International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume XXXIX-B4, 2012
XXII ISPRS Congress, 25 August — 01 September 2012, Melbourne, Australia
IMPLEMENTATION OF THE DISTRIBUTED PARALLEL PROGRAM FOR GEOID
HEIGHTS COMPUTATION USING MPI AND OPENMP
Seongkyu Lee*, Jinsoo Kim®, Yonghwa Jung“, Jisun Choi *, Chuluong Choi ^*
* Dept. of Spatial Information Engineering, Pukyong National University, 45, Yongso-ro, Nam-Gu, Busan, South Korea
- geoslegend@gmail.com, invitation7@nate.com, sun33023@hotmail.com, cuchoi@pknu.ac.kr
® ZEN21, 2nd Floor, RNC building, 981-1 Bangbae 3-dong, Seocho-gu, Seoul 137-848, South Korea -
pknu9680@gmail.com
Commission IV, WG IV/S
KEY WORDS: Distributed parallel computing, Geoid heights computation, MPI, OpenMP, Diskless-based PC Cluster
ABSTRACT:
Much research have been carried out using optimization algorithms for developing high-performance program, under the parallel
computing environment with the evolution of the computer hardware technology such as dual-core processor and so on. Then, the
studies by the parallel computing in geodesy and surveying fields are not so many. The present study aims to reduce running time for
the geoid heights computation and carrying out least-squares collocation to improve its accuracy using distributed parallel
technology. A distributed parallel program was developed in which a multi-core CPU-based PC cluster was adopted using MPI and
OpenMP library. Geoid heights were calculated by the spherical harmonic analysis using the earth geopotential model of the
National Geospatial-Intelligence Agency(2008). The geoid heights around the Korean Peninsula were calculated and tested in
diskless-based PC cluster environment. As results, for the computing geoid heights by a earth geopotential model, the distributed
parallel program was confirmed more effective to reduce the computational time compared to the sequential program.
1. INTRODUCTION
These days, global navigation satellite systems (GNSS) have
been widely used and geoid is a very important element which
makes it possible to determine orthometric height using GNSS
positioning. Moreover, many countries have continued to make
efforts to develop their own high-precision geoid model
(Chandler and Merry, 2010; Kuroishi et al., 2002; Roman et al.
2009; Toth et al. 2000)
Since global geopotential models (GGMs) well expresses long
wavelength component of the Earth's gravity field (Daho et al.
2008; Krynski and Lyszkowicz 2006), they not only provide
basis for the gravity field to develop high-precision geoid model
but also have important meanings as reference surface for
calculating a local geoid (Bae et al., 2011; Dawod et al., 2010).
In addition, countries which are difficult to develop geoid
models have been using GGMs for rough calculation of geoid
heights (or undulations) and gravity anomaly through spherical
harmonic analysis (SHA) (Lee et al., 2008)
Geopotential data that is recently observed by GRACE satellite
has contributed to more precise GGMs development and
especially these (geopotential data) have been greatly
contributed to development of Earth gravitational model 2008
(EGM2008). EGM2008 is complete to degree and order 2,159
and includes additional spherical harmonic coefficients
extending to degree 2,190 and order 2,159 (Pavlis et al., 2008;
Pavlis et al., 2012).
Moreover, because GGMs-derived geoid heights are referred to
a global vertical datum and these are fitted with a local vertical
* Corresponding author.
datum by GPS/levelling-derived geoid heights, the accuracy of
these (geoid heights) could be enhanced. For this purpose, least-
squares collocation (LSC) method is widely used (Kotsakis et
al., 2008; Lee et al., 2008).
As mentioned above, although SHA and LSC are mostly used to
evaluate accuracy and suitability of ultra-high degree
geopotential models for certain local areas or countries, they all
contain time-consuming problem. In geoscience simulations, to
solve this problem, resources are distributed parallel
computation is getting popularity to reduce data processing time.
This paper is intended to carry out LSC to fit ultra-high degree
SHA by EGM2008 using distributed parallel computation on
PC cluster and present its performance evaluation.
2. DATA
2.1 EGM2008 geopotential model
EGM2008 is a spherical harmonic model of the Earth's
gravitational potential, developed by U.S. National Geospatial-
Intelligence Agency (NGA) (Lemoine, 1996; Pavlis et al., 2008;
Pavlis et al., 2012).
EGM2008 is the first-ever global model that is capable of
resolving the Earth's gravity field beyond spherical harmonic
degree 2000. EGM96 that is a widely used GGM has spatial
resolution of 30 arc minutes whereas EGM2008 is a high-
resolution GGM of Earth's gravity field that allows computation
of geoid heights down to a resolution of 5 arc minutes (Dawod
et al., 2010; Pavlis et al. 2008).
212
2.2
GF
evi
bu
An
val
ge
dei
hei
the
po
(Fi
Fig
3.1
SH
cot
hai
sat
La
car
pot
an
spl
(19
(1€
