MODIS Geolocation Status · accuracy goal (50 m) •Large errors occur ~1.5 hr after maneuvers...

(Terra, Aqua) MODIS Geolocation Status

MODIS-VIIRS Science Team Meeting Calibration Workshop

April 15, 2013 5/1/2014 5/18/20156/6/2016

Guoqing (Gary) Lin,

Robert Wolfe, James Kuyper, Zhangshi Yin, and Bin Tan

Geometric Calibration Group

NASA GSFC Code 619

Geolocation C5/C6 results

• Excellent results –Root Mean

Square Error (RMSE) in nadir

equivalent units is better than

accuracy goal (50 m)

• Large errors occur ~1.5 hr after

maneuvers (about 12 per year)

- accuracy in following orbit

suspect

• Good results – RMSE is better

than goal (50 m) in track direction

but slightly over goal in scan

direction (but much better than

specification – 150 m)

• Definitive ephemeris is used for

best results – causes up to 27 hr

processing delay

Terra Aqua

C5 C6 C5 C6

Along-track RMSE (m) 44 43 47 46

Along-scan RMSE (m) 44 44 53 53

Years of Data 15.8 16.0 13.6 13.8

Ground Control Point Match-ups/day

Days missing (no residuals)

260 218

134 62

224 189

Terra Aqua

Results up to April 30, 2016

Note: These results are for MODIS Band 1, which is used in the control point matchup. Other bands must be offset by the band-offsets published by the MODIS calibration team. Lin et al., 6 June 2016

Terra trend and update details

Lin et al., 6 June 2016

Lin et al., 6 June 2016 4

Terra C6 long-term trend (uncorrected)

RMSE with no correction: Track: 49 m (+6 m vs C6) Scan: 48 m (+4 m vs C6)

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

Years since Jan. 1, 2000

Daily 16-day Global 16-day Southern Hemisphere 16-day Northern Hemisphere

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

Actual Terra C5 residuals

C5 RMSE Track: 44 m Scan: 44 m, nadir equivalent

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

Actual Terra C6 residuals

C6 RMSE Track: 43 m Scan: 44 m, nadir equivalent

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

Terra Safe Hold

• Pointing/geolocation look ok after re-activation Lin et al., 6 June 2016 7

16.00 16.08 16.17 16.25 16.33

16.0 16.1 16.2 16.3 16.3

Scan r

Aqua trend and update details

Aqua C6 Long-term Trend (unorrected)

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

RMSE with no correction: Track: 68 m (+22 m vs C6) Scan: 54 m (+1 m vs C6)

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

Actual Aqua C5 residuals

C5 RMSE Track: 47 m, Scan: 53 m, nadir equivalent

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

Actual Aqua C6 residuals

C6 RMSE Track: 46 m, Scan: 53 m, nadir equivalent

next slide

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

11 11.5 12 12.5 13 13.5 14 14.5 15 15.5 16 16.5 17

Daily 16-day GlobalEffects of AMSR-E (in)activities

Lin et al., 6 June 2016 12 • LUT update soon (shown here are plots from C6)

AMSR-E, 0->2 rpm

2013-12-5, 16:50 UTC

AMSR-E, 2->0 rpm

2015-12-4, 5:34 UTC

AMSR-E, 40->0 rpm

2011-10-4, 7:27 UTC

4019 4384 4750 5115 5480 5845 6211

Day (c. 2000)

c6 actual pitch (track)

c6 actual roll (scan)

c6 actual yaw

LUT 6.0.25

2016-02-02 00z

Future work

1) Routine monitor and LUTs update as needed

2)Update to correct AMSR-E effects in Aqua geolocation from late 2011

3)C7 update?

4)Update of Landsat ground control point chip library

Concluding Remarks

Residuals Terra C5 Aqua C5 Terra C6 Aqua C6 VIIRS C1.1

Track mean -1 m 7 m 2 m 2 m 8 m

Scan mean 5 m -3 m -1 m 0 m 5 m

Track RMSE 44 m 47 m 43 m 46 m 72 m

Scan RMSE 44 m 53 m 44 m 53 m 61 m

Data-days 5777 (15.8 yrs) 4951 (13.6 yrs) 5849 (16.0 yrs) 5040 (13.8 yrs) 1566 (4.3 yrs)

Missing days 134 99 62 10 1

Daily matched

GCPs w/ B1/I1

260 224 218 189 130

• Nadir equivalent accuracy (RMSE – Root Mean Square Error)

– Mostly within 20% band B1 HSI (250 m) = 50 m @ nadir ;

– Within 10 % for HKM bands and 5% for KM bands

• Band-to-band mis-registration adds bias to RMSE to other bands:

• Other features for MODIS geolocation

– Aqua uses definitive ephemeris data 27 hour latency (Terra uses TDRSS-based on-board ephemeris)

– C6 uses 0.5 km resolution DEM for terrain corrections (C5 uses older 1 km DEM, same as the VIIRS), plus other improvements over C5

22 RMSE

Backup slides: sun angle dependent Residuals

y = 0.0025x2 - 0.5159x + 20.521R² = 0.6431

y = -0.0074x2 + 1.2704x - 44.415R² = 0.8514, C4 used for C5

-20 0 20 40 60 80 100 120 140 160 180

Sun angle (degrees)

y = 0.0017x2 - 0.3666x + 21.435R² = 0.5203

y = -0.0068x2 + 1.3044x - 49.118R² = 0.8886, C4 used for C5

-20 0 20 40 60 80 100 120 140 160 180

Sun angle (degrees)

y = -9E-05x2 - 0.0406x + 5.9642R² = 0.27

y = -0.0025x2 + 0.5511x - 22.682R² = 0.7863, C5 used for C6

-20 0 20 40 60 80 100 120 140 160 180

Sun angle (degrees)

Terra Sun angle Correction C6 results C5 results

y = 0.0002x2 - 0.0579x + 3.1975R² = 0.0876

y = -0.0031x2 + 0.7185x - 34.362R² = 0.8647, C5 used for C6

-20 0 20 40 60 80 100 120 140 160 180

Sun angle (degrees)

Small overcorrection in C5 – was corrected in C6

y = 0.0011x2 - 0.1535x + 4.7188R² = 0.4607

y = 0.0026x2 - 0.3546x + 4.3033R² = 0.7477, C5 used for C6

0 20 40 60 80 100 120 140 160 180

ift (m

Sun angle (degrees)

y = -0.0022x2 + 0.3632x - 11.523R² = 0.5996

y = -0.0043x2 + 0.6999x - 19.695R² = 0.7678, C5 used for C6

0 20 40 60 80 100 120 140 160 180

ift (m

Sun angle (degrees)

Aqua Sun angle Correction C5 results C6 results

y = 0.0025x2 - 0.3173x + 9.4398R² = 0.7822

0 20 40 60 80 100 120 140 160 180

Sun angle (degrees)

y = -0.0045x2 + 0.757x - 26.315R² = 0.7628

0 20 40 60 80 100 120 140 160 180

ift (m

Sun angle (degrees)

no correction in C5

(SNPP+J1) VIIRS Geolocation Status

NASA VIIRS Characterization Support Team (VCST)

Geometric Calibration Group

Guoqing (Gary) Lin, SSAI/GSFC code 619

Robert E. Wolfe, NASA/GSFC Code 619

Zhangshi Yin, Ping Zhang, Bin Tan, SSAI/GSFC Code 619

John Dellomo, GST/GSFC Code 619

James C. Tilton, NASA/GSFC Code 606

NASA MODIS/VIIRS STM Calibration Workshop

Silver Spring, Maryland

18 May 2015 6 June 2016

Acknowledgements

• Thanks Carol Davidson & her Land SIPS (formerly Land

PEATE) Team for processing control point residuals and

possible two line element (TLE) use from both IDPS and

LSIPS forward-&re-processed VIIRS geolocation

products, and testing Geo LUTs updates

• Thanks Geo JAMs -- Robert Williamson (now moved on)

& Rosalie Marley -- for helping us resolving DRs in the

AMP (formally DPA) at the GRAVITE

• Thanks Fred Patt of NASA Ocean Group for helping us

resolving issues related to ephemeris and attitude data

Lin et al., 6 June 2016 VCST/GEO 2

Outline

• Geolocation Performance and Trends

• I- & M-bands geolocation

• DNB geolocation, up to ~ 9/1/2013

• Remaining issues, concerns and challenges

• Some expectations of J1 VIIRS

• Conclusions

Overall Geolocation Performance Residuals IDPS Land SIPS

MODIS C6

Track mean -1 m 8 m 2 m 2 m

Scan mean -1 m 5 m 0 m -1 m

Track RMSE 75 m 72 m 46 m 43 m

Scan RMSE 61 m 61 m 53 m 44 m

Data-days 1127 (3.1 yrs) 1566 (4.3 yrs) 5040 (13.8 yrs) 5849 (16.0 yrs)

Missing days 21 1 10 62

Daily matched

GCPs w/ I1/B1

130 130 189 218

• Nadir equivalent accuracy (RMSE – Root Mean Square Error) . (MODIS for reference)

– Meet Spec: 133 m (1s); within 20% I1 HSI (375 m) = 75 m @ nadir for VIIRS

– Band-to-band mis-registration adds bias to RMSE to other bands:

– Land PEATE/SIPS time period: 19 Jan 2012 to 3 May 2016

• MODIS – VIIRS differences

Aqua use definitive ephemeris data 27 hour latency

SNPP attitude data is not as good, see Slides10 & 11

DEM resolutions: older 1 km for VIIRS vs newer 0.5 km for MODIS C6

22 s RMSE

-100-75-50-25

0255075

12 12.5 13 13.5 14 14.5 15 15.5 16 16.5

NASA LSIPS C1.1 geolocation errors

Started 1/19/2012

Cryo door open

VCST/GEO 5

C1.1 RMSE Track: 72 m Scan: 61 m, nadir equivalent

y = 4.5999x - 57.255R² = 0.3709

-40-30-20-10

010203040

12 12.5 13 13.5 14 14.5 15 15.5 16 16.5

16-day Global16-day Global Linear (16-day Global)

y = 1.8391x - 21.165R² = 0.1346

12 12.5 13 13.5 14 14.5 15 15.5 16 16.5

Possible Long-Term Trending and Correction

Lin et al., 6 June 2016 Land SIPS Re-processed, will be corrected

Linear trend and annual cycle

VCST/GEO 6

y = 0.0018x2 - 0.4397x + 31.08

R² = 0.3885

0 20 40 60 80 100 120 140 160 180

Sun angle (deg) y = 0.001x2 - 0.1782x + 13.271

R² = 0.1088

0 20 40 60 80 100 120 140 160 180

Scan r

Sun angle (deg)

VIIRS Sun Angle Residuals

Land SIPS Re-processed, will be corrected VCST/GEO 7

DNB geolocation error trending

based on coastal area GCP matching

As of Nov 4, 2013, the DNB geolocation accuracy is

Scan: 8 ± 33 μrad Track: -35 ± 68 μrad Scan: 7 ± 28 m Track: -29 ± 57 m over coastal areas

(nadir equivalent with mean altitude of 838.8 km)

Courtesy

of NGAS

2nd update

2/15/2013 1st update

3/30/2012

4/25/2013 fine tuned

4/18/2013 3rd & last update

8/22/2013

Star tracker

induced shift

Star tracker

shift corrected

VCST/GEO 8

DNB TC

geolocation (appending

fields (lat, lon, height,

QF)_TC to the

ellipsoid DNB

geolocation product)

was TTOed on

5/22/2014, 14:30 GMT

(data observation time)

in IDPS.

DNB errors track

with I1 band errors

no more

monitoring

~ 9/1/2013

Time (UTC, 2015-09-23)

SNPP Inclination Adjust Maneuver

-18-16-14-12-10

-8-6-4-202468

1012141618

23:36:00 23:42:00 23:48:00 23:54:00 0:00:00

Time (UTC, 2015-11-05)

SNPP Collision Avoidance Maneuver

“False” attitude

reported during delta-V

maneuvers

• On-orbit control

reference frame is

changed ~ 17 deg to

orient the SC for

thruster burns between

points A & C.

delta-V

VCST/GEO 9

• Geolocation error ~ 250

km between A & C

• We plan to mask the

VIIRS geolocation with

365 730 1095 1460

Time (days since launch)

SNPP attitude (Spec 3s<108") non-compliance

Max 3*Stdev in RPY

Max 3*Stdev outage in RPY

Max 99.73% outage in RPY

SNPP attitude issues

16:54:30 16:55:00 16:55:30 16:56:00 16:56:30 16:57:00

Time (UTC, 2013-03-25)

RollPitchYawLonLat

9:13:00 9:13:30 9:14:00 9:14:30 9:15:00 9:15:30 9:16:00

Time (UTC, 2015-03-04)

Roll Pitck

Yaw Lon

• SNPP attitude under-performed, in

99.73% interpretation of 3s, as early as

2013-03-25

• The MOT is working with the Flight

Project and the SC vendor to tune the star

trackers

• We plan to flag/mask geolocation for 100

< |R, P, Y| < 1800 arcsec (> 1800 arcsec

already has geolocation masked)

Thanks Fred Patt and NASA FDF for support VCST/GEO 10

~7 km error

60 scans

Attitude error 2016-05-02 06:48:50 – 06:50:40z

Difference in image masking

W. Australian coastline, 2016-04-16 False coastline, 2016-05-02 VCST/GEO 11

RGB (I4, I3, I2), 2016-05-02

128 130 132 134 136 138 140 142 144 146 148 150 152 154 156 158

roll pitch yaw

M1 M2 M4 M3 I1 I2 M7 M5 M6 M13 M12 I4 I3 M10 M11 M8 M9 I5 M16 M15 M14

Band Name

Scan Agg1x1

Scan Agg2x1

Scan Agg3x1

M1 M2 M4 M3 I1 I2 M7 M5 M6 M13 M12 I4 I3 M10 M11 M8 M9 I5 M16 M15 M14

Band Name

Scan Agg1x1Scan Agg2x1Scan Agg1x1Track

J1 VIIRS M/I band offsets wrt I1

• J1 scan rate is nominal @3.517 rad/sec (SNPP 3.531)

• Overall BBR band pair performances are compatible between J1 and SNPP

• J1 Bands on SWMWIR and LWIR FPAs shifted from bands on VisNIR FPA, ~ 7% for

M-bands and ~ 14% for I-bands. Mapping uncertainties are affected

S/MWIR FPA LWIR FPA VisNIR FPA

22 s RMSE

VCST/GEO 12

1000 1200 1400 1600 1800 2000 2200

DNB distance from nadir (km)

Beginning of scan

End of scan

0 200 400 600 800 1000 1200 1400 1600

Distance from Nadir (km)

Along-Scan Direction HSIM-band - No Agg

M-band

I-band

J1 DNB Geometric Performance DNB LSFs are mostly square

Baselined pixel size is ~ 750 m

“Option21” has pixel size up to 1.6 km

within 56.5o

Geolocation is extrapolated post-nadir for

scan angle > ~56.5o (pixel size up to 3.9

km @60.5o)

“Option21” – default, in km

extrapolated

Baseline -- SNPP

J1 DNB

default

scan angle > ~56.5o

Nominal maximum

scan angle (~56.5o)

4546474849505152535455565758596061

1400 1500 1600 1700 1800 1900 2000 2100 2200

DNB Sample#, "Option21"

Pre_nadir

Post-nadir

Scan_Angle

“Option21” – default, in Samp#

scan angle

> ~56.5

BL:2032

extrapolated

VCST/GEO 13

BL:1480

Conclusions

• SNPP VIIRS geolocation performance is good

– The attitude system underperforms, geolocation error of 7 km occurred

occasionally. Attitude/star trackers need maintenance.

• Geolocation mean errors for I- & M-bands are near 0 and

uncertainties are ~ 70 m at nadir, statistically

– Caveat: DNB geolocation errors are not monitored after ~9/1/2013

• J1 geolocation expectations – Bands on VisNIR FPA should be good; Bands on cold FPAs will be ~ 50 m off in the

track direction

– DNB geolocation pixels will be larger beyond S#1480, 1100 km off nadir

– DNB geolocation accuracy will be TBD

Questions?

؟ Do random variables exist?

؟ Where can we find biases & remove them?

؟ Is ergodicity a fair assumption?

؟ What do we mean by “mean values?”

؟ in what time/spatial scales?

؟ in any functional forms?

Backup Slides

On-orbit Geolocation LUT Updates

Update Date Description Comments

a 1/19/2012 Cryo-radiator door open All VIIRS band available,

LPEATE re-process start date

1 2/23/2012 Initial mounting coef. update Removed bias ~ 1.3 km

2 3/30/2012 Initial DNB FPA center update Removed bias ~ 1 km

b 11/22/2012 Scan control electronics (SCE) was switched

from B-side to A-Side Caused bias ~ 300 m

3 12/11/2012 Correction after SCE was switched from B-

Side to A-side Removed bias ~ 300 m

4 2/15/2013 Second, fine DNB FPA center update Removed DNB bias ~ 300 m

5 4/18/2013 Second, scan angle dependent, fine Geo LUT

update

Fine tuned and removed scan

dependent biases

c 4/25/2013 Star tracker maintenance/re-alignment Caused bias ~ 25 m

6 8/22/2013 Correction to the star tracker re-alignment Removed bias ~ 25 m

DNB only All bands impacted Key: External event

VCST/GEO 17

-100-75-50-25

0255075

12 12.25 12.5 12.75 13 13.25 13.5 13.75 14 14.25 14.5 14.75 15 15.25

IDPS (NRT) geolocation errors

Started 2/23/2012

1.3 km error before

300 m error in scan direction

2012 11/22 - 12/11

25 m error Star Tracker adjustment

2013 4/25 - 8/22

VCST/GEO 18

R² = 0.3933

-60 -40 -20 0 20 40 60

Scan angle (deg)

R² = 0.0007

-60 -40 -20 0 20 40 60

Scan resid

Scan angle (deg)

VIIRS Scan Angle Residuals

Land SIPS Re-processed VCST/GEO 19

50 60 70 80 90

Day of Year (2012)

50 60 70 80 90

Day of Year (2012)

30 50 70 90

Day of Year (2012)

30 50 70 90

Day of Year (2012)

Initial on-orbit geolocation LUTs Update

Nadir equivalent units;

Track -21 80

Scan -8 64

Error after

LUT update

(2/23/2012, doy 54)

27 days with average of

142 matchups/day

(minus 12 outliers/day)

Biases removed: Track -755 m, Scan 1118 m

VCST/GEO 20

Scan Control Electronics (SCE) Side Switch,

Geolocation Error and Correction

Lin et al., 6 June 2016 (Nadir equivalent units)

12.86 12.87 12.88 12.89 12.90 12.91 12.92 12.93 12.94 12.95 12.96 12.97

After A-side SCE

coefficient update

in IDPS (Dec. 11,

After switch to A-side SCE (Nov. 22,

Scan direction geolocation

shifted ~325 m for 19 data-days

Operated on B-side

SCE since launch

VCST/GEO 21

Star Tracker Re-alignment and Correction

13.2 13.3 13.4 13.5 13.6 13.7 13.8

Daily 16-day Global

13.2 13.3 13.4 13.5 13.6 13.7 13.8

Scan (

res. (m

8/22/2013

4/18/2013 4/25/2013

4/18/2013:

Geo LUTs fine tuned

4/25/2013:

Star tracker re-

alignment

8/22/2013

Error ~ 25 m found

and corrected

VCST/GEO 22

RTA/HAM sync losses

• Sync losses happen more frequently lately

• Sync losses cause data loss (events detected and flagged in SDR/Geo

• 63 times so far, ~ 100 sec each (60 scans), equivalent to 1.0 orbits total Lin et al., 6 June 2016

0 365 731 1,096 1,461

Time (Days since Launch 2011-10-28, A-Side started 2012-11-22)

SNPP VIIRS Sync Loss History (up to the one on 16 May 2016)

A-Side

B-Side

VCST/GEO 23

An outlier period

0 3 6 9 12 15 18 21 24

Time (UTC Hours, 19 Aug 2015)

Track_shiftScan_shift

1 second time error

onboard of SNPP

14:23Z 21:12Z

“SCC Clock slope was performed on August 19, 2015 at 14:23Z”

“SCC Clock was demonstrating a +999.989 ms drift relative to the

GPS time, “

LPEATE/LSIPS archives exclude the datasets now. Could

“manually” correct time error later.

Error in time code

VCST/GEO 24

SNPP, & J1+, “824 km orbit”

SNPP Altitude: =838.8 km, Max = 856 km, Min = 828 km

Wolfe et al., May 13, 2014

0 365 730 1095 1460

Time (Days since Launch 28 October 2011)

0 365 730 1095 1460

Time (Days since Launch 28 October 2011)

Alt_max (km) Alt_min (km)

VCST/GEO 25

0 365 730 1,095 1,460

Time (Days since Launch on Oct 28, 2011)

SNPP Equatorial Crossing Time

101.49

101.51

0 365 730 1,095 1,460

Time (Days since Launch on Oct 28, 2011)

Orbital Period 16-day Repeat Deviation

23 DMUs (DAMs, CAMs, RMMs)

5 IAMs

MODIS Geolocation Status · accuracy goal (50 m) •Large errors occur ~1.5 hr after maneuvers...

Documents

Image Quality Assessment Plug‐in (RMSE and SSIM)kyungs.bol.ucla.edu/software/quality_assessment/quality...3/26/11 1 Image Quality Assessment Plug‐in (RMSE and SSIM) Kyung Sung,

Overall Statistics RMSE WRF-UA: 159 W m -2 WRF-UCSD: 171 W m -2 STDERR ‘Bias-corrected RMSE’

Performing Basic Vehicle Maneuvers

Doc. No.: Issue · 2017. 6. 6. · A relative geometric accuracy of the about 3800 orthorectified scenes is reached better than 10 m RMSE in each direction with respect to the IMAGE2000

P o Nice - FRANCE Crisis Information Hydrography ...€¦ · estimated geometric accuracy (RMSE) is 12.5 m or better, from native positional accuracy of the background satellite image

Advanced Combat Maneuvers

IIIIMPACT OF PARKING MANEUVERS ON MPACT OF PARKING MANEUVERS ON

Bppv and particle repositioning maneuvers

LEOPOLD’S MANEUVERS - OBGYN Academy

Chapter 4: Basic Maneuvers

Map Maneuvers Sayre

Wrestling Maneuvers & Arenas

Pan-Sharpening With a Hyper-Laplacian Penalty · Gaussian, RMSE = 2.385 Laplacian,RMSE=1.58 Hyper−Laplacian,RMSE=0.961 Figure 3. Fitting curves to empirical image gradient data

Accuracy of Meteonorm (7.1.6.14035) model • Steps of uncertainty model = steps of chain of algorithms • Expert model for measurement uncertainty • Uncertainty = RMSE = approx

Aerobatics Maneuvers

Offensive Maneuvers

Objectives Differentiate accuracy, precision, error, and uncertainty. Discuss the dimensions of geographic data quality. Discuss how to compute RMSE for

RMsE Lecture

SOVIET ARMY TRAINING, MANEUVERS AND MISCELLANEOUS INFORMATION · soviet army training, maneuvers and miscellaneous information ... soviet army training, maneuvers and miscellaneous

Reverse Engineering Maneuvers