Journal of Geographic Information System, 2013, 5, 325-336
http://dx.doi.org/10.4236/jgis.2013.54031 Published Online August 2013 (http://www.scirp.org/journal/jgis)
LAN Tool: A GIS Tool for the Improvement of Digital
Elevation Models Using Drainage Network Attributes
Alexandra Gemitzi1*, Odysseas Christou2
1Department of Environmental Engineering, Faculty of Engineering, Democritus University of Thrace, Xanthi, Greece
2Geoinfo Cyprus Limited, Larnaca, Cyprus
Received March 24, 2013; revised April 24, 2013; accepted May 24, 2013
Copyright © 2013 Alexandra Gemitzi, Odysseas Christou. This is an open access article distributed under the Creative Commons
Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is
properly cited.
ABSTRACT
Digital Elevation Models (DEMs) are constructed using altitude point data and various interpolation techniques. The
quality and accuracy of DEMs depend on data point density and the interpolation technique used. Usually however,
altitude point d ata especially in plain areas do no t provide realistic DEMs, mainly due to errors produced as a result of
the interpolation technique, resulting in imprecise topographic representation of the landscape. Such inconsistencies,
which are mainly in the form of surface depressions, are especially crucial when DEMs are used as input to hydrologic
modeling for impact studies, as they have a negative impact on the model’s performance. This study presents a Geo-
graphical Information System (GIS) tool, named LAN (Line Attribute Network), for the improvement of DEM con-
struction techniques and their spatial accuracy, using drainage network attributes. The developed tool does not alter the
interpolation technique, but provides higher point density in areas where most DEM problems occur, such as lowland
areas or places where artificial topographic features exist. Application of the LAN tool in two test sites showed that it
provides considerable DEM improvement.
Keywords: Digital Elevation Model (D EM ); Geographical Info rmati on Systems (GIS); Drainage Network; Spati al
Interpolation; Hy d r ologic Mode l ing
1. Introduction
Digital Elevation Models (DEMs) are numerical repre-
sentations of areas on the earth’s surface produced ap-
plying interpolation techniques on altitude point data.
DEMs provide vital input to several scientific areas and
have numerous applications, among others in telecom-
munications, military services, location based services
and environmental services. One of the most important
applications of DEMs is their use in hydrologic modeling.
As climate and land use changes have already brought
serious impacts on local communities such as floods,
droughts, soil erosion and pollution from point and
non-point sources, scientists and decision makers are
concerned on having realistic hydrologic models capable
of simulating runoff, and sediment and nutrient loadings,
in order to determine areas and ecosystems that are prone
to be affected in the future.
The importance of DEM quality on hydrologic studies
is already depicted by many researchers. Legleiter and
Kyriakidis [1], indicated the problems usually encoun-
tered while constructing a DEM from a number of dis-
crete altitude data points acquired through field survey
and pointed out the fact that sophisticated methods of
spatial prediction are no substitute for field data. The
impact of DEM quality in hydrologic modeling, hazard
modeling, stream network delineation, floodplain bound-
aries is highlighted in many studies [2-6].
Much research is also focused on detecting and im-
proving the accuracy of DEMs and comparative studies
have been conducted on the accuracy of the results of
various interpolation techniques [7-12]. Several tech-
niques have been developed for error minimization and
DEM improvement including data fusion and other so-
phisticated techniques which however require the avail-
ability of multi-source DEM products [13,14]. Other
methods provide combination or adjustments of various
interpolation mod e ls [15,16].
Early studies have dealt with the problem of improv-
ing DEM quality using drainage enforcement algorithms
incorporating stream line data. Ch en et al. [17], presented
*Corresponding a uthor.
C
opyright © 2013 SciRes. JGIS

A. GEMITZI, O. CHRISTOU 326
a comparative study of the drainage constrained methods,
categorizing them in three main groups, i.e.: stream
burning, surface fitting, and constrained-TIN algorithms.
The stream burning algorithms apply a raster representa-
tion of a vector stream network in order to capture
known hydrological features into a DEM [18,19]. The
most popular surface fitting algorithm was presented by
Hutchinson [20], known as ANUDEM algorithm, in-
tending to remove pits from DEMs. In this methodology,
irregularly spaced elevation data points, streamline and
contour line data are used together with a finite differ-
ence interpolation technique. In the constrained TIN ap-
proaches, TINs are used instead of square grid cells in
the DEM. Recently, Zhou and Chen [21], presented the
compound method, constructing a drainage-constrained
TIN, optimized to keep the important terrain features and
slope morphology. Such algorithms are mainly targeting
at DEM generalization, acting as correctors to the already
constructed DEM, by altering point altitude where nec-
essary and utilizing a specific interpolation method each
time. In that way, although all previously mentioned
methods provided DEM improvements, they achieved it
through reconfigurations of certain interpolation methods
and therefore they focused on a specific interpolation
method. However none of the known interpolation me-
thods provide accurate predictions in all cases and the
development of a DEM improvement methodology using
a certain interpolation technique is quite restrictive. This
paper presents a DEM improving methodology that work s
before any interpolation is attempted and aims at im-
proving input data point density incorporating stream line
data into the original altitude point data set without al-
tering any of the existing data points. Th is is achieved by
assigning automatically an altitude attribute to the river
network points, based on the intersections of the river
lines with the altitude con tours and by in terpolating add i-
tional altitude points between intersection points. In that
way, errors in the drainage network which are common
due to artificial or even natural changes in topography,
artificial drainage network construction, low density of
point altitude data in lowland areas, or even errors pro-
duced by the interpolation may well be eliminated. The
major difference from the previously mentioned methods
for DEM improvement is that the present tool does not
alter any of the initially existing data points. As it acts
prior to DEM construction, it is not dedicated to any in-
terpolation method and therefore it provides greater fle-
xibility for the user to choose any interpolation method
fits better to the case study each time.
Legleiter and Kyriakidis [1], mentioned that root mean-
square error of the predictions provided by interpolation
methods is directly proportional to the spacing between
surveyed cross-sections, even in a reconfigured channel
with a relatively simple morphology. In this context and
because of the fact that the methodology presented herein
provides higher density of point altitude data in the
drainage network, it seems that it is a useful alternative
for DEM improvement especially in cases where hydro-
logic modeling is of prime interest.
The methodology is demonstrated in two parts of the
Vosvozis river basin, in northern Greece (Fi gur e 1 ).
2. Materials and Methods
2.1. Study Area
The study area, Vosvozis river basin, located in northern
Greece covers 357 km2 (Figure 1). The southern lowland
part is an agricultural area. The river course in that part
has undergone many interventions, thus changing com-
pletely the natural drainage pattern during the last dec-
ades. Moreover, many artificial drainage tiles have been
constructed. The northern mountainous part is a forested
area that has remained largely unchanged.
2.2. Methodology and Theoretical Background
LAN tool was developed using the Mapinfo MapBasic
programming language and works as a seamless add on
tool within Mapinfo Professional ver. 7 or later. The idea
is to provide a higher point density as predefined by the
user. In our case linear objects represent the drainage
network of an area. Nevertheless, the same methodology
Figure 1. Location map of the study area.

A. GEMITZI, O. CHRISTOU 327
may well be applied for other linear features, e.g. for the
road network of an area, where enhanced point density is
required.
In any case the user is prompted to provide two vector
files, one with the altitude contours that will be used for
the construction of the DEM and one with the linear ob-
jects that will be used for the production of new altitude
data points (Figure 2).
Subsequently, the user chooses the required point dis-
tance to be produced on the linear objects (Figure 3).
Then LAN tool detects all intersecting points of the
given linear objects with the altitude contours as shown
in Figure 4 and creates nodes at those points. In order to
detect those intersections the well known sweep line al-
gorithm is used [22]. This algorithm uses a sweeping
vertical line from left to right in order to detect the k in-
tersections of n line segments. The status of this sweep
line changes constantly as the line moves and is defined
by the set of segments intersecting it each time. The
sweep line changes its status every time it passes an end
point of a line segment or an intersecting point. As all
data points are processed in sorted order the in tersections
are also provided in sorted order too. The advantage of
this algorithm is that unlike conventional intersection
search algorithms that require O(n2) time, the sweep line
algorithm requires O((n + k)logn) time to run and there-
fore it provides a fast and robust way to define intersec-
tion poi nt s.
Figure 2. Selection of input files within LAN tool.
Figure 3. Selection of input parameters for LAN tool.
Figure 4. Intersection of altitude contour lines with river
network.
Altitude values are assigned to those intersecting nodes
based on the intersecting contour. Following, each line
segment between those newly produced nodes is divided
into equally spaced parts, and a new node is created for
each one, according to the user predefined maximum
required node distance. Then, LAN tool assigns altitude
values to those new nodes performing either linear or
cubic spline interpolation as shown on Figure 5, based
on user’s choice.
Linear interpolation fits a different linear function be-
tween each pair of existing data points, i.e., nodes of
known altitude and returns the value of the relevan t func-
tion at the interpolated points. The cubic spline fits a dif-
ferent cubic function between each pair of existing data
points. Figure 5 shows an example of the application of
those two different techniques. The produced nodes are
spaced at 5 m. It is true that the cubic spline produces a
more realistic representation of the topography; however
when the distance between original nodes is small (sec-
ond and third point on Figure 4) then the two methods
produce almost identical results. Taking into account the
fact that linear interpolation is much faster, it depend s on
the user to choose the optimum method, based on the
density of the original data points. After the new nodes
have been assigned altitude values, LAN tool creates a
new vector file containing all original and newly pro-
duced data points, in order to be used for DEM construc-
tion using conventional interpolation techniques.
It should be noted herein that also in the case of low-
land areas where pits are often a problem when applying
a hydrologic model, LAN tool is a useful alternative to
pit removal algorithms used in most commercial GIS
software, after DEM construction.
The flow chart of the developed software is shown on
Figure 6.
In order to create the DEM for the study area, conven-
tional topographic maps of various scales were digitized.
The northern mountainous part was digitized from con-
ventional maps of the Hellenic Military Geographical
Service of scale 1:25,000 whereas for the southern plain
topographic maps of scale 1:5000 were used in order to
acquire higher density altitud point data in that part of e

A. GEMITZI, O. CHRISTOU 328
Figure 5. Cross sectional diagram of a river showing the results of linear and cubic spline interpolation.
Figure 6. Flow chart of the developed softwar e.
the study basin. All computations were performed using
the GIS software MapInfo Professional ver. 9 and its
raster tool Vertical Mapper ver. 3.1.
The usefulness and accuracy of LAN tool are demon-
strated in two sub areas of the study region, i.e., one sub
basin in the northern mountainous part which will be
referred hereafter as TA1 (Test Area 1) and one in the
southern lowland area, which will be referred as TA2

A. GEMITZI, O. CHRISTOU 329
(Test Area 2) (Figure 1). DEMs were constructed using
the kriging interpo latio n meth od with an d withou t th e u se
of LAN tool and their accuracy was checked using vari-
ous tests.
Kriging is based on the assumption that for a given
point data set those that are close to each other are corre-
lated whereas those at a further distance are statistically
independent [23]. Its basic idea is equivalent to inverse
distance weighted interpolation but instead of using
weights based on an arbitrary function of distance in or-
der to compute an interpolated point, the weights used in
kriging are based on the model variogram. The model
variogram is derived after constructing the experimental
variogram from the point data set to be interpolated, like
the one presented in Figure 7.
The experimental variogram is calculated by deter-
mining the variance of each data point in the data set
with respect to all other data points and by plotting the
variances (or semivariances) versus distance between
points. The model variogram is then defined as a simple
mathematical function that best fits the experimental
variogram. As it can be seen on Figure 7, for small
separating distances the variance of the variable to be
interpolated is small whereas after a certain distance the
variance becomes random. The model variogram is used
to compute the weights used in kriging in order to evalu-
ate an interpolated value F of the variable f based on the
following equation:

n
ii
i1
Fx,y wf

 (1)
where n in the number of points in the set, fi are the val-
ues of those points and wi are the weights assigned to
each data point. In that way, if a new interpolated value P
it to be calculated from three neighboring points P1, P2,
P3 then w1, w2 and w3 have to be determined solving a
system of 3 equations based on the values of the model
variogram evaluated at a distance equal to the distance
between points i and j.
There are various types of kriging. The one described
above is known as ordinary kriging. Universal kriging is
the kriging interpolation methodology that assumes data
stationarity, i.e. the average value of the point data re-
mains the same everywhere. Carter and Shankar [24],
supported the use of kriging, stating several advantages
of this approach: 1) predictions are based on the
variogram; 2) the kriging system is constructed so as to
minimize the variance of prediction errors while ensuring
unbiasedness and 3) the error variance provides an indi-
cation of the uncertainty associated with each prediction.
In every case, however, the LAN tool may well be used
irrespective of the selected interpolation technique as it
does not interfere with it, but enhances the altitude point
data density in certain locations.

A. GEMITZI, O. CHRISTOU 330
Figure 7. Comparison of semi variograms produced with and without the use of LAN tool, to the semi variogram with meas-
red altitude data. u

A. GEMITZI, O. CHRISTOU 331
3. Results and Discussion
3.1. Application to TA1
Figure 8 shows the altitude contours and the drainage
network of TA1. In order to test the efficiency of the
proposed methodology, DEM construction in TA1 was
achieved with and without the use of LAN tool. The
kriging interpolation technique was used in either case,
as mentioned previously. The acquired results were com-
pared to measured values of altitude performed using a
Trimble GPS Total Station R4.
Two independent tests were then performed; the first
one compares the semi variograms produced with 50
measured altitud e points, to those produced with the data
points before and after the use of LAN tool in a small
portion of TA1 highlighted as a red rectangle on Figure
8. In each case the lag distance used was the mean dis-
tance between data points. Figure 7 shows the semi
variograms produced using only measured altitude points
and those produced with and without the use of LAN tool.
Figure 8. Map of TA1. Red rectangle highlights the area where semi variograms with and without the use of LAN tool, were
compared to the semi variogram with measured data. Black spots indicate locations where altitude measurements on a main
iver branch were conduc ted. r

A. GEMITZI, O. CHRISTOU 332
It can be seen that all three semi variograms present the
same trend but the semi variograms of the measured
points and that using points produced with LAN tool are
almost identical.
The second test compares the altitud e values produced
with and without the application of LAN tool to meas-
ured altitude values on 100 randomly sampled data po-
ints located on one of the main branches of the drainage
network of TA1 shown as black points on Figure 8. Fig-
ure 9(a) shows the river profiles produced with the
measured altitude points and with and without the use of
LAN tool. As it can be seen the profile produced using
LAN tool is much closer to the measured one.
The mean absolute error (MAE) was calculated for the
two profiles produced with and without the use of LAN
tool, based on the fol l o wi n g f ormula:
N
tt
t1
1ˆ
MAEZ Z
N

 (2)
Zt being the interpolated altitude valu es, t the meas-
ured altitude values and N the nu mber of points, which in
our case is 100.
ˆ
Z
The MAE for the profiles produced with and without
the use of LAN tool is 0.58 m and 2.39 m respectively.
Figures 9(b) and (c) present the DEMs produced with-
out and with the use of LAN tool, respectively. From a
(a)
(b) (c)
Figure 9. Cross sectional plot of a river branch produced with and without the use of LAN tool and with measured altitude
alues. v

A. GEMITZI, O. CHRISTOU 333
first point of view the DEMs produced using either
method are very similar, but based on the comparison to
the measured values, the LAN tool improved the pro-
duced DEM.
3.2. Application to TA2
In contrast to TA1, which is almost unaffected by human
intervention, TA2 has undergone changes of the natural
route of the drainage network, changing the main course
of Vosvozis river and constructing irrigation and drain-
age channels. Additionally, Ismarida lake which is an
important ecosystem, has decreased seriously in the last
decade due to excess pumping. Thus, the topography of
the area is also altered, which is not represented in the
conventional maps that cannot be updated continuously.
The main problem that arose is that when the DEM of
TA2 was constructed, using topographic data from those
conventional maps, the drainage network that was de-
lineated using hydrologic models did not correspond to
the existing network. Moreover, in plain areas, even in
cases where no major human interventions have taken
place, altitude contours are sparse, thus even the small
errors associated with the interpolation technique may
result in depressions that hinder flow or a drainage pat-
tern which deviates considerably from the existing one.
In those cases, the use of LAN tool proves to be quite
useful for DEM correction. As LAN tool requires the
river network to be provided, an ASTER image of the
study area acquired in August 2009 was used for the de-
lineation of the river network (Figure 10).
Figure 10. Aster image acquired on August 2009, showing
the river network of TA2.
Topographic data were digitized using conventional
topographic maps of the Hellenic Military Geographical
Service of scale 1:5000. Kriging interpolation was used
to create a DEM, with and without the use of LAN tool.
The produced DEMs were introduced to one of the most
popular hydrologic models, i.e. Soil and Water Assess-
ment Tool (SWAT) [25], in order to delineate automati-
cally the drainage network. The results were compared to
the existing drainage network. Figure 10 shows the
drainage network delineated from the ASTER image.
Figure 11(a) shows the drainage network produced by
automatic delineation using the DEM without the use of
LAN tool (Figure 11(b)).
Figure 12(a) shows the drainage network produced by
automatic delineation using the DEM with the use of
LAN tool shown on Figure 12(b). Comparison of the
produced results shows that the river network delineated
based on the DEM with the use of LAN tool is much
closer to the real one. Additionally, one can easily dis-
tinguish the artificial irrigation pond shown on the north
east of Figure 12(b), which is incorporated in the DEM
using LAN tool. Thus LAN tool proved to be quite effec-
tive in TA2.
At this point the find ings of the research conducted by
Chaplot et al. [12] should be mentioned. They found that
irrespective of the surface area, landscape morphology
and sampling density, few differences existed between
the employed interpolation techniques if the sampling
density was high. At lower sampling densities, in con-
trast, the performance of the techniques tended to vary.
Within this context the usefulness of LAN tool is unam-
biguous at it increases sampling density in specified lo-
cations where most of the DEM inaccuracies tend to exist.
Besides the fact that kriging was used for demonstration
purposes, the same improved results on the DEM are
expected with all other interpolation techniques, as LAN
tool does not alter the interpolation itself, but it provides
a higher point dat a de nsi t y in areas of inte res t .
4. Conclusions
This paper demonstrates the use of a newly developed
GIS tool, namely LAN tool, which helps in the im-
provement of DEMs. The main idea is to increase point
density data in areas where conventional interpolation
produces problematic results, such as plain areas or river
beds. For this purpose information provided by the de-
lineation of linear features, such as river network is used.
Two applications of LAN tool were presented, one in a
mountainous area and the other in a lowland area. Inde-
pendent tests were performed in order to prove the effec-
tiveness of LAN tool. In the first case, i.e. TA1, the tests
were based on field altitude measurements. Results
showed that a considerable improvement in DEM was

A. GEMITZI, O. CHRISTOU 334
(a) (b)
Figure 11. (a) River network produced with automatic delineation using SWAT model using DEM without the use of LAN
tool; (b) DEM produced without the use of LAN tool.
(a) (b)
Figure 12. River network pr oduced with automatic delineation using SWAT model using DEM with the use of LAN tool; (b)
DEM produced with the use of LAN tool.
achieved, as indicated by the minimization of MAE.
In the second case, i.e. TA2, where serious interven-
tions on the drainage network had taken place, the im-
provement using LAN tool was obvious, as the produced
DEM represented accurately the current pattern of the
drainage network.
As this application is merely focused on hydrologic
surveys, the drainage network has been used as the line

A. GEMITZI, O. CHRISTOU 335
attribute network. Analogous applications may well be
developed using LAN tool with other line attribute net-
works, like roads o r any other linear feature.
5. Acknowledgements
The field measurements conducted in the present work
were part of the project: “Water resources management
in Eastern Macedonia and Thrace” funded by the Tech-
nical Chamber of Greece (project code: 2237 Democritus
University of Thrace).
REFERENCES
[1] C. Legleiter and P. Kyriakidis, “Spatial Prediction of Ri-
ver Channel Topography by Kriging,” Earth Surface Pro-
cesses and Landforms, Vol. 33, No. 6, 2008, pp.841-867.
doi:10.1002/esp.1579
[2] V. Chaplot, “Impact of DEM Mesh Size and Soil Map
Scale on SWAT Runoff, Sedimen t, and NO3-N Loads Pre -
dictions,” Journal of Hydrology, Vol. 312, No. 1-4, 2005,
pp. 207-222.
[3] L. Kalin, R. S. Govindarajua and M. M. Hantush, “Effect
of Geomorphologic Resolution on Modeling of Runoff
Hydrograph and Sedi mentograph over Small Watershe ds,”
Journal of Hydrology, Vol. 276, No. 1-4, 2003, pp. 89-
111. doi:10.1016/S0022-1694(03)00072-6
[4] A. R. Darnell, A. A. Lovett, J. Barclay and R. A. Herd, “An
Application Driven Approach to Terrain Model Construc-
tion,” International Journal of Geographical Information
Science, Vol. 24, No. 8, 2010, pp. 1171-1191.
doi:10.1080/13658810903318889
[5] J. Kiesel, N. Fohrer, B. Schmalz and M. J. White, “In-
corporating Landscape Depressions and Tile Drainages of
a Northern German Lowland Catchment into a Semi-Dis-
tributed Model,” Hydrological Processes, Vol. 24, No. 11,
2010, pp. 1472-1486. doi:10.1002/hyp.7607
[6] H. Achour, N. Rebai, J. Van Den Driessche and S. Boua-
ziz, “Modelling Uncertainty of Stream Networks Derived
from Elevation Data Using Two Free Softwares: R and
SAGA,” Journal of Geographic Information System, Vol.
4, No. 2, 2012, pp. 153-160. doi:10.4236/jgis.2012.42020
[7] D. Weber and E. Englund, “Evaluation and Comparison
of Spatial Interpolators,” Mathematical Geology, Vol. 24,
No. 4, 1992, pp. 381-391. doi:10.1007/BF00891270
[8] D. Weber and E. Englund, “Evaluation and Comparison
of Spatial Interpolators II,” Mathematical Geology, Vol.
26, No. 5, 1994, pp. 589-603. doi:10.1007/BF02089243
[9] A. Carrara, G. Bi tell i and R. Carla, “Comparison of Tech-
niques for Generating Digital Terrain Models from Con-
tour Lines,” International Journal of Geographical In-
formation Science, Vol. 11, No. 5, 1997, pp. 451-473.
doi:10.1080/136588197242257
[10] S. M. Robeson, “Spherical Methods for Spatial Interpola-
tion: Review and Evaluation,” Cartography and Geogra-
phic Information Systems, Vol. 24, No. 1, 1997, pp. 3-20.
doi:10.1559/152304097782438746
[11] F. J. Aguilar, F. Agüera, M. A. Aguilar and F. Carvajal,
“Effects of Terrain Morphology, Sampling Density, and
Interpolation Methods on Grid DEM Accuracy,” Photo-
grammetric Engineering and Remote Sensing, Vol. 71,
No. 7, 2005, pp. 805-816.
[12] V. Chaplot, F. Darboux, H. Bourennane, S. Leguédois, N.
Silvera and K. Phachomphon, “Accuracy of Interpolation
Techniques for the Derivation of Digital Elevation Mod-
els in Relation to Landform Types and Data Density,”
Geomorphology, Vol. 77, No. 1-2, 2006, pp. 126-141.
doi:10.1016/j.geomorph.2005.12.010
[13] S. J. Buckley and H. L. Mitchell, “Integration, Validation
and Point Spacing Optimization of Digital Elevation Mo-
dels,” The Photogrammetric Record, Vol. 19, No. 108,
2004, pp. 277-295.
doi:10.1111/j.0031-868X.2004.00287.x
[14] M. Karkee, B. L. Steward and S. A. Aziz, “Improving
Quality of Public Domain Digital Elevation Models through
Data Fusion,” Biosystems Engineering, Vol. 101, No. 3,
2008, pp. 293 -305.
doi:10.1016/j.biosystemseng.2008.09.010
[15] W. Z. Shi and Y. Tian, “A Hybrid Interpolation Method
for the Refinement of a Regular Grid Digital Elevation
Model,” International Journal of Geographical Informa-
tion Science, Vol. 20, No. 1, 2006, pp. 53-67.
doi:10.1080/13658810500286943
[16] O. Bonin and D. Rousseaux, “Di gital Terrain Model Com-
putation from Contour Lines: How to Derive Quality In-
formation from Artifact Analysis,” GeoInformatica, Vol.
9, No. 10, 2005, pp. 253-268.
doi:10.1007/s10707-005-1284-2
[17] Y. Chen, J. P. Wilson, Q. Zhu and Q. Zhou, “Comparison
of Drainage-Constrained Methods for DEM Generaliza-
tion,” Computers and Geosciences, Vol. 48, 2012, pp. 41-
49. doi:10.1016/j.cageo.2012.05.002
[18] W. Saunders, “Preparation of DEMs for Use in Environ-
mental Modelling Analysis,” In: D. Maidment and D.
Djokic, Eds., Hydrologic and Hydraulic Modelling Sup-
port with Geographic Information Systems, Environmen-
tal Systems Research Institute Inc., Redlands, 2000, pp.
29-51.
[19] J. N. Callow, K. P. Van Niel and G. S. Bog gs, “How D oe s
Modifying a DEM to Reflect Known Hydrology Affect
Subsequent Terrain Analysis?” Journal of Hydrology,
Vol. 332, No. 1-2, 2007, pp. 30-39.
doi:10.1016/j.jhydrol.2006.06.020
[20] M. F. Hutchinson, “A New Procedure for Gridding Ele-
vation and Stream Line Data with Automatic Removal of
Spurious Pits,” Journal of Hydrology, Vol. 106, No. 3-4,
1989, pp. 211-232. doi:10.1016/0022-1694(89)90073-5
[21] Q. Zhou and Y. Chen, “Generalization of DEM for Ter-
rain Analysis Using a Compound Method,” ISPRS Jour-
nal of Photogrammetry and Remote Sensing, Vol. 66, No.
1, 2011, pp. 38-45. doi:10.1016/j.isprsjprs.2010.08.005
[22] D. Souvaine, “Line Segment Intersection Using a Sweep
Line Algorithm,” Tufts University, 2005.
http://www.cs.tufts.edu/comp/163/notes05/seg_intersectio
n_handout.pdf
[23] J. C. Davis, “Statistics and Data Analysis in Geology,”
2nd Edition, John Wiley and Sons, New York, 1986.

A. GEMITZI, O. CHRISTOU 336
[24] G. S. Carter and U. Shankar, “Creating Rectangular Ba-
thymetry Grids for Environmental Numerical Modelling
of Gravel-Bed Rivers,” Applied Mathematical Modelling,
Vol. 21, No. 11, 1997, pp. 699-708.
doi:10.1016/S0307-904X(97)00094-2
[25] M. Winchell, R. Srinivasan, M. Di Luzio and J. Arnold,
“ARCSWAT 2.0 Interfac e for SWAT 2005, User’ s Gui d e, ”
Blackland Research Center, Texas Agricultural Research
Station and USDA Agricultural Research Service, 2008.