Deriving long-term sea level variations at tide gauge stations in Atlantic North America
Tide gauge recording is an indispensable geodetic tool to study local sea level variations in coastal areas and mean sea level trends on a global scale. Combined with satellite altimetry sea surface heights or GNSS- derived ellipsoidal heights, tide gauge records have been used for deriving estimates of vertical crustal motion elsewhere. The objective of our paper is to study the capabilities of the method of singular spectrum analysis, which is generically related to empirical orthogonal functions/principal component analysis technique, to derive local long-term and secular sea level trends. This method allows one to extract any anomalous sea level signals, which is difficult to achieve by the conventional harmonic analysis. We use a set of tide gauge stations in Atlantic Canada and the USA extracted from the monthly PSMSL data base. To create continuous time series, we fill all time gaps by interpolating the main periodic and trend components of the sea level signal using the least squares harmonic analysis. We analyze all tide gauge time series simultaneously through singular value decomposition of the time-lagged series combined in a data trajectory matrix. This enables us to extract and separate the main modes of variability of the local sea level and to study only the long-term spatio-temporal patterns in the sea level variations. Our preliminary results show that the length of the tide gauge time series and the relative contribution of the signals to the total sea level variance are the two crucial factors that may preclude the separation of the local secular sea level fall or rise from any decadal sea level variability. The outcome of our study will be useful for combined satellite altimetry/TG/GRACE studies of sea level changes in the coastal areas, studies of vertical crustal motion due to postglacial rebound in the region, as well as the definition and realization of a dynamic reference surface for orthometric heights in North America.
New 1-arc-minute Geoids for Alaska and its Vicinity
A new gravimetric geoid with 1 arc minute spatial resolution has been computed for Alaska and its vicinity, ranging from 49N to 72N in latitude and 172E to 234E in longitude. More than 3 million gravity data collected by NGS (National Geodetic Survey) and other agencies such as NGA (National Geospatial-Intelligence Agency), NRCan (Natural Resources Canada), and DNSC (Danish National Space Center) are gridded into 1'x1' mean gravity anomalies, which are combined with the EGM08 coefficient model, up-to degree and order 2160, by a modified Stokes's Kernel. In addition to the improvement on the spatial resolution, the new gravimetric geoid shows about 3cm precision improvement over EGM08, compared with the geoid heights at 199 GPS leveling benchmarks in this region. Finally, a better than 2cm accurate hybrid geoid model is generated by removing both the systematic errors and the correlated signals in the differences of the gravimetric geoid at these GPS leveling benchmarks, which are modeled by a linear system and a multi-matrix covariance (Roman et. al., 2004) function,respectively.
The European Gravimetric Quasigeoid EGG2008
A new European Gravimetric Quasigeoid model EGG2008 was computed within the framework of the European Gravity and Geoid Project (EGGP), a project of IAG Commission 2. The new model includes several updates with respect to the last computation in 2007 (EGG2007). The re-computation was considered necessary, because the evaluation of the 2007 results and especially the comparisons with the ultra-high- degree geopotential models PGM2007A and EGM2008 from NGA indicated that some of the EGGP gravity sources had biases due to incorrect gravity reference system information. After a re-evaluation of the suspicious sources, some land data sets were updated. In addition, several marine gravity data sets were improved, and the terrain reduction procedure was revised. The modelling procedure was adopted from the 2007 computation. The new EGG2008 model is compared with the 2007 model, and an evaluation by national and European GPS and levelling data sets is performed. The new model shows improvements over the 2007 model in selected regions where data updates were realized.
Effects of complex topographical mass-density distributions on geoidal height in the Stokes-Helmert scheme
Geoid computation according to the Stokes-Helmert scheme requires accurate modeling of mass-density variation within topography. Current topographical models consider only horizontal density variations, although in reality topographical density varies three-dimensionally. The lack of accurate knowledge of regional three- dimensional topographical density distribution may prevent precise geoid evaluation from real data. In light of this deficiency, we attempt to at least estimate the order of magnitude of the errors in geoidal heights caused by neglecting the 3-d variation. We do this by calculating, for artificial but realistic mass-density distributions, the difference between results from 2-d and 3-d models. Our previous work has shown that for simulations involving simple mass-density distributions in the form of planes, discs or wedges, the effect of mass-density variation unaccounted for in 2-d models can reach centimeter-level magnitude in areas of high elevation, or where a large region of anomalous density exists over a wide area. However, real mass-density distributions are more complicated than those we have modeled so far, and involve multiple structures. To expand on our previous work, we now present results for effects on geoidal height of more complex mass-density structures. Complex structures are formed by different methods, such as creating arrays of simple structures, or by arbitrarily assigning slanting downward interfaces between rock types given by 2-d geological maps. By these simulations we provide a more precise indication of how concerned we should be about effects of vertical density variations on geoidal height, and under what conditions they are worth worrying about. We also evaluate the potential for use of and other global crustal models for better density modeling in Stokes-Helmert geoid determination.
USGG2009, GEOID09, and DEFLEC09: Updated Models for the United States and Its Territories
Updated gravimetric and hybrid geoid models have been produced for the Conterminous United States
(CONUS), Alaska, Hawaii, Puerto Rico and the U.S. Virgin Islands, Guam, the Commonwealth of the Northern
Marianas Islands, and American Samoa. These models are built on an improved reference gravity model
developed from GRACE satellite gravity mission data and millions of surface gravity points. SRTM 3" data were
used fr most regions to produce consistent terrain effects, although a more complicated model was required in
Alaska to account for the northern latitude cutoff of the SRTM mission. These gravimetric geoid (USGG2009)
models represent to geoid heights between a GRS-80 ellipsoid shell centered at the ITRF00 reference frame.
The hybrid geoid height models (GEOID09) were developed from the USGG2009 model and available GPS-
derived ellipsoid heights on leveled bench marks (GPSBM's). The leveling is tied to different datums for most
regions (NAVD 88 in CONUS and Alaska, PRVD02 in Puerto Rico, etc.). Finally, deflection of the vertical models
(DEFLEC09) were also produced for the same regions. Initial comparisons between these models and
uncorrelated, external data show significant improvements (better than 50%) over the previous models
(USGG2003, GEOID03). Comparison data include minimally constrained GPSBM's, tidal bench marks, and
astrogeodetic DOV's. These models represent an update to the exiting official U.S. national models and, likely,
the final time that current techniques and data will be used. Future models will rely upon these models as a
baseline for efforts to develop a common, seamless, gravimetric geoid height model to serve as a new vertical
reference system for both Canada and the United States.