The Impact and Mitigation of Ionosphere Anomalies on Ground-Based Augmentation of GNSS

The Impact and Mitigation of Ionosphere Anomalies on Ground-Based

Augmentation of GNSS

Sam Pullen, Young Shin Park, and Per Enge

Stanford University

[email protected]

12th International Ionospheric Effects Symposium (IES 2008)

Alexandria, Virginia

Session 4A, Paper #6 14 May 2008

14 May 2008 Impact and Mitigation of Ionosphere Anomalies on GBAS 2

Significance of Ionosphere Spatial Decorrelation in LAAS/GBAS

vig

VPL

VHF Data Broadcast

LAAS Ground Facility

Vertical Protection Level (VPL)

Ionospheric delay

Broadcast Standard Deviation (Sigma) of Vertical Ionosphere Gradient

Vertical Alert Limit (VAL)

VAL

Source: Jiyun Lee, IEEE/ION PLANS 2006


Severe Ionosphere Gradient Anomaly on 20 November 2003


Moving Ionosphere Delay “Bubble” in Ohio/Michigan Region on 20 Nov. 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 UT)

Sla

nt Io

no D

elay

(m

)S

lant

Ion

o D

elay

(m

)

Sharp falling edge; slant gradients 250 – 330 mm/km

Initial upward growth; slant

gradients 60 – 120 mm/km

Data from 7 CORS stations in N. Ohio and S. Michigan

“Valleys” with smaller (but anomalous) gradients


Validation of High-Elevation Anomaly Reported by FAATC (SVN 38, ZOB1/GARF, 20/11/03)

20.9 20.95 21 21.05 21.1

0

50

100

150

200

250

300

350

400

Time (hour of 11/20/2003)

Iono

Slo

pe (m

m/k

m)

ZOB1 and GARF, Slopes Comparison; SV38

DF Slope

L1 Slope

Maximum slope from L1-only data 412.8 mm/km


-102 -100 -98 -96 -94 -92 -90 -88 -86 -84 -82 -80

38

40

42

44

46

48

KNTNSIDN

MCON

PKTN

KNTNSIDN

MCON

PKTN

IPP Tracks of (Low-Elevation) GPS SVN 26 (20 Nov. 2003; 20:30 ~ 21:30 UT in OH-MI)

Satellite Direction of Motion

Iono. Frontat 21:00


20.6 20.7 20.8 20.9 21 21.1 21.2 21.3 21.4 21.50

5

10

15

20

25

30

35

40

45

50Iono Delay for: SV 26; Elevation:10.06890 - 12.0780

Time(Hour of Day); Date:2003 11 20; Filename: sv26

Sla

nt I

ono

Del

ay (

m)

SVN 26 Slant Delays Observed at WOOS, FREO, LSBN, and GARF

FREO

LSBN

WOOS

GARF

• Sufficient similarity between the two sets of ionosphere delays

exists

• Lines-of-Sight from FREO and WOOS are within the bulk of the “enhanced” ionosphere gradient

GUST


20.8 20.9 21 21.1 21.2 21.3 21.4-50

0

50

100

150

200

250

300

350

400

Hour of 11/23/03

Iono

Slo

pe,

mm

/km

WOOS and GARF Slope comparision

Severe Ionosphere Slope Validated with L1 data WOOS and GARF, SVN 26, 20 Nov. 2003

DF Slope

L1 Slope

• Maximum Validated Slope: ~ 360 mm/km

• This observation window is very close to the time that peak ionosphere gradients were observed on higher-elevation satellites.

Estimated Slope using L1 Code-minus-Carrier Data

L1-L2 slope

L1-only slope


Ionosphere Anomaly Front Model:Potential Impact on a GBAS User

Simplified Ionosphere Front Model: a ramp defined by constant slope and width

Front Speed 200 m/s

Airplane Speed ~ 70 m/s

(synthetic baseline due to smoothing ~ 14 km)

Front Width 25 km

GBAS Ground Station

Front Slope 425 mm/km

LGF IPP Speed 200 m/s

Stationary Ionosphere Front Scenario: Ionosphere front and IPP of ground station IPP move with same velocity.

Maximum Range Error at DH: 425 mm/km × 20 km = 8.5 meters

Max. ~ 6 km at DH


Resulting Revised (and Simplified) Ionosphere Anomaly Threat Model for CONUS

(note: plot not precisely to

scale)

5 15 30 45 65 90

SV elevation angle (deg)

Sla

nt

ion

o.

gra

die

nt

bo

un

d (

mm

/km

)

100

200

300

375

425Flat 375 mm/km

Flat 425 mm/kmLinear bound:

ybnd (mm/km) = 375 + 50(el15)/50

Also bounds on:

Front speed wrt. ground: ≤ 750 m/s

Front width: 25 – 200 km

Total differential delay ≤ 50 m


Semi-random Results for Memphis LGF at 6 km DH

0 5 10 15 20 25 30 35 40 450

0.02

0.04

0.06

0.08

0.1

0.12

0.14

User Vertical Position Error (meters)

PD

F

Worst-case error, or

“MIEV”, is 41 m

Most errors are exactly zero due to, e.g., CCD detection and exclusion before

anomaly affects users, but all zero errors have been removed from the histogram.

RTCA-24 Constellation; All-in-view, all 1-SV-out, and all 2-SV-out subsets included; 2 satellites impacted simultaneously by ionosphere anomaly


Simplified Flow Chart for Real-Time Inflation

LAAS Ground Facility (LGF) Real-Time Geometry Screening

SV almanac and current

time Subset Geometry Determination

(N2 constraint)

Worst-Case Ionosphere Error

Determination

Ionosphere Anomaly

Threat Model

Airport Approach

Layout and Ops. Limits

Approach Hazard Assessment

Iterative Sigma/P-Value Parameter

Inflation

Do Any Unsafe Subsets Exist?

Yes

Compare MIEV to Ops. Limits for Available

Subset Geometries

No

Inflated pr_gnd, vig, and/or P-

values

Approved Sigmas/P-Values for Broadcast by VDB

References: J. Lee, et al., Proceedings of ION GNSS 2006

S. Ramakrishnan, et al., Proceedings of ION NTM 2008

LGF acts to make potentially unsafe user geometries unavailable.


MIEV for Memphis at 6 km Prior to Inflation

0 50 100 150 200 250 30010

15

20

25

30

35

40

45

MIE

V (

m)

WC error from histogram on slide 11

OCS Error Limit at DH

Time Index (5-min updates)


0 50 100 150 200 250 3001

1.5

2

2.5

3

3.5

4

4.5

5

5.5x 10

-4

Infl

ate

d P

-va

lue

s (m

/m)


Inflated P-values for Memphis at 6 km from LGF (PA = 0.17, PB = 0.27 m/km)


0 50 100 150 200 250 30014

16

18

20

22

24

26

28

30

MIE

V (

m)

Inflated MIEV (m)

OCS Limit (m)

MIEV for Memphis at 6 km after P-Value Inflation



0 50 100 150 200 250 3002

3

4

5

6

7

8


Pro

tect

ion

Le

vel (

m)

Uninflated VPL H0

Uninflated VPL eph

Inflated VPLeph

Protection Levels for Memphis at 6 km from LGF

Significant margin (> 2.5 meters) relative to 10-meter FASVAL at DH


Summary

• Ionosphere anomaly threat model for CONUS has been developed based on validated ionosphere storm gradients discovered since April 2000

• Maximum ionosphere-induced errors are caused by worst-case extrapolation of events on 20 Nov. 2003

• Given lack of ground system observability, mitigation strategy is to inflate broadcast parameters to exclude potentially unsafe subset geometries from use– This requires GBAS ground station to vary inflation

parameters in real time.

– Two methods (using inflated pr_gnd and P-values that vary across satellites, as shown here, and using inflated vig values) have been demonstrated that retain near-100% availability for RTCA-24 constellation and typical actual constellations (with no satellite outages).


Backup Slides

• Backup slides follow…


Outline

• Nominal Ionosphere Spatial Decorrelation over Short Baselines

• Anomalous ionosphere spatial decorrelation

– Examples from October and November 2003 storms in CONUS

– Worst-case range-domain errors for GBAS users

• Ionosphere Spatial Anomaly “Threat Model”

• Real-Time GBAS/LAAS Threat Mitigation

– Impact of code-carrier divergence (CCD) monitoring

– Simulation of worst-case vertical position errors

– “Geometry Screening” by broadcast parameter inflation

– Resulting impact on CAT I GBAS user availability


0 20 40 60 80 100 120 140 160 180 2000

1

2

3

4

5

6

IPP Separation Distance (km)

vi

g o

verb

ou

nd

(m

m/k

m)

vig

Estimates using JPL-processed CORS data

07/02/00, Dst: 2

09/11/02, Dst: -78

07/26/04, Dst: -9407/27/04, Dst: -197

11/09/04, Dst: -223

11/10/04, Dst: -289

vig Overbound Results from Station Pair Method

JPL post-processed CORS “truth” data

Insufficient number of samples to obtain reliable statistics

Solid lines

show vig +

|vig|.

Inflation factors are 2.2 ~ 4.1


Iono. Anomaly Event Verification Methodology

L1 code and carrier

L1 code and carrier, L2 carrier

“Raw” IGS/CORS Data

JPL Ionosphere “Truth” Processing

Ionosphere Delay Estimates

Find Maximum Apparent Gradients

(Station Pair Method)

Screening Process (Automated)

Database of Extreme Gradients Erroneous Receiver

Steps and L1/L2 Biases Removed Investigate

Remaining Points

Remove “Questionable”Observations

Estimate Gradients from L1 code-minus-carrier

Output Database of Maximum Gradients

Compare L1/L2 and L1 CMC Gradient Obs.

“Validated”Max. Gradient

Database

Note: CORS stations in CONUS are typically 30 – 100 km apart.


LGF

Iono Front

Runway

i (aircraft)

i (LGF)Vi,Proj

j (aircraft)Vj,Proj

j (LGF)

Vfront

Worst-Case Ionosphere Front Scenarios

Generate two sub-cases for each SV pair (i, j) “i worst” and “j worst”

“i worst” => Apply worst range error to SV i and resulting error to SV j

“j worst” => Apply worst range error to SV j and resulting error to SV i


Proposed Worst-Case Ionosphere Error Limits

Distance from LGF (km)

Wor

st-C

ase

Iono

sphe

re E

rror

Allo

wed

(m

)

6 7 8 9 10 11 12 13 1420

40

60

80

100

120

140

160

99% TSE

95% TSE

99.9% TSE

28-meter constraint

at 6 km

DH location


Notes on Real-Time Ground-System Parameter Inflation

• Real-time parameter inflation is fundamentally providing integrity in the position domain

• The real-time satellite-specific P-value inflation method shown here achieves 100% availability for all-in-view geometry (for RTCA-24 constellation) at 4 of 5 airports tested (MEM, DEN, DFW, MCO)– One exception (DCA) is shown on following slide

– For P-value inflation approach, inflation required for 6-km separation dominates that required for larger separations

• However, since thousands of subset geometries must be checked at every real-time parameter update interval, simpler methods that retain acceptable availability are also of interest

Documents

The Impact and Mitigation of Ionosphere Anomalies on Ground-Based Augmentation of GNSS