Disaggregating geological mapping units using expert

Laboratory of Geo-Information Science and Remote Sensing
  Education
  Research
  Publications
  Models
  News & Calendar
  News
  Calendar
  Archive
  News
  Calendar
  2011
  2010
  2009
  2008
  2007
  2006
  2005
  2012
  Staff
  Equipment
  Contact details
  Workshops

28 Aug 2008 10:00 - 28 Aug 2008 10:30
Unit: Laboratory of Geo-Information Science and Remote Sensing
Location: Gaia 1
Organisation: Wageningen University

 By Linda van der Toorn  

Abstract:  
Digital soil mapping aims to derive new data from existing data sources, hence a minimal
amount of extra fieldwork is needed. Topography and landscape position are often linked to
soil development. Since a digital elevation model (DEM) represents topography, and a lot of
topographic parameters can be derived from it, the DEM plays a very important role in this.
Catenary soil development is a difference in soil development on the same parent material
due to differences in topography and differences in developing time. Therefore different
catena units can be created by using the topographic parameters derived from the DEM. In
this research an effort is made to disaggregate a geological map into smaller catena units,
using different derivatives of a DEM, expert rules and existing field points. The used
derivatives are: slope, flow accumulation and mean curvature. The expert rules (on flow
accumulation, slope and area) are based on years of extensive fieldwork in the study area in
Southern Spain. Three different DEMs are used to compare the effect of resolution (10X10
[m] and 25X25 [m]) and DEM origin on the final disaggregated results. Different methods and
combinations of the used topographic parameters are used to create different catena
models. The most time consuming model, which stepwisely increases or decreases the
values of the expert rules until the pre-described area is reached for each unit, performed
best. A less computational consuming model, which only used the expert rules on flow
accumulation and slope, performed less good, but could be a good alternative when
comparing the computation times of both models. Mean curvature can estimate crest and
valley positions best, whereas slope is better to estimate the positions in between the crest
and the valley. It appeared that DEM derivatives are quite accurate in predicting actual field
conditions. Striking is the fact that curvature appears to be a very important topographic
parameter for predicting the catena positions, while on the other hand this is also the
parameter which is most difficult to relate with the actual field measurements. When
validating the models, it appeared that in all cases there are about 25 cells near each field
point necessary, to get good validation results. DEM resolutions did not make a large
difference in this. Possibly these cells are necessary to get a get hold on the shape of a
landscape and to avoid noise in the DEM. When comparing the distances needed for good
validation results, the DEM with the higher resolution performs better. Also the quality of the
DEM with the 10X10 [m] resolution appears to be higher, since the field measurements fit
better to this DEM.
Print this activity