New Zealand Vertical Datum - IAG Commission 1

THE VERTICAL DATUMS OF
AUSTRALIA AND NEW ZEALAND
Sten Claessens 1 Will Featherstone 1 Matt Amos 2 Mick Filmer 1
Christian Hirt 1 Jon Kirby 1
1 The Western Australian Centre for Geodesy &
The Institute for Geoscience Research
Curtin University of Technology
2 Land Information New Zealand
REFAG2010, Marne-la-Vallée, France
Session 5: Definition and Establishment of Vertical Reference Systems
6 October 2010

INTRODUCTION (1/2)
• Australia and New Zealand share strong ties in the development of
geodetic infrastructure, e.g. through the joint Intergovernmental
Committee on Surveying and Mapping (ICSM) and joint research
• However, both countries have recently made different choices
relating to their vertical datums
New Zealand has established a new vertical datum
Australia has decided to maintain its existing vertical datum for
the foreseeable future

INTRODUCTION (2/2)
• Recent vertical datum activities:
•New Zealand Vertical Datum (NZVD) 2009 based on gravimetric
NZGeoid09 released
•AUSGeoid09 combined gravimetric-geometric quasigeoid
established – aids transformation from ITRF to Australian Height
Datum (AHD)
•Experimental version of new ‘scientific’ Australian vertical datum
(AVD:Ev1) created through constrained re-adjustment of
Australian National Levelling Network (ANLN)

NEW ZEALAND VERTICAL DATUM 2009 (1/5)
Situation before 2009:
• 13 major local vertical datums
(LVDs) with one tide gauge
each as origin
• Origins do not represent
current mean sea level
• Levelling data was never
adjusted nationally
The new New Zealand Vertical
Datum (NZVD) 2009 is based
on a new quasigeoid model,
NZGeoid09

NEW ZEALAND VERTICAL DATUM 2009 (2/5)
Iterative quasigeoid datum unification scheme
• Quasigeoid
compared to
GNSS-levelling in
each LVD
• LVD offsets
applied to gravity
anomalies
• Quasigeoid
recomputed

NZGeoid09 specs
NEW ZEALAND VERTICAL DATUM 2009 (3/5)
Resolution 1’
Reference model EGM2008
Reference
degree/order
2160
Kernel FEO
Modification
degree
40
Cap size 2.5˚
Fitted to GNSSlevelling?
RMS differences to
GNSS-levelling
Data sources:
• EGM2008
• DNSC08GRA
• 40,737 land gravity
observations
• 56 m DEM
• 1422 GNSS-levelling points
No
0.062 m
(m)
Residual quasigeoid heights with respect to EGM2008

NEW ZEALAND VERTICAL DATUM 2009 (4/5)
• Offsets between local vertical datums (LVDs) were determined
through the iterative quasigeoid computation process
• Range of offsets between NZGeoid09 and LVDs: -0.06 m to -0.49 m
• Fit to LVD offsets from precise levelling: ±0.064 m
Local vertical
datum
Offset
(m)
Std Dev
(m)
Local vertical
datum
Offset
(m)
Std Dev
(m)
One Tree Point -0.06 0.03 Nelson -0.29 0.07
Auckland -0.34 0.05 Lyttleton -0.47 0.09
Moturiki -0.24 0.06 Dunedin -0.49 0.07
Gisborne -0.34 0.02 Dunedin-Bluff -0.38 0.04
Taranaki -0.32 0.05 Bluff -0.36 0.05
Napier -0.20 0.05 Stewart Island -0.39 0.15
Wellington -0.44 0.04

NEW ZEALAND VERTICAL DATUM 2009 (5/5)
• NZGeoid09 forms the reference surface of the New Zealand Vertical
Datum (NZVD) 2009
• New Zealand was one of the first countries to adopt a geoid-based
national vertical datum
• NZVD2009 implemented by publication of NZVD2009 heights for
existing marks
• NZVD2009 is actively adopted by users and appears to be exceeding
many users’ expectations
• LINZ is currently considering improvements to NZVD2009 through
collecting airborne gravity

AUSGEOID09 / AHD (1/4)
• Like New Zealand, Australia has recently established a new
quasigeoid model: AUSGeoid09
• AUSGeoid09 will be officially released in the coming months
• The ICSM has resolved to maintain the Australian Height Datum
(AHD) for the foreseeable future, despite well-documented
deficiencies
• AUSGeoid09 contains a geometric component to model the offset
between the gravimetric-only quasigeoid and the AHD

AUSGeoid09 specs
Resolution 1’
Reference
model
Reference
degree/order
EGM2008
2160
Kernel FEO
Modification
degree
40
Cap size 1˚
Fitted to
levelling?
Fitting
technique
Correlation
length
AUSGEOID09 / AHD (2/4)
Yes
LSC
75 km
(m)
Residual quasigeoid heights with respect to EGM2008

AUSGEOID09 / AHD (3/4)
Differences between gravimetric-only AUSGeoid09 and GNSS-AHD heights
(m)
• ~ 1 m northsouth
trend in
AHD (~23 mm/
100 km)
• Regional
distortions up to
~ 0.5 m
• Shows
magnitude of
geometric
component of
AUSGeoid09

AUSGEOID09 / AHD (4/4)
The geometric component of AUSGeoid09:
• Computed by Nicholas Brown and colleagues at Geoscience
Australia
• A posteriori fitting by least-squares collocation (LSC) in crossvalidation
mode based on extended dataset of 6,794 AHD-GNSS
points
• Gaussian analytical covariance function determined empirically;
noise based on STD of post-processed GNSS ellipsoidal heights
• Correlation length of 75 km provided smallest RMS misfit in crossvalidation
of ±30 mm

AUSTRALIAN VERICAL DATUM: EV1 (1/6)
• Despite the ICSM decision to retain the AHD as Australia’s vertical
datum for the foreseeable future, research into an improved vertical
datum for scientific purposes has continued
• An experimental version of a new ‘scientific’ Australian vertical
datum has been computed (AVD:Ev1)
• AVD:Ev1 is based on a re-adjustment of the Australian National
Levelling Network (ANLN), supplemented with:
•The gravimetric-only version of AUSGeoid09
•GNSS data at ANLN junction points
•CARS2006 SSTop model
•Tide gauge data

AUSTRALIAN VERICAL DATUM: EV1 (2/6)
Australian National Levelling Network (ANLN)
Yellow: first order
Green: second order Red: third order one-way
Purple: third order Blue: unspecified
Dark green: fourth order
• >100,000 km of
levelling, primarily
of 3 rd order
• Many loops
>1000 km
• ~8% of 1366
loops exceed
closure tolerance
• Largest loop
misclosure 0.93 m

• ODRW based on
Baarda’s data
snooping, MDB and
loop closures
• No improvement in
fit to AUSGeoid09
before and after
ODRW
AUSTRALIAN VERICAL DATUM: EV1 (3/6)
Differences STD (m)
Pre-ODRW minus
post-ODRW
Pre-ODRW minus
AUSGeoid09
Post-ODRW minus
AUSGeoid09
Outlier detection and re-weighting (ODRW)
0.091
0.180
0.200
(m)
Differences between re-adjusted MC ANLN before and
after outlier detection and re-weighting (ODRW)

AUSTRALIAN VERICAL DATUM: EV1 (4/6)
Sea Surface Topography around Australia from CARS2006
• Most of the N-S trend in the
AHD can be attributed to Sea
Surface Topography (SSTop)
• N-S trend (mm/100km):
CARS2006 at tide
gauges
EGM2008 and GNSS-
AHD
Figure 8.2. Filmer
CARS2006 SSTop
23.7
22.3
ANLN lev
(m)

AUSTRALIAN VERICAL DATUM: EV1 (5/6)
Adjustment strategy:
• CARS2006 fixed to MSL at Albany tide gauge and subsequently
corrected for bias with AUSGeoid09 (0.132 m)
• Tide gauge/CARS2006 data introduced in adjustment with STD of
±0.005 m
• AUSGeoid09 at 268 GNSS-levelling points (selected from set of
1042) introduced in adjustment with STD of ±0.010 m
• ODRW process repeated (2 AUSGeoid09 points removed as
outliers; 11 added to robustify areas with multiple levelling errors)
• Result validated by comparison to GNSS-levelling points not used

Validation of AVD:Ev1
AUSTRALIAN VERICAL DATUM: EV1 (6/6)
Differences STD (m)
AUSGeoid09 minus
AVD:Ev1
AUSGeoid09 minus
AHD
AHD minus
AVD:Ev1
0.098
0.207
0.186
• No trend is discernable
in differences between
AUSGeoid09 and
AVD:Ev1
• AVD:Ev1 provides a
substantial improvement
over AHD
Differences between AHD and AVD:Ev1
(m)

SUMMARY (1/1)
• Australia and New Zealand have both recently computed a new
quasigeoid model
• New Zealand has adopted NZGeoid09 as the reference surface of
its new vertical datum, NZVD2009
• Australia has ‘warped’ the gravimetric version of AUSGeoid09 to
better fit the Australian Height Datum
• An experimental version of a new Australian vertical datum has
been created through a combined adjustment of the ANLN,
gravimetric-only AUSGeoid09, GNSS, tide gauge and SSTop data

PAPERS IN PRESS (1/1)
• Claessens SJ, Hirt C, Amos MJ, Featherstone WE, Kirby JF (in
press) The NZGeoid09 model of New Zealand. Survey Review.
• Featherstone WE, Kirby JF, Hirt C, Filmer MS, Claessens SJ, Brown
NJ, Hu G, Johnston GM (in press) The AUSGeoid09 model of the
Australian Height Datum. Journal of Geodesy.