laas study of slow-moving ionosphere anomalies and their potential impacts
DESCRIPTION
LAAS Study of Slow-Moving Ionosphere Anomalies and Their Potential Impacts. Ming Luo, Sam Pullen, Seebany Datta-Barua, Godwin Zhang, Todd Walter, and Per Enge Stanford University (with funding from FAA SatNav LAAS Program Office, AND-710) ION GNSS 2005Long Beach, CA.Session E5 - PowerPoint PPT PresentationTRANSCRIPT
LAAS Study of Slow-Moving Ionosphere Anomalies and Their Potential Impacts
Ming Luo, Sam Pullen, Seebany Datta-Barua,
Godwin Zhang, Todd Walter, and Per Enge
Stanford University
(with funding from FAA SatNav LAAS Program Office, AND-710)
ION GNSS 2005 Long Beach, CA. Session E5
16 September 2005
16 September 2005 2
Presentation Outline
• LAAS Ionosphere Anomaly Threat Model
• Ionosphere Anomaly Data Analysis
– 20 Nov. 2003 data in MI/OH (summary)
– 31 Oct. 2003 data in Florida
• Potential Impact on LAAS
– “Worst-case” threat model assessment
– “End-around check” data-replay assessment
• Conclusions and Ongoing Work
16 September 2005 3
Max Iono delay
Front Speed
Nominal IonoWidth
LAAS Model of Iono. Spatial Anomaly
Iono Front
• Moving wave front scenario:Iono wave front moves in the same direction as the airplane does and “catches” the airplane from behind before reaching the LGF
• Stationary front scenario:
Ionospheric wave front stops moving before reaching the LGF
Simplified model: a wave front ramp defined by the “slope” and the “width”.
Slope
An illustration of the impact on LAAS users
Front Speed
Airplane Speed 70 m/s
45 km
LGF
16 September 2005 4
Iono. Anomaly from JPL IGS/CORS Data (20 Nov. 2003; 20:15 – 21:00 UT)
16 September 2005 5
Subset of OH/MI Stations that Saw Similar Ionosphere Behavior on 11/20/2003
0 50 100 150 200 250 300 3500
5
10
15
20
25
30
35
WAAS Time (minutes from 5:00 PM to 11:59 PM)
Sla
nt Io
no D
elay
(m
)S
lant
Ion
o D
elay
(m
)
Sharp falling edge; slant gradients 300 mm/km from
previous work
Weaker “valley” with smaller (but still anomalous)
gradients
Initial upward growth analysis
continues…
Stations from Groups B and D
16 September 2005 6
Iono. Anomaly from JPL IGS/CORS Data: 10/31/03 01:00 ─ 02:40 UT
16 September 2005 7
Iono. Anomaly from JPL IGS/CORS Data: 10/31/03 03:00 ─ 04:40 UT
16 September 2005 8
Iono. Anomaly from JPL IGS/CORS Data: 10/31/03 05:00 ─ 06:40 UT
16 September 2005 9
CORS Stations in Florida and SE Region
16 September 2005 10
Slant Delay Observed at GNVL (Gainesville) and PLTK (Palatka), FL: PRN 29, 31 Oct. 2003
• GNVL and PLTK are ~ 60 km apart.
• PRN 29 is at 15-20
• Estimated Slant Slope: 210 mm/km
• Speed between the two stations appears ~ 200 m/s
4 4.2 4.4 4.6 4.8 5 5.2 5.4 5.6 5.8 6-15
-10
-5
0
5
10
15
20
25GNVL
PLTKDifference
Hours Past Midnight UT on 31 Oct. 2003
Sla
nt I
ono.
Del
ay (
m)
L1 Code-minus-Carrier Data
16 September 2005 11
Slant Delay Observed at GNVL and PLTK, FL: PRN 10 (SVN 40), 31 Oct. 2003
• GNVL (Gainesville) and PLTK (Palatka) are ~ 60 km apart.
• PRN 10 is at 70-80
• Estimated Slope: 100 mm/km
• Appears to be slow-moving:
~ 60 m/s between these two stations.
2 3 4 5 6 7 8 9-4
-2
0
2
4
6
8
10
12
14
GNVL
PLTKDifference
Hours Past Midnight UT on 31 Oct. 2003
Sla
nt I
ono.
Del
ay (
m)
L1 Code-minus-Carrier Data
16 September 2005 12
Slant Delay Observed at GNVL and PLTK, FL: PRN 10 (SVN 40), 31 Oct. 2003
• Plot of same event using L1–L2 data is very similar to L1 code-minus-carrier result
• Data gaps are due to (semi-codeless) L2 loss-of-lock
Hours Past Midnight UT on 31 Oct. 2003
Sla
nt I
ono.
Del
ay (
m)
Post-Processed L1 – L2 Data
3 4 5 6 7 8
1
2
3
4
5
6
7
8
9
10
11PLTK satellite 40
GNVL satellite 40difference between two IPPs
16 September 2005 13
Slant Delay Observed at PLTK and JXVL (Jacksonville), FL: PRN 10, 31 Oct. 2003
Using JXVL instead of GNVL
shows very similar “slow-moving” event
on PRN 10
(PLTK and JXVL are ~ 75 km
apart)
2 3 4 5 6 7 8 9
-5
0
5
10
15
20
Hours Past Midnight UT on 31 Oct. 2003
Sla
nt I
ono.
Del
ay (
m)
JXVL
PLTK
Difference
L1 Code-minus-Carrier Data
16 September 2005 14
Slant Delay Observed at NAPL (Naples) and MTNT (West of Miami), FL: PRN 10, 31 Oct. 2003
For PRN 10, a slow-moving pattern
similar to that seen from NE Florida is also observed in SW Florida (~ 400
km away)
2 3 4 5 6 7 8-6
-4
-2
0
2
4
6
8
10
12
NAPL
MTNT
Difference
Hours Past Midnight UT on 31 Oct. 2003
Sla
nt I
ono.
Del
ay (
m)
L1 Code-minus-Carrier Data
16 September 2005 15
LAAS Threat Model Parameter Bounds approved in Sept. 2004
Elevation Speed Width Slope Max Error
Low elevation < 12
0 – 1000m/s
25 – 200km
30 – 150mm/km
25 m
High elevation ≥ 12
70 – 1000 m/s
25 – 200km
30 – 500mm/km
25 m
0 – 70m/s
25 – 200km
30 – 250mm/km
25 m
• Max error and slope are in the vertical (zenith) direction
• Two changes proposed based on more-recent data analysis:– Interpret numbers in slant direction (change max. slope const. to
~ 50 m) still bounds all verifiable observed events
– Restrict max. slope of slow-moving events to 200 mm/km or less
16 September 2005 16
Availability Assessment for Stationary Fronts at Memphis PSP Site (7/18/05 almanac, all SV healthy)
• No geometry (among 145) has vertical error greater than 10 m.
• The maximum VPLH0 among these geometries is
about 5 m.
• Note that max. error exceeds VPLH0 for all geometriesVPLH0
16 September 2005 17
With one satellite out:
• Maximum vertical error is 22 m
• 46 geometries (out of 4060) have errors > 10 m
• (46/4060 = 0.0113)
Availability Assessment for Stationary Fronts at Memphis PSP Site (7/18/05 alm, all one-SV-out cases)
16 September 2005 18
Availability Assessment for Stationary Fronts at Memphis with Slant Slope Limit ≤ 200 mm/km
Even with one SV out, no geometry (among 4060) has
vertical error greater than 10 m.
(7/18/05 alm, all one-SV-out cases)
16 September 2005 19
Differential Slant Delay Observed between PLTK and GNVL, FL: All Satellites, 31 Oct. 2003
0 2 4 6 8 10-15
-10
-5
0
5
10
15D
iffe
rent
ial S
lant
Ion
o. D
elay
(m
)
PRN 4
PRN 5
PRN 6
PRN 10
PRN 17
PRN 24
PRN 28
PRN 29
Hours Past Midnight UT on 31 Oct. 2003
Recall that GNVL and PLTK are ~ 60 km apart
PRN 29 (low-elevation; fast-moving iono. @
210 mm/km)
PRN 10 (high-elevation; slow-moving iono. @
100 mm/km)
16 September 2005 20
Range and Position Error between PLTK (“user”) and GNVL (“LGF”), FL: 31 Oct. 2003
• Range errors from all satellites are
included
• Based on actual GPS constellation on Oct 31 of 03
• Max vertical error is about 6 m at about 04:30 UT
• If scaled down to typical ≤ 5-km phys.
separation between user and LGF, diff. error would be
significantly smaller0 2 4 6 8 10 12
-15
-10
-5
0
5
10
15
Dif
fere
ntia
l Err
or (
m)
PRN 4
PRN 5
PRN 6
PRN 10
PRN 17
PRN 24
PRN 28
PRN 29
Vertical Position Error
Hours Past Midnight UT on 31 Oct. 2003
16 September 2005 21
Conclusions and Ongoing Work
• Iono. anomaly data analysis has turned up at least one verifiable slow-speed event in Florida – ~ 60 m/s event on high-elevation PRN 10 is confirmed by
multiple reference stations spread around Florida
– OH/MI gradients are more severe, but all verified points analyzed to date are moving faster than 140 m/s
– Data analysis continues to support “finalized” threat model
• Slow-speed events should remain in GBAS threat model, but reduction of max. gradient is advisable
• Impact on LAAS availability (of integrity) is not severe if maximum slow-speed slant slope is ≤ 200 mm/km
• “End-around-check” replay of Florida data shows that worst-case position error is well below 10-meter VAL (and may be “boundable” by inflated VPLH0)
Backup Slides follow…
16 September 2005 23
Iono. Anomaly from JPL IGS/CORS Data (29 Oct. 2003; 20:00 – 20:45 UT)
16 September 2005 24
Iono. Anomaly: Threats to WAAS
• WAAS corrections are based on planar fits to measured iono. delays
• Thus, threats include:
» deviations from linearity (mitigated by chi-square “storm detector”)
» bubbles of enhanced or depleted iono. delay that fall inside WAAS iono. pierce points (mitigated by “undersampled” threat model)
16 September 2005 25
11/20/2003 Ionosphere Storm as Seen From OH/MI CORS Cluster (for SVN 38)
17:40 18:30 19:20 20:10 21:00 21:50 22:30 23:300
5
10
15
20
25
30
35All Stations in OH-MI area, SV 38, 11/20/03, 17:00 - 24:00; Figure filename: sv38_17-24.fig
Sla
nt I
ono
Del
ay (
m)
Sla
nt I
ono
Del
ay
(m)
Time (hours, UTC)
16 September 2005 26
CORS Stations in Ohio/Michigan Region
Group A
Group B
Group D
Group C
Group E
Group F
-87 -86 -85 -84 -83 -82 -81 -80 -79
38
39
40
41
42
43
44
45
46
MTVRPIT1
METR
UNIV
PARY
LANS
TIFF
UPTC
PWELMPLE
SUP3
SUP2
WOOS
CLREBRIG
GALP
GUST
COLB
GALB
PAPT
PTIR
UVFM
KNTN
AVCA
HRUF
BAYR
SAG1
PCK1
BFNY
WLCI
LEBA
SIDN
SOWR
ERLA
HBCH
YOU2YOU1
CASS
GRTN
SIBY
STKR
ADRI
LSBN
MCON
DEFI
FREO
PKTN
FRTGGRAR
NOR3NOR2
NOR1
HRN1
DET2
GARFTLDO
IUCO
OKEE
VAST
MIO1
LOU1
16 24
27
37
40
57
62
70
74
79
83
84
88
89 91
92
95
97
105
106
119
120
122
124
128
132
134
149
150
151
171
175
177
178
186
192193
196
213
217
234
236
248
249
261
265
275
285292
301302
303
307
316
330337
340
345
347
356
375
16 September 2005 27
Histograms of Velocity Normal to Front (Vn) Based on Three-Station Trigonometric Fit
Normal Velocity Vn (m/s)
-100 0 100 200 300 400 500 6000
5
10
15
20
25
No.
of
Occ
urre
nces
At Sharp Falloff In “Valley” Section
Further analysis placed doubt on low-speed results
150 200 250 300 350 400 4500
1
2
3
4
5
6
7
8
Normal Velocity Vn (m/s)
16 September 2005 28
Satellites In View on 31 October 2003 for GNVL (Gainesville), Florida
x
4 60
30
0
30
210
60
240 120
300
150
330
N
E
S
W
'x' denotes last point in timeelevation is 90 degrees at center
6.0 Hours of Coverage from 10/31/03 0:00:00 for GPS Visibility for GNVL, Florida
x
5
x
6
x
7
x
8
x
9
x
10
x11
x
13
x
17
x
20x21x
24
x
26
x27 x
28x
29
x
30
16 September 2005 29
Slant Delay Observed at PLTK and JXVL, FL: PRN 24, 31 Oct. 2003
While gradient is smaller (~ 30 mm/km), note persistence of gradient over
2.5-hour period
0 1 2 3 4 5 6 7 8 9
-4
-2
0
2
4
6
8
10
12
14
16
18
JXVL
PLTKDifference
Hours Past Midnight UT on 31 Oct. 2003
Sla
nt I
ono.
Del
ay (
m)
L1 Code-minus-Carrier Data
16 September 2005 30
Slant Delay Observed at NAPL and MTNT, FL:PRN 29, 31 October 31 2003
4 5 6 7 8 9 10-10
-5
0
5
10
15
20
Hours in Oct 31, 2003 (hr)
Sla
nt
De
lay
(m
)
Slant Delay Observed at NAPL and MTNT, FL, Oct 31, 03, PRN 29
NAPL
MTNTDifference
For PRN 29, a faster-moving
pattern similar to that seen from NE Florida is
also observed in SW Florida
(~ 450 km away), but MTNT data
jump makes precise analysis
difficult
16 September 2005 31
Slant Delay Observed at PNCY and MRKB, FL:
PRN 10, 31 Oct. 2003
2 3 4 5 6 7 8 9-4
-2
0
2
4
6
8
10
Hours in Oct 31, 2003 (hr)
Sla
nt
De
lay
(m
)Slant Delay Observed at PNCY and MRKB, FL, Oct 31, 03, PRN 10
PNCY
MRKBDifference
16 September 2005 32
Slant Delay Observed at PNCY and MRKB, FL: PRN 29, 31 Oct. 2003
4 5 6 7 8 9 10-4
-2
0
2
4
6
8
10
Hours in Oct 31, 2003 (hr)
Sla
nt
De
lay
(m
)
Slant Delay Observed at PNCY and MRKB, FL, Oct 31, 03, PRN 29
PNCY
MRKBDifference
16 September 2005 33
Slant Delay Observed at DFNK and TALH, FL: PRN 5, 31 Oct. 2003
1 2 3 4 5 6 7-4
-2
0
2
4
6
8
Hours in Oct 31, 2003 (hr)
Sla
nt
De
lay
(m
)
Slant Delay Observed at DFNK and TALH, FL, Oct 31, 03, PRN 5
DFNK
TALHDifference
16 September 2005 34
Slant Delay Observed at DFNK and TALH, FL: PRN 10, 31 Oct. 2003
2 3 4 5 6 7 8 9 10-4
-2
0
2
4
6
8
10
12
Hours in Oct 31, 2003 (hr)
Sla
nt
De
lay
(m
)Slant Delay Observed at DFNK and TALH, FL, Oct 31, 03, PRN 10
DFNK
TALHDifference
16 September 2005 35
Slant Delay Observed at DFNK and TALH, FL:
PRN 24, 31 Oct. 2003
0 1 2 3 4 5 6 7 8 9-4
-2
0
2
4
6
8
10
12
Hours in Oct 31, 2003 (hr)
Sla
nt
De
lay
(m
)
Slant Delay Observed at DFNK and TALH, FL, Oct 31, 03, PRN 24
DFNK
TALHDifference
16 September 2005 36
Slant Delay Observed at DFNK and TALH, FL: PRN 29, 31 Oct. 2003
4 5 6 7 8 9 10-6
-4
-2
0
2
4
6
8
10
12
Hours in Oct 31, 2003 (hr)
Sla
nt
De
lay
(m
)
Slant Delay Observed at DFNK and TALH, FL, Oct 31, 03, PRN 29
DFNK
TALHDifference
16 September 2005 37
Slant Delay Observed at DFNK and TALH, FL: PRN 28, 31 Oct. 2003
0 0.5 1 1.5 2 2.5 3 3.5 4
0
5
10
15
20
Hours in Oct 31, 2003 (hr)
Sla
nt
De
lay
(m
)
Slant Delay Observed at DFNK and TALH, FL, Oct 31, 03, PRN 28
DFNK
TALHDifference
16 September 2005 38
Florida Data Analysis Summary
• CORS data from Florida region on 10/31/03 (UT) provides the clearest example of slow-moving iono. fronts seen thus far
– Event on high-elevation PRN 10 is confirmed by multiple reference stations spread around the state
– L1-L2 results look very similar to L1 code-minus-carrier
• Slopes of slow-moving events studied to date are as large as ~ 100 mm/km (slant)
– Fast-moving events show possible larger gradients
• Fortunately, gradients of this size are unlikely to be hazardous to LAAS
– LAAS threat simulation results to come…
16 September 2005 39
Impact of Florida Anomaly on WAAS
• To be filled in if needed…
16 September 2005 40
Updated Candidate Threat Model (proposed by FAA in Aug. 2005)
Max error and slope are in the slant direction
Slow-speed possibility is removed
Elevation Speed Width Slope Max Error
Low elevation < 12
100 – 1000m/s
25 – 200km
30 – 150mm/km
50 m
High elevation ≥ 12
100 – 1000 m/s
25 – 200km
30 – 500mm/km
50 m
16 September 2005 41
Sam’s Proposed Threat Model as of Today…
Max error and slope are in the slant direction
Elevation Speed Width Slope Max Error
Low elevation < 12
0 – 1000m/s
25 – 200km
30 – 150mm/km
50 m
High elevation ≥ 12
100 – 1000 m/s
25 – 200km
30 – 500mm/km
50 m
0 – 100m/s
25 – 200 km
30 – 150mm/km
50 m
16 September 2005 42
0.005 0.01 0.015 0.02 0.025 0.03 0.035 0.04 0.045 0.050
20
40
60
80
100
120
140
160
180
200
Ionospheric Rate (m/s)
Tim
e-to
-det
ect
(s)
Simplified Time-to-detect vs Ionospheric Rate
by LGFby Airborne
GMA Only
MQM
CUSUM
Time-to-detect vs. Ionospheric Rate (From Stanford IMT)
LGF Monitoring:
• When Iono rate ≥ 0.02 m/s: MQM Ramp detects <= 5 seconds
• When 0.01 ≤ Iono rate 0.02 m/s: CUSUM detects first.
• When Iono rate 0.01 m/s: No LGF detection
Airborne Monitoring:
• Assume only GMA Code Carrier Divergence with time constant of 200s
• Iono rate ≥ 0.035 m/s: 5 s
• Iono rate = 0.01 0.035 m/s: 200 – (rate – 0.01) × 8 × 103 s
• Iono rate 0.01 m/s: No detection
16 September 2005 43
Max Error at Memphis, Stationary Iono. Front (All SVs Healthy, LGF Monitoring, 7/18/05)
050
100150
200250
0
50
100
150
2000
2
4
6
8
10
Iono Slope (mm/km)
Max Error at Memphis, Stationary Front, LGF Monitoring, July 18,05
Iono Width (km)
Ma
x E
rro
r (m
)
•Sensitive to slope but not to width.
•The maximum error is 9.9 m.
16 September 2005 44
Geometry Screening via Reduced VAL for Memphis PSP Site (7/18/05 almanac, all one-SV-out cases)
• VPLH0 limit (to eliminate all errors
> 10 m) 3.13
• Resulting availability loss: 945/4060 = 0.2328
• For all-SV-healthy, same VPLH0 limit
gives availability loss of 24/145 = 0.1665
16 September 2005 45
2 2.5 3 3.5 4 4.5 52
3
4
5
6
7
8
9
10
VPL-H0 (m)
Ma
x V
ert
ica
l E
rro
rs (
m)
Vertical Error vs VPL-H0, Memphis, 071805, All SVs, Stationary Front, LGF
Although no error exceeds 10 m for
all-SV-healthy case; since VPLH0 must
be 3.13 to protect all 1-SV-out
scenarios, the resulting
availability loss for all-SV-healthy case becomes 24/145 =
0.1665
Geometry Screening via Reduced VAL for Memphis PSP Site (7/18/05 almanac, all-SV-healthy case)