AFREF Computational Methodologies

1
AFREFComputational Methodologies
Rui Fernandes
DI-UBI, Univ. Beira Interior, Covilhã, Portugal.
Instituto Geofísico D. Luíz, Lisboa,
Portugal. DEOS, Delft Univ. of Technology, The
Netherlands.
2
Towards the implementation of AFREF results of
case-studies for the computation of the reference
solution
R.M.S. Fernandes H. Farah L. Combrinck H.
Khalil S. Leinen
UBI, IDL, CGUL, Portugal DEOS, The
Netherlands RCMRD, Kenya HartRAO, South
Africa NRIAG, Egypt IPG, TUD, Germany
3
Way to go...
Non-uniform systems
Uniform system

GNSS stations ITRS

• GNSS stations will realize and maintain AFREF
• AFREF will be based on ITRFxx

4
Relation with IGS

• IGS stations will be the backbone of the AFREFxx
  realization by providing the link to ITRFxx

5
Tectonic Plates

• PB2002 Bird, 2003 - 52 tectonic units

6
Estimated Velocities ITRF2005
7
AFREF08 AFREF09Study Cases

• Set of coordinate positions for a number of GNSS
  stations distributed by the entire African
  continent.
• On contrary of ITRFxx, AFREFxx must be static
  reference frames.
• Therefore, one single epoch must be selected as
  reference.
• AFREF08 Reference 05 May 2008 (GPSweeks 1477
  1478)
• AFREF09 01 January 2009 (GPSweeks 1512 1513)
• Approach A using the first week as reference and
  the second for comparison and evaluation of the
  reference solution (done at coordinate level and
  internal repeatabilities).
• Approach B using the two weeks to compute the
  reference solution (internal repeatabilities).

8
Roadmap for the AFREFxx solutions

• Selection of the AFREF stations
• What was (and will be?) an AFREF station?
• Continuously operating
• No end of operation foreseen
• Reliable Internet access
• Data transfer after few hours of acquisition
• Data Publicly available
• No restrictions to data distribution
• Installation according to IGS standards
• stable monument, self-centering mounting device,
  equipment recognized, any local ties very well
  determined, etc...
• Uniform Distribution as good as possible
• Current objective no more than 1000Km between
  AFREF core stations

9
How to implement the network?

• Active way
• Install stations to densify the AFREF network
• which can also be used for other applications
• Passive way
• Classify stations already installed in the
  framework of other projects as AFREF stations
• necessary cooperation/support with these projects

• Interaction with other applications needed

10
Site distribution
Scientific Applications - Crustal Dynamics -
Tectonic motions (long term / low frequency) -
Earthquake/Volcano monitoring (real time / high
frequency) - Tsunami Earlier Warning Systems
(TEWS) - Sea Level Monitoring / Vertical Datum
Definition - Link to Tide-gauges (TEWS
again) - Meteorological and Climate studies -
Precipitable Water Vapor (PWV) estimations -
Communications - Total Electron Contents (TEC)
maps - Real-time Scientific applications -
e.g., Precise orbit estimation
11
Site distribution
Technical Applications - Definition and
Maintenance of National Grid - Densification of
permanent stations at national level. -
Mapping - Pos-processing - Real-Time
(differential corrections) - Navigation -
Real-Time (differential corrections) -
Augmentation systems (EGNOS)
12
Site distribution
Political constraints - 61 Territories -
Largest Sudan (2 505 810 Km2) - Smallest
Melilla, Spain (12 Km2) Gambia (11 300
Km2)
Every territory (definitively, every country)
should have a station part of AFREF
13
Current Situation
Survey of CGPS sites 116
14
South AfricaTrignet IGS
Too many fiducial stations at continental scale
15
Final Selection AFREF08
Position solutions was computed for a total of 47
stations.
16
Final Selection AFREF09
Some new additions but also
17
Final Selection AFREF09
Some some stations not avalable. Position
solutions was computed for a total of 43 stations.
18
IGSProblems with network maintenance
Only 13 stations with available data for the
desired period
23 stations officially part of IGS network
19
AFREF08 AFREF09 solutions

• 3 Independent Solutions using 3 Different
  Software Packages
• RCMRD (Kenya) IDL (Portugal) used GIPSY
• HartRAO (South Africa) used GAMIT
• NRIAG (Egypt) TUD (Germany) used Bernese
• 2 Different Mapping approaches to align to
  ITRF2005
• RCMRD/IDL used a global set of reference mapping
  stations
• HartRAO and NRIAG/TUD used a regional set of
  reference mapping stations
• Unique combined solution
• Dedicated scripts based on GIPSY tools
• Differences allowed us to detect errors due to
  metadata/software packages/models used.

20
Software GIPSY GAMIT BERNESE

• GIPSY (undifferenced data)
• Precise Point Positioning strategy
• Each site individually processed
• JPL orbits clock corrections
• Use of ambizap algorithm to solve for ambiguities
• GAMIT BERNESE (double differences)
• Free Network strategy
• All station positions estimated in one inversion
• a-priori IGS orbits (adjusted)

21
Mapping Approaches
RCMRD/IDL
NRIAG/TUD similar to HartRAO
22
AFREF08 ResultsComparison Global/Regional mapping
IGS sites
Solution 1..n
daily mapping
Positions of AFRICA stations w.r.t. ITRF2005 at
day 1..n
23
AFREF09 ResultsComparison Global/Regional mapping
IGS sites
Solution 1..n
daily mapping
Positions of AFRICA stations w.r.t. ITRF2005 at
day 1..n
24
Weekly Combination AFREF08 example
AFREF sites
25
Solution Combinations
AFREF sites
26
Comparison between the BERNESE GIPSY
AFREF09
AFREF08
27
The role of tectonics.
Africa contains two major tectonic plates (Nubia
and Somalia) plus some few minor tectonic blocks
(e.g. Victoria)
in Fernandes et al. 2004
in Stamps et al. 2008
28
Example Ethiopia case

• Stations located on the opposite sides of the
  East African Rift will move apart about 6-7 mm/yr
• This is a significant change in the relative
  positions that must be taken into account if
  someone decides to use stations located on both
  sides to further densify the network using data
  collected in the future

29
Internal consistency of AFREF08

• AFREF08 will be linked to the Nubian plate ?
  Stations located in this plate will be stationary
  (no motion).
• Reference stations located on other tectonic
  blocks will have a differential motion with
  respect to Nubia that should be taken into
  account when these stations are used for
  densification purposes.X(t) X(t0) V . t
• t sometime in futuret0 reference epoch (01
  May 2008)

After the computation of the position of a new
station at an epoch t, the angular velocity model
must be applied backwards in order to compute the
position of this station at epoch t0 (01 March
2008). This is only necessary for stations not
located on Nubia.
30
Summary

• AFREFxx will be estimated using a consistent set
  of stations distributed by the entire continent
  (plus some stations located in neighborhood
  regions).
• Almost all available CORS stations in Africa will
  be part of the AFREFxx solutions (the few
  exceptions are due to good reasons e.g.,
  Trignet).
• The coordinate positions will be computed by
  combining independent solutions using different
  software packages.
• AFREFxx will be formed by a set of coordinates
  linked to ITRFxx at a certain epoch together with
  an angular velocity model that will express the
  relative motions of the existing tectonic blocks
  in the continent with respect to the Nubian plate.

31
And the effort will continue...
THANK YOU