The Hunting of the
SNARF
Giovanni F. Sella
Seth Stein
Northwestern University
Timothy H. Dixon
University of Miami
"What's the good of Mercator's North
Poles and Equators,
Tropics, Zones, and Meridian Lines?"
So the Bellman would cry: and the
crew would reply
"They are merely conventional signs!
Ideal Rigid Plate
Motion described by a
rotation pole and angular
velocity
Points rotate about pole
along small circles and
their rate increases as sin 
No vertical motion
Real Rigid Plate
Internal deformation due to
platewide (e.g. ridge push)
regional (e.g. GIA) and
local (e.g. density
anomalies) effects causes
both horizontal and
vertical motion
Stable plate vs rigid plate
Geographically representative
distribution of sites
Stable plate (geologic)
Use geologic a priori criteria to exclude
sites not expected to be on stable plate
>100 km from plate boundaries,
seismicity, active faults (Avoid seismic
cycle effects)
Rigid plate (geodetic)
Minimize any plate wide effects e.g.
GIA, subsidence, intra plate
deformation. Avoid any local effects e.g.
fluid withdrawal or injection
Choice of GPS sites
Detecting GIA using GPS
Large vertical signal observed
~10 mm/yr
GPS vertical velocities
with respect to
ITRF2000
Regionally Coherent Vertical Velocity Pattern
Clear pattern of positive
velocities in and around
Hudson Bay that decreases
going southwards to zero
(hinge line), beyond which
velocities are initially negative
and then rise to near zero
All sites north of the hinge line
and those up to a 100 km south
of it are interpreted as being
GIA affected
Distribution of GPS Sites on North America with Seismicity
CGPS - Continuous GPS
24hrs, 365 days
EGPS - Episodic GPS
8-12hrs, 2-3 days, every 2-3 yrs
If plate is: rigid, good geographic
distribution and errors are accurate
2 =1
Rigid North America defined using 83
CGPS sites (black diamonds)
2=1.08
include 46 GIA (red circles) affected
sites 129 sites
2 =1.33
Horizontal Residual Velocities
Residual horizontal velocities after removing the motion of rigid North
America defined using an 83 site solution
Other sites:
Larger magnitude with a clear
pattern of northward and
southward directed motions
Rigid plate sites:
Small magnitude with a
random distribution
Coherent patterns in GPS Velocities
Vertical Velocities
Residual (intraplate) Velocities
GIA signal:
Vertical: Around Hudson Bay rapid uplift, slower subsidence south of Great Lakes
Horizontal: Outward near HB and Western Canada (secondary ice load)
Non-GIA signal
Central US, small coherent uplift with randomly oriented horizontal
Implication of GIA
Largest detectable motion within stable plate is GIA
Essential to account for it
How?
Remove GIA model predictions - Helps or hurts?
Omit GIA affected sites?
GIA: Many models, many predictions
We use
•
•
•
•
Earth assumed laterally homogeneous with seismically realistic
depth-varying density and elastic parameter profiles
ICE-3G ice loading history from the Last Glacial Maximum (18,000
yrs to the present)
120 km thick elastic lithosphere, upper mantle viscosity of 1021 Pa s
and lower mantle viscosity of 2 x 1021 Pa s (LM2) and 4.5 x 1021 Pa s
(LM4.5)
Load increases from 0 at 100,000 yrs to maximum at 18,000 yrs then
decreases in 1,000-year increments
Model
Predictions
Vertical: predicted uplift
(rebound) north of Great
Lakes, subsidence to
south, both models similar
Horizontal: Orientation of
vectors similar but large
difference in the far field
Vertical Velocities
Observed
Vertical
minus
Predicted
Vertical
LM4.5
Equals
If
perfect
match
then
should
be all LM 2
white
Good agreement between model & observations
In Hudson Bay LM 4.5 (higher mantle viscosity)
better fit
Test removal of GIA predictions from horizontal Vel.
GPS intraplate
Rigid North America
defined using 83 CGPS
2=1.08 (Better)
-5.0N,85.3N,0.195°/Myr
,1.0max,0.3min
include 46 GIA affected
sites 129 sites
2 =1.33 (worse)
LM2
Remove LM2
Predictions to 83
CGPS
2=1.11
LM4.5
Remove LM4.5
Predictions to 83
CGPS
2=1.41
Pole Positions wrt ITRF00
83 sites RIGID -4.96 N
-85.34 E
0.1953/Myr 1.0max 0.3 min
83 + 46 GIA
-2.23 N
-83.60 E
0.1989/Myr 1.9max 0.5 min
83 LM2 Corr
-4.44 N
-84.40 E
0.1996/Myr 1.0max 0.3 min
83 LM4.5 Corr -4.30 N
-81.79 E
0.2011/Myr 1.1max 0.3 min
GPS intraplate
LM2
LM4.5
Horizontal Velocities
GPS intraplate
LM2
LM4.5
Near field: GPS horizontal velocities are larger than LM4.5
Far field: GPS horizontal velocities are small like LM2 but random
LM4.5 very poor fit in the far field
Reasons:
Model assumptions of a laterally homogeneous earth are not valid
Ice load history may be incorrect affecting near field
GPS field too sparse in the North
Implications for SNARF
Largest residual (intraplate) signal across stable North America is GIA
GIA models generally predict observed GPS vertical velocities but do
poorly in the horizontal
Removing GIA model predictions does not reduce misfit
In contrast omitting sites that appear affected by GIA reduces misfit
Improved or other GIA model may do better
Pole Positions wrt ITRF00
0.1953/Myr 1.0max 0.3 min
83 sites RIGID -4.96 N
-85.34 E
83 + 46 GIA
-2.23 N
-83.60 E 0.1989 /Myr 1.9max 0.5 min
83 LM2 Corr
-4.44 N
-84.40 E
0.1996 /Myr 1.0max 0.3 min
83 LM4.5 Corr -4.30 N
-81.79 E
0.2011/Myr 1.1max 0.3 min