Precision Agriculture, 4, 179±192, 2003
2003 Kluwer Academic Publishers. Manufactured in The Netherlands.
Comparison of DEM Data Capture and
Topographic Wetness Indices
FRANK SCHMIDT firstname.lastname@example.org
Foerstereistrasse 38 01099 Dresden, Germany
ANDREAS PERSSON email@example.com
Swedish Institute of Agricultural and Environmental Engineering (JTI), P.O. Box 7033,
750 07, Uppsala, Sweden
Abstract. Digital elevation models (DEMs) can be captured and analysed by various methods. Elevation
capturing with RTK-GPS and airborne laser scanning is presented and evaluated in terms ofheight accuracy of
raw data and interpolated DEMs for study sites in Sweden and Germany. Applications for precision agriculture
are based on the connection ofland surface and the movement ofwater in the landscape. Three methods of
deriving potential flow accumulation from DEMs are discussed. Results indicate that the Topographic Wetness
Index can be used to assess the pattern ofpotential soil moisture on a field and changes in soil texture caused by
erosion processes. The quality ofthe soil moisture assessment varies with both flow algorithm and terrain
characteristics. Correlations up to r
0.64 were found. Best results were obtained on undulating farmland using
the DEMON algorithm and a formbased approach. However, in areas with low relief, the concept did not lead to
valuable soil moisture maps.
Keywords: GPS, laser scanning DTM, topography, wetness index
Terrain reliefcontrols the movement ofwater in a landscape. It influences the spatial
pattern ofsoil attributes and is one ofthe most important natural factors causing
heterogeneity on arable land and yield (Afyuni et al., 1993, Stone et al., 1985). The
functions of the terrain can be represented using digital elevation models (DEM). The
DEM is a stable factor compared to other data sources needed for precision agriculture.
DEMs have been under investigation for agricultural applications for a number of years
(e.g. Bishop and McBratney, 2002; Nolan et al., 2000; Nugteren and Robert, 1999;
Russel et al., 2000; Yao and Clark, 2000) and will be more common with increasing
availability and quality in the near future. Digital terrain analysis can support the creation
of application maps for soil tillage, site-specific seeding, irrigation, fertilizing and
pesticide spreading. For this purpose, topographic attributes such as slope, aspect,
drainage area (flow accumulation) or the Topographic Wetness Index (TWI) have to be
derived from the DEM. The Wetness Index ln(A
/tan b) is a compound terrain attribute
calculated from specific catchment area of a point (A
) and the local slope gradient tan b.
The concept was first presented by Beven and Kirkby (1979) and further developed in the
1990s (Wilson and Gallant, 2000).
In this article, the importance ofsample geometry for the generation ofdigital terrain
models and algorithms for terrain analysis will be presented for two study sites and two