Will COSMIC Data Improve Ionospheric Specification 17 October 2006

1
Will COSMIC Data Improve Ionospheric
Specification?17 October 2006
Capt Craig Baker Space Vehicles Directorate Air
Force Research Laboratory
2
Acknowledgements

• USC/JPL GAIM (Tony Mannucci, Lukas Mandrake,
  Brian Wilson)
• Data
• Ground Data (JPL)
• CHAMP GPS Data (GFZ, JPL/Attilla Komjathy)
• Beacon Data (ARL/Gary Bust, Trevor Garner)
• Millstone ISR Data (Haystack/Larisa Goncharenko)
• Ionosonde Data (UMLCAR NGDC)
• COSMIC Data (NSPO UCAR)
• AFRL (Odile de la Beaujardiere, Bill Borer,
  Dwight Decker)

3
Introduction

• The ionosphere affects communication, navigation
  and surveillance systems
• Refraction (HF communications, satellite
  tracking)
• Scintillation (GPS, satellite communication)
• Good ionospheric specification vital to
  ionospheric effects mitigation and forecasting
• Climatological models capture mean behavior but
  not short-term variations (weather)
• Assimilation models attempt to improve on
  climatological models by making adjustments
  consistent with available measurements
• The question Does more satellite-based data
  (e.g., COSMIC) yield a better specification?

4
Overview

• USC/JPL GAIM Introduction
• Quick look at GAIM with assimilated data
• Slant TEC from ground GPS receivers
• Slant TEC from LEO GPS receivers (radio
  occultation)
• Relative TEC from multi-frequency beacon
  receivers
• Combined data
• Limited comparison with ionosonde and incoherent
  scatter radar (ISR) data at Millstone Hill

5
Global Assimilative Ionospheric Model (GAIM)

• Physics-based model of the ionosphere (O)
  developed by USC/JPL
• Neutral density, composition and wind from
  climatologically-driven models (e.g., MSIS)
• Kalman filter approach to assimilation of
  measurements
• Can assimilate variety of data
• Slant TEC
• Ground-based dual-frequency GPS receivers (e.g.,
  IGS network)
• LEO GPS receivers (e.g., CHAMP and COSMIC)
• Relative slant TEC
• Beacon receivers (e.g., ARL CIDR receivers)
• Electron density
• In-situ sensors (e.g., CHAMP PLP, DMSP SSIES)
• Ionosondes
• User-defined confidence for each data type

6
Data Type 1 Slant TEC from Ground GPS Receivers

• Slant TEC derived from ground-based
  dual-frequency GPS receivers
• Continuous coverage of large regions, no coverage
  of other regions
• Figure shows ionospheric pierce points of slant
  rays for 30 minute period
• Color scale shows equivalent VTEC

7
GAIM with Slant TEC from Ground GPS Network

• Assimilated slant TEC from network of ground GPS
  receivers
• Figure shows Vertical TEC, 13 Aug 05, 1800 UT
• Top No data assimilation
• Middle Ground GPS assimilated
• Bottom Difference (upper - lower)
• Significant differences most notable near large
  gradients, e.g., near anomaly peak

8
Data Type 2 Slant TEC from LEO GPS Receivers
(Occultation)

• Slant TEC derived from satellite (LEO)
  dual-frequency GPS receivers (e.g., CHAMP,
  COSMIC)
• Figure shows tangent points for links during one
  (green), two (blue) and three (magenta) hours
  prior to 1800 UT latest marked with red
• Periodic global coverage from single satellite

9
Combined Data GAIM with Ground GPS and LEO GPS

• Assimilated CHAMP GPS and ground GPS data
• Figure shows Vertical TEC at 13 Aug 05, 1800 UT
• Top Ground GPS assimilated
• Middle Ground GPS and CHAMP GPS
• Bottom Difference (upper - lower)
• As expected, differences follow satellite
  (receiver) track
• COSMIC dramatically increases number of GPS
  occultation measurements

10
CHAMP GPS Occultation Movie

• Movie spanning 13 Aug 05 in 15-minute time steps
• Top GAIM vertical TEC with ground and CHAMP GPS
  data assimilated
• Middle Difference between model using only
  ground GPS and model using both ground and CHAMP
  GPS
• Bottom CHAMP tangent trace during 1 hour prior
  to model time step
• Differences localized (near CHAMP) and fade over
  a few time steps

11
Data Type 3 Relative Slant TEC from Beacon
Receivers

• Relative TEC derived from ground based receiver
  of dual-frequency beacon (e.g., Transit, COSMIC
  or C/NOFS) for single pass (continuous arc)
• Quasi-continuous regional coverage COSMIC will
  help fill in time gaps
• Figure shows ionospheric pierce points for three
  receivers (marked with ) during one (blue), two
  (green) and three (red) hours prior to 2006 08 13
  1800 UT

12
GAIM with Relative TEC (Beacon Data)

• Assimilated relative TEC from 3 ground receivers
  in eastern US
• Data mostly from a few Transit beacons
• COSMIC beacons significantly increase available
  data
• Figure shows vertical TEC at 13 Aug 05, 1800 UT
• Top No data assimilation
• Middle Relative TEC assimilated
• Bottom Difference (upper - lower)
• Differences limited to region near receivers (as
  expected)

13
Combined Data GAIM with Ground GPS and Relative
TEC

• Assimilated relative TEC (regional) and ground
  GPS data (global)
• Figure shows vertical TEC at 13 Aug 05, 1800 UT
• Top Ground GPS assimilated
• Middle Ground GPS and relative TEC
• Bottom Difference (upper - lower)
• Very small differences (note scale), due to good
  agreement between data sets

14
Vertical Slice along Millstone Hill Longitude

• Figure shows electron density (Ne) for vertical
  slice at fixed longitude (285 deg E) at 13 Aug
  05, 1800 UT
• Top Ground GPS assimilated
• Second Ground GPS, CHAMP GPS, and relative TEC
  assimilated
• Third Absolute difference (upper - lower)
• Bottom Percent difference
• Complicated effect on 3-d specification

15
Diurnal NmF2 Profile GAIM with Ground and CHAMP
GPS

• Assimilation of continuous ground GPS data allows
  GAIM to update at each time step
• CHAMP GPS add smaller, transient adjustment at
  time of pass
• Figure shows NmF2 vs UT at Millstone Hill, MA, 13
  Aug 05

16
Diurnal NmF2 Profile GAIM with Ground and COSMIC
GPS

• NmF2 vs UT at Millstone Hill for 26-28 Jun 06
• Each panel shows GAIM climate, ground GPS
  assimilation, and ground plus COSMIC GPS
  occultation assimilation
• As with CHAMP, COSMIC adjustments are transient

17
COSMIC GPS Occultation Movie

• Six-hours on 28 Jun 06
• COSMIC constellation still clustered
• Situation will improve as constellation moves to
  final configuration
• Model developer challenge to address local
  transients

18
How to tell whether the added data improves the
results?

• Ionosonde as ground truth
• Provides electron density up to F peak
• Continuous operation at tens of sites around
  globe
• Incoherent scatter radar (ISR)
• Can provide full altitude profile of electron
  density
• Limited to handful of sites and intermittent
  operation
• COSMIC benefit should be greatest in data-poor
  regions (e.g., over oceans)
• Hard to get ground truth in data-sparse regions

19
Simulate Data-Sparse Region

• Simulate data-sparse region near ground truth
  site by excluding ground GPS data near Millstone
  Hill
• NmF2 very near climatology when data excluded

20
Diurnal Profile of NmF2 Comparison with Ionosonde

• Allow COSMIC GPS data to fill hole in ground
  GPS and compare to ionosonde
• Ground GPS improves comparison for much of
  profile
• Large COSMIC adjustments appear to agree with
  ionosonde, but transient effect of data makes
  comparison difficult
• Quality of ionosonde data dependent on skill of
  analyst (computer or human)

21
Ionosonde Data

• Autoscaling programs extract characteristic
  parameters quickly and routinely, but with
  variable effectiveness
• Hand-scaling time-consuming, but (often) more
  reliable
• Community effort to improve auto-scaling
  reliability

22
Altitude Profile GAIM with Ground and COSMIC GPS

• Comparison with Millstone Hill ISR electron
  density profile on 28 Jun 06, 01 UT
• No COSMIC data at this time, but ground GPS
  allows model to make significant adjustment in
  peak
• Ground-based TEC measurement provides limited
  information about vertical structure

23
Altitude Profile GAIM with Ground and COSMIC GPS
(Example 2)

• Comparison with Millstone Hill ISR electron
  density profile on 28 Jun 06, 2145 UT
• COSMIC data provides information about vertical
  structure and gives good agreement with ISR at
  this time step
• Ground GPS data improves profile relative to
  climatology
• Combined data gives intermediate result
• Model developers challenged to combine data in
  most intelligent fashion

24
Summary

• Limited look at USC/JPL GAIMs use of ground GPS,
  LEO GPS and relative TEC data
• Does COSMIC data improve ionospheric
  specification?
• We can expect that COSMIC data, especially GPS
  occultation at final orbital configuration, will
  improve ability of assimilation models to
  represent ionosphere in data-sparse regions
• Much more validation needed
• JPL tuning GAIM to best handle COSMIC data
• Need to understand complexities of transients and
  interaction of different data types
• Quality ground truth data still limitation

25
BACKUPS
26
Closer Look at Differences

• Large differences due to shift in location of
  steep gradients e.g., location of anomaly peaks

27
GAIM with All Three Data Sets

• Assimilated ground GPS, CHAMP GPS and relative
  TEC data
• Figure shows vertical TEC at13 Aug 05, 1800 UT
• Top Ground GPS assimilated
• Middle All data types (ground GPS, CHAMP GPS,
  relative TEC)
• Bottom Difference (upper - lower)
• Effect of relative TEC negligible due to good
  agreement with ground GPS data
• Beacon data has other uses

28
Comparison of Differences Along Millstone
Longitude

• Effect of combining ground GPS data with
• CHAMP GPS (top),
• Relative TEC (middle),
• And both (bottom)
• Figure shows Lat-Alt slice of differences in Ne
  for each combination, 13 Aug 05, 1800 UT
• Note different scale for relative TEC case
  (middle)
• When combined with CHAMP GPS data, relative TEC
  data had negligible effect

29
Vertical Slice along Millstone Hill Latitude

• Figure shows Ne vs Longitude and Altitude at 44
  deg N latitude, 13 Aug 05, 1800 UT
• Top Ground GPS assimilated
• Second Ground GPS, CHAMP GPS, and relative TEC
• Third Absolute difference (upper - lower)
• Bottom Percent difference
• Effect of CHAMP pass visible in both hemispheres

30
GAIM with Ground and COSMIC GPS Diurnal Profiles

• NmF2 vs UT at Millstone Hill for three days 26-28
  Jun 06
• Is it better?