International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B4. Istanbul 2004 Intern 
IAU Equatorial| Polar Wo . |Wo-Wo (142000) 3.2 Horizontal Reference qus nm 
definition radius radius J i} a D| Dz 
[km] [km] [.] [>] GRASS supports any user-defined sphere or biaxial ellipsoid. 
The radii have to be enlisted in the file etc/ellipsoid.table (see ——] 
: 72 6 : 
IAUI985 | 3393.40 | 3375.80 | 176.729 0.099 Table 2). G 
JAU1988/ | Ak (D | ete 
ec 3397.00 | 3375.00 |176.868 0.238 : ; a Mi 
IAUI99] 3.2.1 Planetographic Latitude: As within GRASS all 
IAUI994 | 3397.00 | 3375.00 |176.901 0.271 ellipsoidal latitudes are interpreted to be planetographic, 2) 
IAUZ000/ datasets of this latitude definition could be used directly with M 
IAU2003 3396.19 | 3376.20 |176.630 0 the appropriate ellipsoid. However, recent datasets in the Gy. 
= planetocentric/east coordinate system would have to be M 
: : reprojected for this purpose. The required resampling process (4) A 
Table 1: Excerpt of IAU Reports. [8], [9] may reduce data quality. It should be mentioned that working M 
with planetographic latitudes using a spherical reference surface (5)|ME 
IIC PRINCIPLES OF GRASS instead the ellipsoidal one would cause a loss of accuracy T 
3. CARTOGRAPHIC D A similar as described in 3.2.3 and APPENDIX A. (6) 
To implement a geodetic accurate and consistent database, : M 
which is essential for comparative analyses, all datasets have to 3.00 Planctocenttic Latitude: Accordinepto Duxbüry erat (DI 
be transferred to the same geodetic reference frame (i.e. the [2], all future Martian datasets are recommended to fit to the (e: 
€ ce € "s . . 
same coordinate system/projection, W, and reference surface). pladetocentriceas coordinate sr How eum GRASS | 
* assumes all latitudes to be planetographic. The only possibility p 
31 Coordinate Systems to directly work with planetocentric datasets is to use a sphere M 
, . s as the reference body. In this case planetographic and (i 
GRASS supports a variety of different coordinate systems and planetocentric latitude coincide. To conform to future datasets, (p 
projections. These can be divided in two groups: we chose this system/reference body for our GIS database. The 
T drawbacks of this solution are the reduced accuracy of distance T 
3.1.1 Planar Coordinate System — Map Projection: GRASS er area P p Ee due to the less e RET Cl | 
supports a large number of conventional (GRASS term "other") the planet by a sphere instead of a more precisely fitting biaxial 
map projections (e.g. Transverse Mercator) using metre, yard, — Cllipsoid (see 3.2.3) and the necessity to reproject older data As a 
etc. as map units. Measurements of distances, areas or volumes from the ellipsoid to a sphere. defini 
are quantified in map units and are flawed by distortions of the us : : ; spheri 
map projection. Under consideration of the errors, these 3-23 Estimation of Errors due to the Spherical Approxi- confo 
coordinate systems are suitable for investigations within local ation: The simplification that is made by choosing a spherical A unii 
regions close to the centre of projection GIS database for Mars (see chapter 3.2.1) would not influence 
point positions — the measurements of latitude and longitude GRAS 
3.L2 Ellipsoidal Coordinate System — Database Projection: itself — but distances and areas derived from these ellipsoidal forma 
On ellipsoidal and accordingly spherical bodies, angular units ~~ Parameters using spherical formulae. the fo 
are being used to define a certain location at the planets surface ls SN topog 
(GRASS term: "lat/lon"). Raster data then fit to the simple While. such errors at the equator are negligible — small (conv 
cylindrical database projection (also known as geographic or deviations occur due to the slight difference between spherical Planci 
zs . . 2 - Sg eT . 1 $ 1 1 : "dm 1 d v d 1 ÿ P / 0 
“unprojected”), ic. longitude and planetographic latitude are radius and equatorial axis —, distances differ by 0.6 % at ihe hand 
assigned to the GRASS internal planar x/y coordinate system (x poles. Areal errors are about twice as much and SE Up 10 1.2 files r 
- longitude; y — planetographic latitude). No map properties are 7. If larger latitude ranges are surveyed, the particular errors no res 
preserved as with most conventional map projections. The are intermediate accordingly. In conclusion it should be pointed feces 
simple cylindrical database projection is similar to the spherical Out. that the errors act systematically and measurements give latituc 
form of the cylindrical equidistant map projection. almost too large results that increase with latitude. For à straig 
detailed Investigation — including graphs of such errors versus Datas 
In contrast to measurements within a planar coordinate system, latitude — see Appendix A. correc 
distances in the latitude/ longitude coordinate system are $i GRAS 
determined by true geodetic measurements and therefore no | 3-3 Vertical Reference conve 
errors are imposed. For this reason, we opted to base our . ; 
p ; P ; Xe As the GRASS database already is projected (i.e. two 
database on this coordinate system. Nonetheless, GRASS was S : : ; Patti s Subse 
developed for the terrestrial geographic coordinate system dimensional), no vertical reference is required in the first place, heade 
Which holds Some differences cor ihe JAU systems for Mat. ; though topographic elevation may be implemented as attribute GRAS 
t information. We incorporated the MOLA Mission Experiment Table 
* Longitude ranges from -180? to +180° increasing east Gridded Data Record (MEGDR) topography [11] dataset (sce planet 
a e" D. oT chapters 2.3 and 5). 
e Latitude is always assumed to be planetographic. ap : ) from 
map 
CEN alten qu A ct, ee Never 
“img definitions of the prime meridian (W;) are not 4. IMPLEMENTATION 
supported. 
* 
2 : M: 
We implemented a number of global datasets for Mars, " M. 
assortments of which are listed in Table 2 along with their ** Th 
cartographic parameters. Va