Transcript measuring and modeling

Measurement and Modeling of
Cryosphere‐Geosphere Interactions
J. Sauber, S. Luthcke, S-Ch Han, D. Hall, (NASA GSFC)
Collaborators:
R. Bruhn, R. Forster, M. Cotton, E. Burgess, J. Turrin (U of Utah)
B. Molnia, K. Angeli and ASC staff (USGS)
N. Ruppert (UAF, AEIS), R. Muskett, C. Lingle (UAF)
R. King, T. Herring (MIT), S. McClusky (ANU)
Campaign
GPS
ISLE
GRACE
Terra
ISLE
AB35
DON
GPS
AB35
PBO
GPS
•Cryosphere-Geosphere signals in campaign and PBO GPS observations
•Cryosphere change from GRACE
•Post-1964 earthquake gravity change & ice mass change trends
PyLith & other numerical models
MacBook Air
Dual Core
Mac Desktop
6 Cores
GRACE processing of L1B data:
Mac Cluster
~50 cores
DISCOVER (GSFC HEC)
Use ~150 cores
The GPS station positions for both
sites are predicted to move to the
north-northwest (NNW) and up due
to (steady) tectonic forcing (PCFC
relative NOAM)
1993-1995 surge interval versus
post-surge time period, 1996-2001:
• A higher uplift rate for ISLE. Site
is moving faster to the northwest.
Ice unloading to the south of the
site would cause uplift & north
directed motion.
Sauber and Molnia, 2004
• In contrast, the station DON is
undergoing vertical subsidence
during the surge but uplift
subsequently (as predicted for
tectonic loading). The rate of the
northward motion is lower during
the surge.
1993-1995 Bering Glacier Surge: A Solid Earth Geophysicist View
Represent 10s of meters of drawdown
(unloading) in surge reservoir and
thickening & terminus advancement in
receiving/terminus (loading) region as disc
loads
As a “local load”, use elastic half-space
model or radially stratified, gravitating Earth
model to predict vertical and horizontal
displacements [Farrell, 1972].
Elliott et al.,2013 estimates of horizontal site
velocities between 2005-2008:
DON: ~43 mm/yr @N30oW
ISLE: ~20 mm/yr @N14oW
Predicted Vertical Displ.
ISLE 38 mm uplift
DON -42 mm
Sauber et al., 2000
Surge Related
Predicted Horizontal Displ.
N: 7 mm E:4 mm
ISLE
reservoir
receiving
DON
N: 12 mm E: -5.5
Assumptions & Limitations in predicted values:
1. 1993-1995 net transfer of ice mass from surge reservoir to
receiving area over the duration of surge is equal (~14km3).
GPS motions reflect magnitude of (un)loading & can be used
to invert for process scaling factor.
2. Ignores seasonal snow/ice build-up & summer melting.
Horizontal component is complex due to tectonic strain as
well as possible push moraine toward DON due to surge .
Surge dynamics on Bering Glacier, Alaska, in 2008–2011
E.W. Burgess , R. R. Forster , C. F. Larsen , and M. Braun, Cryosphere, 2012
2008-2011 was a smaller surge than 1993-1995, with less terminus advancement
reservoir
receiving
Above: Repeated aircraft laser
altimetry of elevation change (m/yr)
GPS PBO observations in this area
began ~ 2007
Below: L-, C- and X-band SAR derived ice
horizontal displacement rate (m/day) via
offset/speckle tracking methods along a
longitudinal profile of the Bering Glacier
MONTHLY Mean Land Surface Temperature (LST)
from Terra MODIS (11.02 and 12.02 mm)
March 2005
-4 to -1 -1 to 0
0 to 1 1 to 2
March 2012
2 to 3
AB35
Melt onset
AC09
Sauber et al., AGU, 2013
AB42
> 3 oC
North
East
AB42
AC09
Up
AB32
•Similar rates of horizontal deformation at the 3 coastal GoA sites .
•Uplift rates are more variable.
(SOPAC GPS Explorer combined solution)
Detrended North, East and Up for AB42, AB35, and AC09 (SOPAC, GPS Explorer)
Up annual term (mm): AB42: 12.9 + 0.1 AB35 = 10.5 + 0.1 AC09 = 7.9 + 0.1
GPS multipath for snow depth estimation (K. Larson & colleagues):
AC09
Up to almost a meter of snow near AC09. In this marine environment the snow has
high water content and undergoes compaction as the season progresses.
Snow loading begins ~Mid-September to October and reaches a maximum in late FebMarch. 1 km (hydrologist) versus 10 km (geophysicist) versus regional estimates.
CIG Pylith finite element model:
--Single 10km load at (0,0) with
time history given by
AC09 snow profile & r = 1.
--20 km elastic layer over 80
km Maxwell viscoelastic layer
meters
No of days*10
90 days
April
November
Solid Earth response
to Snow Loading
March
meters
May 2005
-4-4toto-1-1-1-1toto0 0
0 0toto1 1 1 1toto2 2
2 2toto3 3 > >3 3oC
May 2012
May 2012
AB35
AC09
Snow mostly gone
AB42
GRACE estimates of mass change
• 41,168 equal area 1-arc-degree mascons are directly estimated from GRACE KBRR
L1B data from the GRACE project [Tapley et al., 2004] with spatial and temporal
exponential taper constraints applied.
• 10-day temporal resolution
• Spatial constraint: 100 km correlation distance
• Temporal constraint: 10-day correlation
• Includes: LIA (Larsen et al., 2005)
and ICE5G(Peltier, 2004) corrections
Gulf of Alaska mascons (61)
Overall G of Alaska Trend = -69 ± 11 Gt a-1
but note variability from year to year
Gt
such as the cold (2012) versus warm
water (2005) years.
GSFC Mascon solution, Luthcke et al., J. of Glaciology, 2013 12
*
NC
Regional Variability:
*
Gulf of Alaska
*SE
Mascon 1457: Central Coastal
(study area)
Moderate seasonal, 50 -75 cm w.e.
Largest 10-year trend
Mascon 1484: North Central
Smaller seasonal signal
Moderate 10-year trend
Mascon 1425, South Eastern Alaska:
Large seasonal, up to 100 cm w.e. (p to p)
Small 10 year-trend in ice mass loss
1484
cm
w.e.
1425
1457
Year
GRACE MASCONS near three cGPS sites
1449 (AB42)
cm
w.e.
1457 (AC09)
1456 AB35)
Reminder: GRACE is sensitive to changes at spatial scales >300km
NEXT SLIDE: We use the estimate of the temporal history of mass change
to estimate surface load changes again assuming a 10km load near a site.
CIG Pylith finite element model:
--Single 10km load at (0,0) with
time history ((cm w.e. profile)
given by GRACE water year 2012
August - September.
February
November
--20 km elastic layer over 80
km Maxwell viscoelastic layer
Days x 30
Simulation of the present-day postseismic gravity change
1964 Mw = 9.2 Prince William Sound (Alaska) earthquake
Spherical harmonic
coefficients from RL05
CSR L2 monthly data,
degree up to 60; ~330
km resolution.
Postseismic gravity change
in EWH [cm/yr]
~1 cm/yr
Epicenter of the 1964 Alaska earthquake
- Used Johnson et al. [1996] finite fault model:
inversion of tsunami and geodetic data to estimate
1964 coseismic slip
- We used the viscoelastic Earth model that is
consistent with other studies in this region [Suito and
Freymueller, 2009; Sato et al., 2010]
- Global normal mode relaxation code was used to
compute the gravity change over the period 2002 2014. (Courtesy of R. Riva and F. Pollitz)
=> ~1 cm/yr of gravity
change (in w.e.) is
predicted, primarily
dependent on the
asthenosphere viscosity
(1019 Pa s)
Schematic
Time-series at the epicenter (60N 212W)
GRACE secular trend ~ –8.3 cm/yr (observed RL05 CSR L2 )
1964 EQ postseismic change ~ +1.0 cm/yr (model predicted)
The postseismic gravity change
could be as large as 10% of the
observed mass change, even 50
years after the 1964 earthquake.
Summary
How does cryosphere mass change on times scales of months to years in southern
Alaska?
1. The individual GRACE mascons located in distinctly different regions capture important interannual variations in the magnitude of the seasonal signal wastage trend.
2. These GRACE differences are important for estimating the timing and magnitude of broadscale differences between regional GPS sites; however, more local estimates of changes in
snow/ice extent and magnitude are needed to model cryosphere signal in GPS time series.
As estimated from EarthScope GPS data and FEM modeling, how do the surface
displacements due to inter-annual and seasonal cryosphere mass change compare to
tectonic displacement rates?
1. In the study region the surface horizontal displacements due cryosphere changes are
generally <10% of the tectonic displacements whereas the vertical displacements can be
comparable in magnitude to predicted tectonic uplift rates in localized regions near glaciers.
1. With a longer history of continuous GPS, and careful management of the sites, we may be
able to use GPS derived changes in vertical and horizontal displacements to constrain
process oriented models of cryosphere changes.