er 
De 
> 
18 
sat 
ial 
ur 
on 
nd 
ut 
nd 
yill 
tal 
  
  
Felkner, John 
e. comparison of predicted maps of land use change with an actual map of land use change derived from a 
classified 1999 Landsat 7 satellite image, through GIS map comparison methods. Subsequent evaluation will 
determine which combination of inputs into the tree classification (e.g. economic only, environmental only, or 
both) produces the most accurately predicted map. 
4 Data Sources 
4.1 Vectorization of Contour Lines and Road Networks 
Highly accurate contour maps obtained from the Thai government were first scanned into digital form. The contour 
lines and road networks were then vectorized — that is, converted to digital objects that are recognized by the GIS. 
4.2 Creation of Digital Elevation Models (DEMs) 
"Once the contour lines were vectorized, a continuous digital surface was fitted to these contour lines through 
interpolation. This computationally intensive process was completed using ArcInfo software on a unix platform (using 
the ArcInfo TopoGrid system). The result is a surface which can be divided into a regular grid of cells of any specified 
spatial resolution, with each cell having a value that corresponds to its elevation in meters above sea level. Once this is 
completed, a simple mathematical process can be run in the GIS to calculate the slope of each pixel. These digital slope 
and elevation models make it possible to consider slope or elevation in any GIS modeling. 
4.3 Creation of Digital Road Networks 
Once vectorized, the road networks became GIS digital objects made up of a series of road line segments (called “arcs” 
in GIS terminology) that together form a continuous digital road network. Each road segment was assigned a 
distance/cost value corresponding to the approximate average travel speed along that road. For example, the average 
travel speed along a major highway may average around 65 mph, while on a dirt road the average may be around 30-40 
mph. These distance/cost values were assigned to each road segment in the network depending on its road class, and 
distances along the network were calculated in terms of travel time (for example, radiating outward from a single point 
on the network in all directions along all roads for a distance of 2 hours travel time, taking into consideration the 
average speed of travel along the different road segments). 
In addition to average travel times depending on road class, certain road segments were be assigned higher distance/cost 
values if they crossed certain barriers (such as an international boundary or are on very steep slopes) under the 
assumption that it will take more than the average time to cross these barriers using these roads. 
These travel times along road networks were used in the creation of the continuous socio-economic models, described 
below, under the assumption that factors such as population and income are spatially distributed using road networks. 
4.4 Soil and Rainfall Data Sources 
In addition to data taken from the Thai government maps and the CDD data, soil sample data taken in Sisaket and 
Chachoengsao Provinces was used to interpolate soil quality maps for the Provinces using the kriging method (Myer 
1991; Bonham-Carter 1994) (Laslett, McBratney et al. 1987) to generate a continuous spatial interpolation which was 
then used as an input to the soil moisture index model. The soil moisture model also used data collected by the Thai 
government from rainfall gauges as a precipitation input. 
5 FUTURE DIRECTIONS FOR RESEARCH 
  
International Archives of Photogrammetry and Remote Sensing. Vol. XXXIII, Part B7. Amsterdam 2000. 437