1423
A COMPARISON BETWEEN CANADIAN DIGITAL ELEVATION DATA (CDED)
AND SRTM DATA OF MOUNT CARLETON IN NEW BRUNSWICK (CANADA)
Frédéric HAPPI MANGOUA, Kalifa GOÏTA
Frederic.happi.mangoua - usherbrooke.ca, Kalifa.Goita - USherbrooke.ca
Centre d’Applications et de Recherches en Télédétection (CARTEL), Université de Sherbrooke,
2500 Boulevard de l’Université, Sherbrooke, Québec, J1K 2R1 Canada
Commission IV, WG IV/9
KEY WORDS: Relative Accuracy, Canadian Digital Elevation Data (CDED), SRTM, Icesat, Lidar, Mount Carleton
ABSTRACT:
Digital elevation models (DEM) are basic part of the information about an area. Knowledge about DEM quality is important for their
use in management projects, engineering projects and géomorphologie studies. Errors and imprecision of DEM can impact a lot on
the resulting models one makes or uses in a project. It’s essential to have accurate topographical information from a DEM.The Centre
for Topographic Information (CIT) of Natural Resources Canada produced a particular DEM for the Canada country. These are
called Canadian Digital Elevation Data (CDED). The CDED DEM has been used for many types of studies and projects mostly in
Canada.The relative accuracy of Canadian Digital Elevation Data of Mount Carleton was assessed using Shuttle Radar Topographic
Mission (SRTM) model and profiles/points from Geoscience Laser Altimeter System (GLAS) onboard ICESat. This relative
accuracy was examined as a function of surface slope and land cover. Specifically, we analyzed the effect of slope and vegetation
type on topographic information (elevation).The particularity of Mount Carleton is that Mount Carleton is the highest mountain in the
Maritimes Provinces with the peak at 817 meters and it’s heavily wooded. More than 50% of the vegetation is dominated by
coniferous trees and the average slope is 5.45° ± 4.72°. Terrain was segmented into three sloping regions (< 5°, 5° < slope < 15°, >
15°), and also was segmented to aspect regions, standardized to eight geographical directions.From the correlation between CDED
and SRTM, we founded a systematic error of less than 2.0 m in absolute value with a standard deviation of around 16 m. We
observed that those values are slope-dependent and the influence of their orientation is not significative. A relative influence is
observed for the north directions. The broadleaf is the species which has the highest concentration of errors and the obtained root-
mean-square error (RMSE) for SRTM model comparing to CDED fulfill the 16 m RMSE specification mission. ICESat
profiles/points used in the study confirmed the good accuracy of CDED.
1. INTRODUCTION
The Canadian Digital Elevation Data (CDED) consists of an
ordered array of ground elevations at regularly spaced intervals.
The source digital data for CDED at scales of 1:50,000 are
extracted from the hypsographic and hydrographic elements of
the National Topographic Data Base (NTDB) or various scaled
positional data acquired from the provinces and territories
(Geobase, 2008). The Customer Support Group at the Centre
for topographic information (CIT-Sherbrooke) specifies that the
tile reference scheme for CDED models the National
Topographic System (NTS) mapping series. That the coverage
for each CDED corresponds to half an NTS tile, which means,
there is always a western and eastern CDED cell to a whole
NTS tile. And that cell coverage varies according to the
geographic area. The North American Datum 1983 (NAD83) is
used as the reference system for planimetric coordinates.
Elevations are orthometric and expressed in reference to mean
sea level, Canadian Vertical Geodetic Datum 1928 (CVGD28).
For more information about the CDED level 1, see
http://www.geobase.ca/doc/specs/pdf/GeoBase_product_specs_
CDED_3.pdf.
Control Markers (ASCMs) elevation data. The ASCM is in
good agreement with ICESat, although local terrain effects were
not been considered. The effects of snow depth, vegetation,
slope, needs to be studied in more detail. Those factors have
been considered in the various applications of CDED. The
applications on which CDED have been used are not limited to
mapping. Considering the multiple uses of DEM data,
especially for the use in predictive models, it is important to
consider the accuracy of topographic input data that are used
(Thompson J.A et al, 2001). CDED should be considered as
input data. Accuracy refers to the closeness of an observation to
a true value (Maune et al., 2001). Accuracy is computed by a
comparison of elevations in the DEM with corresponding
known elevations. The root-mean-square error (RMSE) statistic
is used to describe the vertical accuracy of a DEM (Eq. (1)),
encompassing both random and systematic errors introduced
during the production of data (ASPRS, 1990; Maune et al.,
2001).
RMSE =
(Zi - z'i) □
(1)
Beaulieu et al. (2007) assessed the accuracy of a new
production of the Canadian Digital Elevation of the North
(Canada), using ICESat LIDAR, obtaining, for a group of 21
CDED an accuracy of 0.34 m± 6.22 m - i.e. 10 m at 90%
confidence level. Braun A. et al. (2007) evaluated the
differences among SRTM data, ICESat data and Alberta Survey
Where Zi is the DEM elevation of a test point, z'/, the true
elevation and n the number of test points.
The comparison between CDED level 1 and SRTM data of
Mountain DEM over Mount Carleton will be assessed using