The Cryosphere Discuss., 7, C1625–C1630, 2013
www.the-cryosphere-discuss.net/7/C1625/2013/
© Author(s) 2013. This work is distributed under
the Creative Commons Attribute 3.0 License.
Open Access
The Cryosphere
Discussions
Interactive comment on “Constraining
GRACE-derived cryosphere-attributed signal to
irregularly shaped ice-covered areas” by
W. Colgan et al.
W. Colgan et al.
william.colgan@colorado.edu
Received and published: 30 August 2013
We thank Anonymous Referee #2 for their interest in our work. We appreciate the keen
eye with which they have reviewed our methodology. We would like to address their six
major comments (numbered 1 through 6) while the discussion forum is still "open", in
case Anonymous Referee #2 can provide further insight. My co-authors and I intend
to address the remaining minor comments in final discussion.
Re: #1 Clarification of precise inversion data – We apologize for being unclear in our
discussion paper: We are indeed inverting the gridded ultimate rate of mass change
field, not individual spherical harmonic coefficients. The inversion is therefore executed
C1625
in the Cartesian (or "node") domain, rather than the spherical harmonic domain. We
will explicitly state this clarification in a revised copy of the manuscript. We will similarly
clarify that we regard the "characteristic scaling length" of a Gaussian filter as being
equivalent to its "standard deviation". We note that spherical harmonics have only been
available up to degree 60 since the inception of GRACE. Thus, it is not possible to trun-
cate spherical harmonic solutions at higher orders (Tapley et al., 2004). We also note
that rather than being prescribed a priori, the Gaussian filter length scale subsequently
used by the inversion algorithm (200 km) was established through a sensitivity analysis
to establish the optimal length scale at which the inversion minimized the root-mean-
squared error when compared with the input GRACE data (discussion paper Figure
7). Thus, we contend that the combination of degree 60 spherical harmonic solution
and Gaussian filter length scale of 200 km does indeed preserve maximum informa-
tion of the magnitude and spatial distribution of mass changes throughout the inversion
process, while honouring the fundamental spatial resolution of the GRACE satellites.
Re: #2 Generating spherical harmonics from ground-level data – We completely agree
that there is a pressing need for coarser resolution spherical harmonic solutions to be
derived from higher resolution inverted ground-level mass change fields (i.e. discussion
paper Figure 10; Barletta et al., 2012) to facilitate further inversion validation. This
would certainly close the circle: generating higher resolution ground-level fields through
inversion, and then forward modeling the corresponding coarser resolution spherical
harmonic fields, and so on. We have given significant thought to forward modelling the
spherical harmonic solution space that corresponds to the inverted field we present.
As we are inverting/inferring a trend over a given period, rather than absolute mass
anomalies at each monthly GRACE time point, we would have to carefully stitch this
cryospheric trend into a forward model of global mass anomalies over the time period.
This forward model would also have to incorporate the usual suite of processes that
can substantially influence gravimetry observations (i.e. corrections for atmospheric,
oceanic, isostatic rebound, etc.). We view this as a non-trivial task. We are currently
working towards a framework to develop and implement such a complimentary forward
C1626