)
Sun angle (degrees)
Scan
y = -0.0003x2 - 0.0116x + 3.12R² = 0.3189
y = -0.0025x2 + 0.5511x - 22.682R² = 0.7863, C5 used for C6, C6.1
-100
-75
-50
-25
0
25
50
75
100
-20 0 20 40 60 80 100 120 140 160 180
Sh
ift (m
ete
rs)
Sun angle (degrees)
Track
y = 0.0001x2 - 0.0561x + 3.7938R² = 0.1107
y = -0.0031x2 + 0.7185x - 34.362R² = 0.8647, C5 used for C6, C6.1
-100
-75
-50
-25
0
25
50
75
100
-20 0 20 40 60 80 100 120 140 160 180
Sh
ift (m
ete
rs)
Sun angle (degrees)
Scan
Lin et al., 18 Oct 2018 21
Terra Sun angle CorrectionC6.1 resultsC6 results
y = -0.0021x2 + 0.3683x - 11.3785R² = 0.5607
y = -0.0043x2 + 0.6999x - 19.695R² = 0.7678, C5 used for C6, C6.1
-100
-75
-50
-25
0
25
50
75
100
0 20 40 60 80 100 120 140 160 180
Sh
ift (m
ete
rs)
Sun angle (degrees)
Scan
y = 0.0009x2 - 0.0951x + 2.22R² = 0.4793
y = 0.0026x2 - 0.3546x + 4.3033R² = 0.7477, C5 used for C6, C6.1
-100
-75
-50
-25
0
25
50
75
100
0 20 40 60 80 100 120 140 160 180
Sh
ift (m
ete
rs)
Sun angle (degrees)
Track y = 0.0008x2 - 0.0841x + 0.0915R² = 0.4656
y = 0.0026x2 - 0.3546x + 4.3033R² = 0.7477, C5 used for C6, C6.1
-100
-75
-50
-25
0
25
50
75
100
0 20 40 60 80 100 120 140 160 180
Sh
ift (m
ete
rs)
Sun angle (degrees)
Track
y = -0.0020x2 + 0.3559x - 12.2284R² = 0.5040
y = -0.0043x2 + 0.6999x - 19.695R² = 0.7678, C5 used for C6, C6.1
-100
-75
-50
-25
0
25
50
75
100
0 20 40 60 80 100 120 140 160 180
Sh
ift (m
ete
rs)
Sun angle (degrees)
Scan
Lin et al., 18 Oct 2018 22
Aqua Sun angle CorrectionC6 results C6.1 results
Root cause – example by VIIRS
• Earth rotation casts a component of speed in the VIIRS track & satellite inclination direction
• A sensor Field of regard (FOR) in the track direction FOR = n ASI h needs to cover all the
time everywhere the scan-to-scan travel distance, i.e. FOR >= Ds2s=T[VECI -Vearth0cos(i)] . Otherwise, scan-to-scan underlap occurs
• For details, see Lin et al (2016),“Trending of SNPP ephemeris and its implications on VIIRS geometric performance”, Proc. of SPIE, Vol. 9972, 99721K, doi: 10.1117/12.2239043.
Lin et al., 18 Oct 2018 23
i= 98.67o
VECI = 6,584 m/s
- Vearth (in ECI) = 464 m/s
VECR = 6,670 m/s
SSP1(to)
SSP2
(to +1s)
Equator
VRTA = 3,000 to 18,000 km/s, perpendicular to VECI.16/32 detectors registeron the ground in parallel with VECI
simultaneously
E
- Ve_track = 70 m/s
- Vearth = 464 m/s
top related