Operational Handling of Conjunction Warnings in EUMETSAT€¦ · in EUMETSAT for operational assessment Based on JSpOC data, conjunction analysis prototype system (Visual Basic in

Slide: 1 SpaceOps WS 2011 – Pier Luigi Righetti

Operational Handling of Conjunction Warnings in EUMETSAT

Pier Luigi Righetti


Presentation Overview

• Introduction and History

• Available Data

• Conjunction Analysis Processing

• Operations

• GEO CA Activities

• Conclusions

• Accuracy Analysis of Metop Orbit (tomorrow)

• Accuracy Analysis of Meteosat Orbit (tomorrow)


Introduction and History

• EUMETSAT operates 1 LEO (Metop-A) and 3 GEO (Meteosat 7, 8 and 9) satellites for meteorological operations

• Launch of 1 LEO (Metop-B) and 1 GEO (Meteosat 10) foreseen in 2012; more launches in the near future (Sentinel-3A&B and Meteosat 11 in 2014)

• No conjunction analysis performed by EUMETSAT till 2008; TLE accuracy considered insufficient for satisfactory risk mitigation (SpaceOps 2010)

• First emergency conjunction message received from JSpOC (via ESOC) on 24 December 2008 for the Metop-A satellite

• Since then, several interaction with JSpOC to better understand data needs and contents

• Currently, high accuracy conjunction data from JSpOC available routinely in EUMETSAT for operational assessment

• Based on JSpOC data, conjunction analysis prototype system (Visual Basic in Excel) developed for the LEO satellite: on-going migration to FDF


Available DataNASA/GSFC Conjunction Assessment Summary Report

• Daily e-mail from a.i. Solutions (NASA-GSFC) since 06/2009

• All (offending) objects within 1km from the asset (Metop-A) for 5 days

• Time of close approach (TCA)

• Total Miss-distance

• Miss-distance components in asset orbital frame

• Probability of Collision (PoC); based on 20m radius (good agreement with EUM computation)

• No velocity, no covariance• Parameters evolution in time


Available Data JSpOC Conjunction Assessment Results

• Daily e-mail received from JSpOC since 10/2010

• All objects within 5km from the asset for 5 days• Time of close approach and miss-distance (total and components)• Equivalent to NASA/GSFC Conjunction Assessment Summary Reports

METOP-A SP Screening Results for 20Oct--------------------------------------

CONJ# Conjunction Time (UTC) Miss (Radial, In-Track, and Cross-Track)

260 22 OCT/14:27:21.282 4011m (-805.5 2272.4 -3205.9)12848 23 OCT/11:31:41.347 4590m (-3432.8 -28.2 3047.8)24973 23 OCT/03:51:42.210 2695m (633.2 801.4 2494.8)30006 23 OCT/13:58:52.999 4262m (1874.8 -3353.3 1847.3)31156 23 OCT/08:04:48.525 2757m (-736.1 2603.9 531.6)31987 25 OCT/13:19:32.421 3353m (-3221.8 -883.2 -290.7)32460 24 OCT/10:14:03.469 3121m (-2154.2 1524.6 -1666.4)35918 22 OCT/21:57:57.577 1812m (1464.0 471.8 -958.3)


Available DataJSpOC Conjunction Summary Messages (CSM)

• CSM posted by JSpOC on EUMETSAT FD account in Space-Track website whenever an object detected into the screening region around the asset

• Initial screening region (since 08/2010) miss-distance below 1km and radial below 200m

• Current screening region (since 10/2010)miss-distance within high risk ellipsoid (300m R, 2500m A, 1250m C)

• Earlier detection of dangerous objects with large along-track separation

• Full state vector and covariance matrix for asset and object

• Auxiliary information on data quality (estimation arc, residuals, data age)

• Excellent quality observed for asset data (results presented later)


Conjunction Analysis Processing

• Analysis of the conjunction geometry• Depth of intrusion (DoI) is computed for all available conjunctions • The probability of collision is computed for all conjunction with DoI below 1

• Computation of the probability of collision (PoC)• If PoC goes over 1 / 15000 then a collision avoidance manoeuvre is prepared

• Computation of collision avoidance manoeuvre• Several manoeuvres reducing the residual error to below 1 / 109 are computed• Optimal one selected for implementation (together with operations management)• Post manoeuvre orbit assessed against known objects (DoI)• Post-manoeuvre orbit provided to JSpOC for conjunction assessment

• Collision avoidance manoeuvre implementation• If risk still over 1/10000 at time of generation of the manoeuvre commands


Conjunction Analysis ProcessingAnalysis of the Conjunction Geometry

• Analyse asset trajectory in the reference frame of the object

• Default velocity assumed for the object (7.4 km/s, circular orbit)

• Estimation of how deep the asset enters the covariance ellipsoid of the object

• Default covariance used for the object (R, AT, CT : 250, 2000, 250 m)

• Check if depth of intrusion below 1• Asset enters into reference covariance

ellipsoid of the object

• All objects within high risk ellipsoid have DoI below 1


Conjunction Analysis ProcessingComputation of the Probability of Collision (PoC)

• Simplified Alfriend’s formula used(similar to ESOC, CNES and GSFC)

• Weighted probability considered:at miss-distance (PoCmd);at md – r/2 (PoCmdm);at md + r/2 (PoCmdp);-> PoC = ¼ (PoCmdm + 2PoCmd + PoCmdp)

• Accurate computation of impact area taking into account:- solar array rotation - commanded attitude

• Data confidence factor applied (F<1 for “bad” data)

• Observations per day (Fo=1 if > 10)• Data arc versus optimal (Fd=1 if 1)• RMS of residuals (Fr=1 if ~1)• Age of the data (Fa=1 if < 24 hours)

• F = Fo * Fd * Fr * Fa


Conjunction Analysis Processing Computation of Collision Avoidance Manoeuvre

• Manoeuvre effect on asset position and covariance computed

• Analytical propagation using Clohessy-Wiltshire equations

• Conjunction conditions not respected; marginal degradation of accuracy

• Different sizes and execution times analyzed

• N+0.5 orbits before TCA considered• N+0.75 sometimes even better

• Residual risk curves computed• For lateral conjunction, benefit in

anticipating the manoeuvre• Target residual risk < 1/109


OperationsTimeline and Intervention Rules


OperationsHigh Risk Conjunctions Observed

• High risk conjunction on 28/12/2009• No CSM available; worst case analysis performed and avoidance manoeuvre prepared• Radial covariance delivered shortly before upload of commands; • Worst case analysis repeated and manoeuvre cancelled

• High risk conjunction on 09/09/2010• High risk observed for large covariance when event far in the future• Further tracking on last days before the event reduced the covariance and risk vanished

• High risk conjunction on 20/11/2010• High risk observed for large covariance when event far in the future• No tracking performed on last days before the event (very small object)• Manoeuvre not performed due to the age of the last observation (>48 hours)

• High risk conjunction on 08/03/2011• Same behaviour as for conjunction on 09/09/2010

• High risk conjunction on 01/05/2011 -> next slides


OperationsConjunction on 1st of May 2011: First Report

• High risk conjunction reported on 26/04/2011• Offending Object: piece of COSMOS 2251 (NORAD ID #34451)• TCA: 01/05/2011-06:00:08.4• Miss-distance : -26m (radial), 6m (along), 49m (across)• Impact angle (relative velocity): 170 deg; nearly frontal impact• DoI: 0.2 (well below 1)

• CSM received• Object covariance: 17m (r), 2375 (a) 10m (c)• Asset covariance: 5m (r), 160m (a), 3m (c)

• Results of PoC computations• PoC: 1/3980

(reference radius 3.5m)• Confidence level: 0.7

(old observations, high RMS)• High risk confirmed


OperationsConjunction on 1st of May 2011: Data Evolution

• High risk conjunction confirmed on 27,28,29/04/2011• DoI increases (except for last day)• Covariance ellipsoid size decreases• Risk decreases (except for last day)• Data quality increases

(new measurements performed)• Escalation to management

(meeting on 30/04/2011 AM)• Preparation of preliminary

avoidance manoeuvre


OperationsConjunction on 1st of May 2011: Manoeuvre Preparation

• Manoeuvre impact on residual PoC• No real benefit in anticipating the manoeuvre

for positive thrust ( detrimental reduction in across separation)

• Little benefit for negative thrust (little increase in across separation)

• Manoeuvre selection• 15mm/s manoeuvre 0.5 orbit before TCA• -15mm/s manoeuvre 1.5 orbits before TCA• Latest JSpOC data expected

on 30/05/2011 late PM• Positive manoeuvre detrimental

for GT maintenance• Negative much better• Negative manoeuvre selected


OperationsConjunction on 1st of May 2011: Last Report

• Last Report received late PM on 30/04/2011• Miss-distance : -8m (radial), -6m (along), -54m (across)• DoI: 0.2 (well below 1)

• CSM received• Object covariance: 9m (r), 205 (a) 8m (c)• Asset covariance: 3m (r), 15m (a), 2m (c)

• Results of PoC computations• PoC: 1/316

(reference radius 3.5m)• Confidence level: 0.8

(still high RMS)• Very High risk confirmed

• Covariance ellipsoid size decreased for the object…• …but miss-distance decreased too (still within previous day covariance ellipsoid)• Asymmetric shrinking of the covariance ellipsoid!

D-1

D-2


OperationsConjunction on 1st of May 2011: Manoeuvre Execution

• Assessment of a-priori manoeuvre effect• Post manoeuvre orbit analyzed against objects in JSPOC conjunction assessment

results: no further high risk conjunction identified• Effect of the manoeuvre analyzed using Clohessy-Wiltshire approach:

• Miss-distance of some 135 meters with 50 m radial separation• Residual PoC : 1/3E7 (target at planning level 1/1E9): 30 times bigger than target

(computed with previous CSM), but still acceptable• Post manoeuvre orbit sent to JSpOC for conjunction assessment: positive feedback• Effect of the manoeuvre analyzed using operational

Flight Dynamics:• Same results as with C-W observed

• Manoeuvre upload authorized at 20:00Z• Manoeuvre upload performed at 23:05Z• Manoeuvre executed at 3:28Z (on 01/05)• TCA at 6:00, in visibility; no collision


OperationsConjunction on 1st of May 2011: Manoeuvre Effect

• Assessment of real manoeuvre effect• Little underperformance observed in the manoeuvre execution• Post manoeuvre orbit sent to JSpOC to speed-up recovery of orbit determination accuracy• Effect of the real manoeuvre analyzed using Clohessy-Wiltshire approach

(object covariance set to latest pre-manoeuvre CSM):• Miss-distance of some 130 meters with 40m radial separation• Residual PoC : 1/1E7 (expected 1/3E7): 3 times bigger than expected

(computed with a-priori manoeuvre), but still acceptable• Effect of the manoeuvre analyzed using post manoeuvre CSM (sent via FTP to EUM):

• Further decrease of the covariance size of the object (8m, 32m, 7m)• Reported Metop-A state vector at TCA still not very accurate:

only 20m radial separation with object reported20 meters difference in radial with operational orbit observed large radial covariance of 10m

• A-posterior PoC, computed using object orbit from CSM and operational orbit for Metop A: 1/5E38 (zero)

• Risk reduction of the manoeuvre confirmed


GEO

• JSpOC support• CSM made available in case of object getting closer that 5km for Met7, 8 and 9• No daily report provided• Post manoeuvre orbit provided by EUM to JSpOC before manoeuvre execution for

identifying possible conjunction on the new orbit• False alarm received for a conjunction with an operated satellite which had performed a

manoeuvre shortly before the issue of the CSM (not yet included in JSpOC orbit)• Conjunction handling system starts being also used for the GEO satellites;

applicability of same rules and methods is still under analysis (support may be needed)

• SDA support• EUM participates in the SDA conjunction assessment program and EUM provides daily

SDA with orbital information for their GEO satellites (including manoeuvre effect)• SDA provides twice a day a screening report on all objects within 50 km• Conjunction assessment performed using the orbit provided by the operator for the

around 200 satellites controlled by participants of SDA (as Inmarsat, SES, Intelsat...)• Conjunction assessment performed using the TLE provided by Space-Track for all

other objects


Conclusions and Next Steps

• Conclusion• Two years ago first conjunction warning received for the Metop-A satellite

• Full set of SW tool and operational procedures in place to handle those data efficiently

• More than 30 conjunction reported within the high-risk ellipsoid (since August 2010)

• Four of those where identified as high risk conjunction

• First collision avoidance manoeuvre successfully performed on first of may 2011

• Next steps• Full integration of the SW tool within the FDF framework

• Addition of covariance to orbital data provided to JSpOC (SW under development)

• Consolidation of GEO conjunction analysis process

• EUMETSAT thanks NOAA, JSpOC, CNES, ESOC, GSFC for their support


Accuracy Analysis of Metop OrbitJSpOC Orbit Post Manoeuvre Recovery Time

• Assessment from received JSPOC conjunction assessment results • Effect of manoeuvre on events after conjunction observed when a-priori predicted

manoeuvre included in Metop-A orbit• Effect of executed manoeuvre in the same events observed from first data provided

on day after the manoeuvre• Excellent agreement between relative changes in miss-distance and manoeuvre

execution efficiency

30-Apr PM without predicted mano cross totalper mm/s along total

per mm/s

33766 04 MAY/02:47:55.885 2291m (-77.5 149.1 -2285.8) Later conjunction -2285.8 149.134451 01 MAY/06:00:08.350 55m (-8.4 -6.0 -54.1) Critical conjunction82820 03 MAY/21:39:01.267 3935m (3904.2 93.5 489.4) Later conjunction 489.4 93.5

30-Apr PM including predicted mano33766 04 MAY/02:47:55.125 1550 (-26.4 102.5 -1546.7) Mano Effect on later conjunction -1546.7 739.1 49.3 102.5 -46.6 -3.1134451 01 MAY/06:00:08.328 123 (51.6 -15.1 -111.0) Mano Effect on critical conjunction82820 03 MAY/21:39:00.555 4054 (3881.3 -170.5 -1159.5) Mano Effect on later conjunction -1159.5 -1648.9 -109.9 -170.5 -264 -17.60

01-May PM (including manoeuvre calibrated - by JSpOC directly?)33766 04 MAY/02:47:55.249 1675m (-26.9 109.0 -1671.3) Mano Effect on later conjunction -1671.3 614.5 50.0 109 -40.1 -3.2682820 03 MAY/21:39:00.625 4002m (3896.7 -129.6 -905.0) Mano Effect on later conjunction -905 -1394.4 -113.4 -129.6 -223.1 -18.14

02-May PM (including manoeuvre calibrated - by JSpOC directly)33766 04 MAY/02:47:55.261 1685m (-37.2 106.6 -1682.0) Mano Effect on later conjunction -1682 603.8 49.1 106.6 -42.5 -3.4682820 03 MAY/21:39:00.612 3985m (3891.0 -121.8 -856.2) Mano Effect on later conjunction -856.2 -1345.6 -109.4 -121.8 -215.3 -17.50

03-May PM (including manoeuvre calibrated - by JSpOC directly)33766 04 MAY/02:47:55.255 1673m (-19.7 109.8 -1670.3) Mano Effect on later conjunction -1670.3 615.5 50.0 109.8 -39.3 -3.20

displacement (m) displacement (m)


Accuracy Analysis of Metop OrbitJSpOC Orbit Comparison versus POD

• Assessment from received JSPOC CSM• Good agreement in along-track and radial direction (both absolute and relative to sigma)• Some discrepancy in the cross-track direction: TDR/J2000 conversion mismatch? • Consistency degrades marginally for CSM close to TCA (smaller covariance)


Accuracy Analysis of Meteosat OrbitJSpOC Orbit Comparison versus EUM MET7 Orbit

• MET7 orbit provided to JSpOC• Large bias in along-track observed• Error in the UT1-UTC value in EUM

operational SW (set always to zero)• Operational SW corrected to set the

right UT1-UTC value (-0.143 sec)• Along track bias disappeared• Post manoeuvre orbit still presenting

a drift (gravity field mismatch?)


Accuracy Analysis of Meteosat OrbitJSpOC Orbit Comparison versus EUM MET9 Orbit

• MET9 orbit provided to JSpOC• Still large along-track bias observed• Station calibration error?• Even larger bias observed for MET8• Important drift observed (?)


Accuracy Analysis of Meteosat OrbitMSG Orbit Determination Refinement with JSpOC CSM

• Calibration of ranging bias using JSpOC CSM• Reference orbit generated from MSG CSM• Large bias still observed wrt operational orbit

(also after removing UT1-UTC error)• Generation of pseudo-observables from this orbit• Fix transponder delay and estimate ranging bias

of both operational ground stations• Perform orbit determination using 2-way ranging

data from both operational ground stations and pseudo-observables from CSM

• Estimated ranging bias presenting differences with a-priori values from close loop calibration(over 20 meters for secondary GS)

• Differences with reference orbit from JSpOC CSM now much smaller• Good match between observed differences and reported position uncertainty• Orbit determination performed on different dates using as fixed values the

calibrated ranging bias provides similar accuracy

Documents

Operational Handling of Conjunction Warnings in EUMETSAT€¦ · in EUMETSAT for operational assessment Based on JSpOC data, conjunction analysis prototype system (Visual Basic in