Sie sind auf Seite 1von 6

2009 World Congress on Computer Science and Information Engineering

An Integrated GIS Database Server for Malaysian Mapping, Cadastral and


Location-based Systems (LBS)

Teoh Chee Hooi (teohch@utar.edu.my)


Department of IT and Engineering, Universiti Tunku Abdul Rahman (UTAR), Malaysia

Abstract done using the Rectified Skew Orthomorphic System


(RSO), which is an oblique mercator projection
This paper discusses the design and development of technique defined by Hotine in 1947 [1].
a Malaysian Geographical Information System (GIS) The reference ellipsoid used for MRT is the
database server for location data sources available in Modified Everest, with Bukit Kertau as the origin. The
Malaysia, such as the Global Positioning System Timbalai 1948 system also is based on the Modified
(GPS), WGS84, Geocentric Datum for Malaysia Everest reference ellipsoid with its origin centered at
(GDM2000), Geodetic Reference System 1980 Mount Timbalai. The Kertau 1948 system was revised
(GRS80), local geodetic measurements such as in 1968 to become the MRT 1968 and the Timbalai
Malayan Revised Triangulation 1948 (MRT48 or 1948 system in East Malaysia was revised to become
KERTAU 48), Borneo Triangulation 1948 (TIMBALAI the Borneo Triangulation (BT) 1968. In practice, the
1948), Malayan Revised Triangulation 1968(MRT68), Kertau Datum of the MRT is defined in terms of only
Borneo Triangulation 1968 (BT68), local cadastral the horizontal components (f, l) of the Modified
information such as the ECadastre Coordinated Everest Ellipsoidal coordinates. However, there is no
Cadastral System (CCS) system developed by JUPEM explicit definition of the ellipsoidal height (h) [2].
and cellular coordinate information based on Cell- For cadastres however, the Cassini Soldner system
based ID, E-OTD etc It describes the pros and cons of (CS) is employed in Peninsular Malaysia whereas the
the abovementioned systems, analyze the accuracies same RSO system is used in East Malaysia. The
and tolerance of the data source. It also highlights the Cassini projection system is based on local datums and
disparities between different systems, and proposes computed with their corresponding equations and
fast methods to correlate and transform these data into respective state origin coordinates [3]. The existing
a unified, reliable data source. Lastly it will cover the Cassini projection for cadastral mapping was derived
storage and retrieval process for the textual and from the RSO system referenced to the Modified
graphical GIS data. Everest ellipsoid.

1. Introduction 2. Issues and Challenges of the Existing


Datum
Malaysian survey and mapping system is very
complicated due to the use of multiple projection The use of multiple datums in survey and cadastres
techniques for different purposes. This can be traced has not only created inconveniences such as lengthy
back to the history of Malaya until now. conversion from RSO to Cassini but also created data
The Malaysian geodetic surveying and mapping is incompatibility issues with newer satellite-based
based on a few standards. Prior to independence (31 navigation systems such as Global Positioning Systems
August 1957), the system used was called The (GPS) [4]. Apart from that, data duplication and
Malayan Revised Triangulation (MRT1948), also redundancy may exist when generating the conversion
known as Kertau 1948. It was based on manual lookup table to speed up the transformation process.
observations and combined with the Repsold The need for a standardized geodetic infrastructure
Triangulation done in 1913-1916 by the British. The for integration of spatial data was first discussed in a
datum format for West Malaysia and Singapore until summary report by Kadir et.al.[5].They proposed a
1968 was the Kertau 1948, whereas the datum for East new geocentric datum for the requirements of 3-
Malaysia, including Sabah and Sarawak was Timbalai dimensional geodetic spatial referencing. The
1948 (BT1948). Both mapping of MRT and BT are Department of Survey and Mapping Malaysia

978-0-7695-3507-4/08 $25.00 © 2008 IEEE 162


DOI 10.1109/CSIE.2009.938
(DSMM) has since taken the initiative to introduce a
geocentric datum format in year 2003 that will Where a, b and ε are respectively, the semi-major-axis
alleviate the problems mentioned above. The standard length, the semi-minor-axis length and the second
is called the Geocentric Datum for Malaysia numerical eccentricity of the reference ellipsoid
(GDM2000) [6]. The GDM2000 is based on the associated with the new datum
International Terrestrial Reference Frame (ITRF2000)
and Geodetic Reference System 1980 (GRS80) P = X 2 +Y2
reference ellipsoid. It entails deployment of Malaysian e2
Active GPS Stations (MASS) for that work closely ε=
1 − e2
with the nearby 17 International GPS Service (IGS)
The parametric latitude u is given by:
sites for baseline processing. This MASS network will
b
also double as the Zero Order Geodetic Network for u = arctan[ tan φ ]
Malaysia. a
The longitude is given by the closed solution:
3. Coordinate Transformations Y
λ = arctan[ ] (5)
X
The standard transformation procedures adopted by The approximate height h is given by:
the national Department of Survey and Mapping h = P cos φ + Z sin φ − a 1 − e 2 sin 2 φ (6)
Malaysia (DSMM) are discussed below.

3.1. Forward Transformation 3.3. The Bursa-Wolf transformation model

According to Heiskanen and Moritz [7], the The Bursa-Wolf transformation model [10, 11] is a
forward transformation from geodetic co-ordinates (λ, seven-parameter model for transforming three-
φ, h) to geocentric Cartesian co-ordinates (X, Y, Z) is dimensional Cartesian co-ordinates between two
given by the following: datums. This model is well suited for satellite datums
on a global scale [12]. There are altogether three
X = (N + h) cos φ cosλ (1)
geocentric datum shift parameters (ΔX, ΔY, ΔZ), three
Y = (N + h) cos φ sinλ (2) rotation elements (Rx, Ry, Rz) and one scale factor
2 (1+ΔL).
b
Z =( N + h) sin φ (3)
a2 The matrix-vector form of the model is:
Given that: ⎡X ⎤ ⎡ΔX ⎤ ⎡1 + ΔL RZ − RY ⎤ ⎡ X ⎤
a is the semi-major axis of the reference ellipsoid ⎢Y ⎥ ⎢ ⎥ ⎢
b is the semi-minor axis of the reference ellipsoid ⎢ ⎥ = ⎢ ΔY ⎥ ⎢ − RZ 1 + ΔL RX ⎥⎥ ⎢⎢ Y ⎥⎥
e is the first eccentricity of the reference ellipsoid ⎢⎣ Z ⎥⎦WGS 84 ⎢⎣ ΔZ ⎥⎦ ⎢⎣ RY − RX 1 + ΔL ⎥⎦ ⎢⎣ Z ⎥⎦ MRT
(7)
Generally it is not usually necessary to use the
where the subscripts WGS84 and MRT refer to the
ellipsoid height (h) in equations (1), (2) and (3)
WGS84 and Malaysian geodetic datums, respectively.
because the forward conversion is only weakly
In order to convert the geocentric coordinates of
dependent upon height h.
(X,Y,Z) to the geodetic coordinate (λ, φ, h), ellipsoid
properties for the respective datums must be known.
3.2. Reverse Transformation
3.4. Geocentric RSO Projection
Bowring’s method [8] is well known to be the most
efficient and fast method for the reverse transformation
The Rectified Skew Orthomorphic (RSO)
from geocentric Cartesian co-ordinates (X, Y, Z) to
projection is an Oblique Mercator projection
geodetic co-ordinates (λ, φ, h). Although closed (non-
developed by Hotine in 1947[1, 13]. RSO is used in
iterative) solution to the reverse transformation exists
surveying of Malaysia since it is a country with large
by using quartic equations [9], we will use Bowring’s
extend of areas in one single direction. Geocentric
method due to its speed and convenience.
RSO uses GRS80 as its ellipsoid parameter. The
Z + ε 2b sin 3 u parameters and constants for RSO are excluded to
φ = arctan[ ] (4)
P − e2 a cos3 u conserve space.

163
RSO Forward Conversion
The following will be computed to find (E, N) from E = FE + v(A-TA3/6) – (8-T + 8C)T * A5/120) (12)
a given (Ø, λ): N = FN + M – M0 +
⎛π φ ⎞ e
t = tan⎜ − c ⎟ /[(1 − e sin φ c ) / (1 + e sin φ c )]2
v tan( φ ) [A2/2 + (5 – T + 6C) A4/24] (13)
⎝4 2⎠ where
Q = H /tB A = (λ-λ0 ) cos ( φ ), T = tan2( φ )
S = (Q − 1 / Q ) / 2 C = e2 / (1-e2) cos 2( φ ) , V = a / (1-e2sin2( φ )) ½
T = (Q + 1 / Q) / 2 M = meridian arc distance
V = sin( B (λ − λ0 )) = a [ (1-e2/4 – 3e4/64 – 5e6/256 - …) φ -
U = −V cos(γ 0 ) + S sin(γ 0 )) / T (3e2/8+3e4/32+45e6/1024 +…) sin 2 φ +
v = A ln[(1 − U ) /(1 + U )] / 2 B
(15e4256+45e6/1024 +…) sin 4 φ -
A ⎡ S cos(γ 0 ) + V sin(γ 0 ) ⎤
u= arctan ⎢ ⎥ (35e6/3072+…) sin 6 φ …]
B ⎣ cos( B(λ − λ0 )) ⎦
with φ in radians.
Hence:
N, E = Computed Cassini coordinates
E = v cos(γ c ) + u sin(γ c ) + FE (8) FE, FN = Cassini State origin coordinates
λ0 = longitude of state origin
N = u cos(γ c ) + v sin(γ c ) + FN (9) λ = longitude of computation point
φ = latitude of computation point
RSO Reverse Conversion
To find (Ø, λ) from a given (E, N), the following Cassini Reverse Conversion
parameters must first be computed: The formula to derive longitude and latitude pair is
v' = ( E − FE ) cos(γ c ) − ( N − FN ) sin(γ c )
shown below:
u ' = ( N − FN ) cos(γ c ) + ( E − FE ) sin(γ c )
φ = φ 1- v1 tan φ 1 /ρ1 [D2/2 – (1 + 3T1) D4/24] (14)
Q' = exp(− Bv' / A)
S ' = (Q'−1 / Q' ) / 2
λ = λ0 + [D – T1D3/3 + (1+3T1).T1. D5/15] / cos φ 1
T ' = (Q'+1 / Q' ) / 2 (15)
U ' = V ' cos(γ 0 ) + S ' sin(γ 0 )) / T '
where,
v1 = a (1-e2) / (1-e2sin2 φ 1) 3/2 ρ1 = a / (1-e2sin2 φ 1) 1/2
[
t ' = H / (1 + U ' ) /(1 − U ' ) ]1/ B

e1 = 1 - (1-e2) ½ / 1+ (1-e2) ½ μ1 = M1 / (1-e2sin2 φ 1) 3/2


χ = π / 2 − 2 arctan(t ' )
Hence: φ 1 = μ1 + (3 e1/2 - 27 e13/32 + …)sin2 μ1 +
φ = χ + sin(2 χ )(e2 / 2 + 5e4 / 24 + e6 / 12 + 13e8 / 360) + (21 e12/16 - 55 e14/32 + …)sin4 μ1 +
(151 e13/96 + …)sin6 μ1 +
sin(4 χ )(7e4 / 48 + 29e6 / 240 + 811e8 / 360) + (1097 e14/512 + …)sin8 μ1 + …
sin(6 χ )(7e6 / 120 + 81e8 / 1120) + M1 = M0 + (N – FN)
Note: M0 is the value of M calculated from the
sin(8χ )(4279e8 / 161280) Latitude of the origin
(10)
T1 = tan2 φ 1
⎡ S ' cos(γ c ) − V ' sin(γ c ) ⎤
λ = λ0 − arctan ⎢ ⎥ B (11) D = (E – FE) / v1
⎣ cos( Bu ' / A) ⎦
4. Design and Implementation
3.5. Cassini-Soldner 4.1. GSM Cellular Positioning

The Cassini-Soldner is a non-conformal transverse This project leverages on public GSM cell-based
cylindrical projection method [13]. It has the ID database information at Telematica Instituut,
advantage of ease of implementation but is eventually Netherlands [14]. It leverages on public contribution to
being replaced by conformal transverse Mercator. estimate the location of GSM cells based on all
measurements of people that have traversed through
Cassini Forward Conversion that cell. The database contains Malaysia’s Celcom’s
The formula to derive Easting and Northing (E, N) GSM network. A total of 247 cell-positions are
are relatively simple, as shown below: observed with an average cell distance of 949m, all

164
with up to 90% signal strength. To determine a cellular source geographical information system are GRASS
location1 we just need to access the getLocation (Geographic Resources Analysis Support System)
method in the CellStore2 database. [16], and MapServer [17], and PostGIS [18]. We have
The 9 parameters has the following form: finally chosen MySQL to experiment with the
{CountryID (Integer), NetworkID (Integer), Cell geometry and binary large object (blob) data type.
Name (String), Date (Long), GPS-track positions
regarding specified cell, Latitude(Double), Longitude 4.5. Transformation and Conversion Engine
(Double), Range(Integer), LocalAreaCode (Integer),
signalStrength(Integer)} The transformation engine architecture is illustrated
in Figure 1. There are several important
4.2. Coordinate Positioning considerations, namely:
1. Speed of conversion –Some formula has large
There are 13 different datum transformation asymptotic (or more commonly known as
parameters values and 7 different coordinate operation computational complexity) or is iterative and
parameters to be maintained for map projections, as hence require a lot of computation time. Formula
shown in Table 1 and Table 2. such as φreverse RSO has an order of (e8).
2. If decent speed cannot be achieved, it may be best
Table 1. Datum Transformation Parameters to precompute the parameters and values and put
them in a lookup table.
Coordinate operation
3D
Coordinate Operation Method
Bursa- Geocentric
3. Data storage requirements arising from storage of
parameter name Geographic Wolf Coord. System precomputed values.
Coord. 3D model
System 4. The delay time may increase for long lookup
Type Ellipsoidal Cartesian Cartesian tables so care must be taken to minimize such
Coordinates Coord. Coord.2
Latitude of system(Øs) 1 effect.
Longitude of system(λs)
Ellipsoidal height of system(hs)
2
3
Cell-ID and GPS data can be consumed directly as
Datum shift parameter (Δx) 1 it is normally embedded with positioning information.
Datum shift parameter (Δy) 2 The positioning information can be conveniently
Datum shift parameter (Δz) 3
Rotation element (Rx) 4 converted into WGS84 format and later transform to
Rotation element (Ry)
Rotation element (Rz)
5
6
geocentric datum or the non-geocentric RSO or
Scale factor (1+ΔL) 7 Cassini. This will allow potential use of GPS enabled
Geocentric coordinate X
Geocentric coordinate Y
1
2
mobile device to become an auxiliary device for
Geocentric coordinate Z 3 surveying and mapping.

4.3. Coordinated Cadastral System (CCS) Table 2. Map Projection Parameters


Coordinate operation parameter Coordinate Operation Method
name RSO Cassini Soldner
The DSMM and UTM has started the development Type Map Projection Map Projection
of a Coordinated Cadastral System (CCS) and national (survey) (cadastral)
Latitude of natural origin (Ø0) 1
Digital Cadastral Data Base (DCDB)in 2002 [15]. We Longitude of natural origin (λ0) 2
will focus on extracting only coordinates of area of Latitude of projection centre (Øc) 1
Longitude of projection centre (λc) 2
interest. The rest of the data structures such as lot, Scale factor on initial line (k ) 3
stone and boundary will not be included in our Azimuth of initial line (αc)
c
4
database design to avoid data duplications. Angle from rectified to skew grid (γc) 5
False Easting (FE) 6 3
False Northing (FN) 7 4
4.4. Spatial Database
4.6. Data Transformation Issues
Spatial databases have a different set of requirements
from conventional databases. We have decided to use
Since RSO and Cassini conversions are
open source database platform rather than commercial
undoubtedly slow, it is better to precompute and build
products like Oracle Spatial database, Boeing’s Spatial
lookup tables for faster results. To save storage space,
Query Server etc. Among the popular choices of open
we can filter the data along the coastline boundaries
leaving only the land.
1
http://cellid.telin.nl/wasp/services/CellStore2Service GPS uses WGS84 as its coordinate reference
2
with its center at earth’s centre of mass frame, i.e. reference ellipsoid GRS80. The difference

165
between WGS84 and GDM2000 is negligible, due to LBS normally will require the support of data
the fact that their parameter and values are very similar aggregator and value added supplier, which in turn will
(a, f). For simple application that does not require high favor the use of rule-based processing for data
precision, the GPS readings can be directly used with conflation in cartography and centralized rules-
the geocentric GDM2000. Proper conversion is still repository for quality control [21].
required for scientific applications such as tide gauge This will necessitate the use of rules database to
or gravity computation. support rule-based processing. Rule based processing
Another concern is the induced error due to the is centered on the Fact-Pattern-Action dynamics. Rule-
inaccuracy of the parameters. This can be amplified based processing is not new in GIS systems [22].
during transformation from one coordinate system to Detailed description of the role of rules-based
another. Errors can also be increased with the increase processing in spatial data management and update is
of transformation steps. It is important to keep the given by Woodsford [23].
transformation straightforward and minimal.
The use of Geocentric Cassini and Geocentric RSO
has simplified many of the surveying tasks as the point
RSO new Cassininew
surveyed using GPS can now be directly projected to (N, E) (N, E)

their respective geocentric plane. Support for the


existing non-geocentric Cassini and RSO are also GDM2000
(λ, φ, h)
GDM2000
(λ, φ, h)

maintained in our system to support backward


compatibility. GDM2000
(X, Y, Z)
Another problem that may arise is the positional
shift when converting from non-geocentric datum such Bursa-Wolf Bursa-Wolf

as RSOMRT1948 to geocentric datum such as 7 parameter 7 parameter

RSOGDM2000. Table 3 shows the disparity between the WGS84


results of RSOMRT1948 and RSOGDM2000. This has (X, Y, Z)

resulted in a projection shift with direction = 271o 31’


Cell-ID GPS WGS84 WGS84
25” and magnitude = 193.82m. This shift is also (λ, φ, r) (λ, φ, h) (λ, φ, h) (λ, φ, h)

subjected to amplification by the map scale used in the


projection. For example, a point on a map with a scale Bursa-Wolf Bursa -Wolf

of 1:50000 were shifted by 3.8764mm whereas on a 6 parameter 7 para meter

town sheet of scale 1:10000, the shift can be amplified MRT1948 BT1968
to as big as 1.982cm! (λ, φ, h) (λ, φ, h)

Table 3. Differences between M RSOold


(N, E)
BRSO old
(N, E)

RSOMRT1948 and RSOGDM2000


Parameter ΔE ΔN Cassini old
(N, E)
Cassiniold
(N, E)
Mean -193.75 6.09
Standard Dev, σ 0.73 0.47 Peninsular Borneo
Root Mean Square 0.72 0.47

4.7. Location-based System (LBS) Support Figure 1. Transformation Engine

Location based System allows a user to search, 4.8. Open XML support
identify and check for location information. Two
components that can support location determination for A web front-end written in Java is created to offer
LBS are present in our system – GPS and GSM search and lookup services to the external clients. A set
localization, making it a suitable candidate for location of lookup web service is also created for client
finding services. Two components that can support application to directly connect to the GIS database and
location determination for LBS are present in our manipulate the data. To reduce compatibility problem
system – GPS and GSM localization, making it a and facilitate ease of transfer, an open XML standard
suitable candidate for location finding services. The needs to be defined. A few candidates have been
user actions may include orientation and localization, implemented for our system, Keyhole Markup
search for point of interest (POI), people or subjects, Language (KML) and Geography Markup Language
navigation, identification, and event check [20]. (GML) and Scalable Vector Graphic (SVG). By

166
following the above open standard our system is able Summary Report submitted to Dept. of Surveying and
to provide support for both textual and graphical GIS Mapping Malaysia, 2002.
data. Both versions are now undergoing active testing [6] DSMM, A Technical Manual on the Geocentric Datum of
to assess its features and usability. Malaysia (GDM2000), Dept. of Surveying and Mapping
Malaysia, Kuala Lumpur, August 2003, chapter.1 – 2.
[7] Heiskanen W.H and Moritz H., “Physical Geodesy”.
Pre-processor
Freeman, San Francisco, 1967 pp 182.
Cadastral RSO
[8] Bowring, B.R., “The accuracy of geodetic latitude and
height equations”, Survey Review, vol. 28, no. 218, 1985
pp. 202-206.
Precomputed
[9] Borkowski, K.M., “Accurate algorithms to transform
geocentric to geodetic co-ordinates”, Bulletin
Cache
Database parameters
MySQL*

Géodésique (translated) vol. 63, 1989, pp. 50-56.


[10] Bursa, M., “The theory for the determination of the non-
Integrated GIS Spatial Server LBS Services parallelism of the minor axis of the reference ellipsoid and
Middleware (Itineries,
Identification,
the inertial polar axis of the Earth, and the planes of the
Application Map Service Feature Service
KML SVG
Search etc,) initial astronomic and geodetic meridians from observations
Index / Catalog Transformation of artificial Earth satellites”, Studia Geophysica et
Service Service
XML
CellStore2 Geodetica, no. 6, 1962, pp. 209-214.
Web Services [11] Wolf, H., “Geometric connection and re-orientation of
Google SVG three-dimensional triangulation nets”, Bulletin Géodésique,
Earth / Controller
Maps
Client Web
Access
GIS Apps no. 68, 1963, pp. 165-169.
[12] Featherstone W. E, “A comparison of existing co-
Mobile GPS ordinate transformation models and parameters in Australia”,
Client
Cartography, vol. 26, no. 1, 1997, pp. 13-26.
*PostGIS in future [13] EPSG, EPSG Guidance Note No. 7 Part 2, European
Petrol Survey Group, Nov 2005.
Figure 2. Web Services and Back-end [14] Espen Talberg, “Positioning with Cell ID”, Context
Preprocessor Design Aware Systems, Institute of Informatics, Østfold University
College, Norway, 2006
5. Conclusion [15] Abd.Majid et. al., “Studies toward the Development of
Implementation Plan of Coordinated Cadastral System for
In this paper, we have proposed a geocentric, Peninsular Malaysia”, International Symposium and
Exhibition on Geoinformation, Kuala Lumpur, Oct 2002.
integrated GIS database server application that can
[16] Neteler, M. and H. Mitasova, "Open Source GIS: A
support mapping, cadastral, positioning and Location- GRASS GIS Approach. 3rd Edition", Springer, New York,
based Systems (LBS) for various Malaysian data 2008.
sources. WE have also addressed the issues of [17] MapServer Home Page: http://mapserver.gis.umn.edu/
disparities between the conversion of different [accessed 14/10/2008]
mapping and projection systems in section 4.6. [18] Steiniger and Bocher, "An Overview on Current Free
and Open Source Desktop GIS Developments”,
6. References http://www.spatialserver.net/osg [accessed 14/10/2008]
[1] Martin Hotine, “The Orthomorphic Projection of the [19] John P. Snyder, “Map Projections — A Working
Spheroid”. Empire Survey Review, No.64, Vol. 9, 1947, pp. Manual”, U.S. Geological Survey Professional Paper 1395,
25-166. Washington, 1987
[2] Wan Aziz, Majid. K, Sahrum, S., Teng, C.B, “Cadastral [20] Stefan Steiniger, Moritz Neun and Alistair Edwardes,
Reform in Malaysia: Vision 2000s”, Proceedings of Map “Foundations of Location Based Services", University of
Asia 2002, Bangkok, Aug 2002. Zurich, 2006, pp 3-26.
[3] Melita Kennedy, “Understanding Map Projections”, [21] Mike Jackson, and Peter Woodsford, “Spatial Database
Environmental Systems Research Institute, USA, 2000. Challenges for Framework Data and Location based
[4] Majid Kasir, Sharum Ses, Kamaludin Omar, Ghazali Services”, The International Archives of the
Desa, Abdullah Hisam Omar et al, “Geocentric Datum Photogrammetry, Remote Sensing and Spatial Information
GDM2000 For Malaysia: Implementation and Implications”, Sciences. Vol. XXXVII. Part B4. Beijing 2008, pp 1431-
Official Launching of the GDM2000, Dept. of Surveying and 1436.
Mapping Malaysia (DSMM), Kuala Lumpur, 28th August [22] Chance, A., Newell, R.G., Theriault, D.G, “An Object-
2003. Oriented GIS – Issues and Solutions”, Proceedings of the
[5] Kadir, M., Desa, G., Ses, S. and K. Omar, “Studies EGIS '90 Conference, Amsterdam, April 1990.
toward the development of implementation plan of [23] Woodsford. P. A.., “Spatial Database Update and
Coordinated Cadastral System for Malaysia”, Executive Validation – Current Advances in Theory and Technology”,
ISPRS Archives Vol XXXVI, Part 4/W54, 2007, pp.1-7

167

Das könnte Ihnen auch gefallen