GNR402: Introduction to Geographic Information 

Systems:
Terrain modeling and TIN

Surya S. Durbha, Ph.D.

Professor
CSRE, IIT(B)
 Need

In high relief areas, variables such as altitude,

aspect and slope strongly influence both human
and physical environments
a 3D data model is therefore essential
Digital Terrain Model (DTM) enables to derive
information on:
❑ height (altitude), aspect and slope (gradient)
❑ watersheds (catchments)
❑ solar radiation and hill shading
❑ cut and fill calculations
❑ etc.
 3 basic methods for representing a surface
DEM (digital elevation model): set of regularly spaced sampled ground points in
the x and y dimensions (although spacing not necessarily the same in each)
accompanied by an elevation measure (z dimension). The DEM terminology was
introduced by USGS. Two concepts used for determining elevation at points
within the grid cells:
Lattice: each point represents a value on the surface only at the center of the
grid cell
Surface grid considers each sample as a square/rectangular cell with a constant
surface value.

TIN (Triangulated Irregular Network) a set of adjacent, non-overlapping triangles

with x, y coordinates and z vertical elevations for their vertices, along with
topological relationship between the triangles and their adjacent neighbors.

Contour lines: lines of equal elevation, drawn at a given interval (e.g. every 6 or 25
feet)
The general term digital terrain model (DTM) may be used to refer to any of the
above surface representations when in digital form.
DEM sometimes used synonymously with DTM—don’t.
Source:Briggs GISC 6382 UT-Dallas
 Aster DEM
http://www.madmappers.com/msdetails/AsterDEM.htm
 DSM sources

Shuttle Radar Topographic Mission (SRTM) Feb 2000 30-90m

digital surface model represents the earth's

surface and includes all objects on it.
 2-5m Ikonos 1999

http://www.satimagingcorp.com/galleryimages/ikonos-high-
resolution-dem-eritrea.jpg
 LiDAR 10cm– 10m(Mt.St. Helens)

http://www.nasa.gov/centers/goddard/images/content/67409main_dh2.gif
Photogrammetry: Tatras, Slovakia2m
Photogrammetry: Tatras, Slovakia2m
Anaglyph- Malaspina
 Application - Flood Risk

3D height data changing water levels-danger

areas
 Gridded models

A Gridded Elevation Model is defined on the

basis of a domain partition into regular
polygons
 RSGs

The most commonly used gridded elevation

models are Regular Square Grids (RSGs) where
each polygon in the domain partition is a square

The function defined on each square can be a

bilinear function interpolating all four elevation
points corresponding to the vertices of the
square
RSG: an example
Calculations from a
Gridded DEM
 Slopeand Aspect
• Calculated from a grid of elevations (a
digital elevation model)
• Slope and aspect are calculated at each point
in the grid, by comparing the point’s elevation
to that of its neighbors
– usually its eight neighbors
– but the exact method varies
– in a scientific study, it is important to know exactly
what method is used when calculating slope, and
exactly how slope is defined
 Slope Definitions
• Slope defined as an angle
• … or rise over horizontal run
• … or rise over actual run
• various methods
– important to know how your favorite
GIS calculates slope
SlopeDefinitions (cont.)
 Slope
• measured from an elevation or bathymetry
raster
– compare elevations of points in a 3x3
neighborhood
– slope and aspect at one point estimated
from elevations of it and surrounding 8
points
• number points row by row, from top left from 1 to 9

1 2 3
4 5 6

7 8 9
 Typical Slope Calculation
The values of the center cell and its eight
neighbors determine the horizontal and vertical
deltas. The neighbors are identified as letters
from a to i, with e representing the cell for which
the aspect is being calculated.

The rate of change in the x direction for cell e is

calculated with the following algorithm:

[dz/dx] = ((c + 2f + i) - (a + 2d + g) / (8 * x_cellsize)

The rate of change in the y direction for cell e is calculated
with the following algorithm:

[dz/dy] = ((g + 2h + i) - (a + 2b + c)) / (8 * y_cellsize)

Typical Slope Calculation
 SlopeCalculation: Example
The cell size is 5 units (both and X and Y direction).
The default slope measure of degrees will be used.

The rate of change in the x direction for the center cell

e is:

[dz/dx] = ((c + 2f + i) - (a + 2d + g) / (8 * x_cellsize)

= ((50 + 60 + 10) - (50 + 60 + 8)) / (8 * 5)
= (120 - 118) / 40
= 0.05
SlopeCalculation: Example
( Contd.)
SlopeCalculation: Example
( Contd.)
Slope
 Aspect Calculation
The rate of change in the x direction for cell e
is calculated with the following algorithm:

[dz/dx] = ((c + 2f + i) - (a + 2d + g)) / 8

The rate of change in the y direction for cell e is

calculated with the following algorithm:

[dz/dy] = ((g + 2h + i) - (a + 2b + c)) / 8

Taking the rate of change in both the x and y direction for cell
e, aspect is calculated using:

aspect = 57.29578 * atan2 ([dz/dy], -[dz/dx])

 Aspect Calculation
The aspect value is then converted to
compass direction values (0-360 degrees),
according to the following rules:

if aspect < 0
cell = 90.0 - aspect
else if aspect > 90.0
cell = 360.0 - aspect + 90.0
else
cell = 90.0 - aspect
 Aspect Calculation: Example
The rate of change in the x direction for the center
cell e is:

[dz/dx] = ((c + 2f + i) - (a + 2d + g)) / 8

= ((85 + 170 + 84)) - (101 + 202 + 101)) / 8
= -8.125
The rate of change in the y direction for cell e is:

[dz/dy] = ((g + 2h + i) - (a + 2b + c)) / 8

= ((101 + 182 + 84) - (101 + 184 + 85)) / 8
= -0.375
Aspect Calculation:
Example
(Contd.)
Aspect
Aspect

The range 325deg- 45deg (broadly

north-facing) has been colored pale
orange, east-facing yellow, west-
facing blue and southerly pale green.
 Connectivity Function Example:
Viewshed Analysis

Image Source: Chrisman, Nicholas.(2002). 2nd Ed. Exploring Geographic Information Systems. p 198. fig. 8-14 .


 Polyhedral terrain models

- They can be used for any type of sampled pointset (regularly and
irregularly distributed)
- They can adapt to the irregularity of terrains
- They represent continuous surfaces
 Triangulated Irregular Networks
The most commonly used polyhedral terrain models are
Triangulated Irregular Networks (TINs), where each polygon of the
domain partition is a triangle
 TINs

Example of a TIN based on irregularly distributed data

 DEMs and TINs

DEM with sample points TIN based on same sample points

 Delaunay Triangulations

• Intuitively: given a set V of points, among all the triangulations

that can be generated with the points of V, the Delaunay
triangulation is the one in which triangles are as much
equiangular as possible
• In other words, Delaunay triangulations tend to avoid long and
thin triangles: important for numerical problems

t P

Does P lie inside t or on its boundary?

 Why Delaunay Triangulations (cont.d)

It has been proven that they generate the best

surface approximation (in terms of roughness)
independently of the z values

There are several efficient algorithms to calculate

them

 Why Delaunay Triangulations (cont.d)

❑ Delaunay triangulation has several advantages over other triangulation methods:

❑ The triangles are as equi-angular as possible, thus reducing potential numerical precision
problems created by long skinny triangles
❑ Ensures that any point on the surface is as close as possible to a node
❑ The triangulation is independent of the order the points are processed

 Delaunay Triangulations

Delaunay triangulation is a proximal method that satisfies the requirement

that a circle drawn through the three nodes of a triangle will contain no
other node

 TIN

The TIN model is attractive because of its simplicity

and economy and is a significant alternative to the
regular raster of the GRID model.
 TIN

❑ Irregularly spaced sample points can be adapted to

the terrain, with more points in areas of rough
terrain and fewer in smooth terrain
❑ an irregularly spaced sample is therefore more
efficient at representing a surface
❑ In a TIN model, the sample points are connected by
lines to form triangles
❑ Within each triangle the surface is usually
represented by a plane

 TIN
❑ By using triangles we ensure that each piece of the
mosaic surface will fit with its neighboring pieces - the
surface will be continuous - as each triangle's surface
would be defined by the elevations of the three corner
points
❑ But there will be sharp changes of slope at triangle
edges
❑ other possibilities exist, especially useful in finite
element modeling, involving curved surfaces and
quadrilaterals, that ensure no sharp changes of slope

 TIN
❑ For vector GISs, TINs can be seen as polygons having
attributes of slope, aspect, and area,
❑ with three vertices having elevation attributes and three
edges with slope and direction attributes
❑ The TIN model is attractive because of its simplicity
and economy

 TIN
❑ Certain types of terrain are very
effectively divided into triangles with
plane facets

❑ This is particularly true with fluvially-

eroded landscapes, however, other
landscapes, such as glaciated ones, are
not well represented by flat triangles

❑ triangles work best in areas with

sharp breaks in slope, where TIN
edges can be aligned with breaks, e.g.
along ridges or channels

 TIN
❑ Contours are a common source of digital elevation
data. In general all the vertices of the contour lines
are used as mass points for triangulation. In many
cases this will cause the presence of flat triangles in
the surface.

❑ Flat triangles are created whenever a triangle is

formed from three nodes with the same elevation
value

TIN
 TIN

The slope- The green areas indicate

Slope = 0 (flat triangles)

 TIN

How can we avoid the flat triangles ?

❑ By adding more mass points
❑ Generalizing the contours
❑ By adding break lines

 TIN
Break lines

Linear features which define and control surface behavior in terms of

smoothness and continuity are called break lines.
A line in a TIN that represents a distinct interruption in the slope of a surface,
such as a ridge, road, or stream. No triangle in a TIN may cross a breakline (in
other words, breaklines are enforced as triangle edges). Z-values along a
breakline can be constant or variable.

 TIN
❑ Creating a TIN model requires many choices:
❑ how to pick sample points
❑ In many cases these must be selected from some
existing, dense DEM or digitized contours
❑ Normally, a TIN of 100 points will do as well as a
DEM of several hundred at representing a surface

 TIN
❑ How to connect points into triangles
❑ How to model the surface within each triangle

❑ Almost always resolved by using a plane surface

 How to pick points

• Given a dense DEM or set of digitized
contours, how should points be selected so
that the surface is accurately represented?
 VIP Algorithm
• Each point has 8 neighbors, forming 4
diametrically opposite pairs, i.e. up and down,
right and left, upper left and lower right, and
upper right and lower left
• For each point, examine each of these pairs of
neighbors in turn
– Connect the two neighbors by a straight line, and
compute the perpendicular distance of the central
point from this line

 VIP ALGORITHM
Average the four distances to obtain a measure of
"significance" for the point
Delete points from the DEM in order of increasing
significance, deleting the least significant first
This continues until one of two conditions is met:
The number of points reaches a predetermined
limit
The significance reaches a predetermined limit

VIP ALGORITHM

 Triangulation Calculation

Given a set of points, calculate a triangulation

Particular properties, e.g., equi-angularity (Delaunay
triangulation)

 Watson’s algorithm (1981)

• Given a set V of points, calculate a Delaunay

triangulation with vertices at points of V

• Watson’s algorithm is one of the so-called on-line

methods: based on the modification of an existing
Delaunay triangulation when a new point is inserted

• In on-line methods, the first step consists of building a

Delaunay triangulation of the domain (containing all
data points)

• Then all points of V are added incrementally

 Watson: initial step

• In Watson’s algorithm, the initial triangulation of the

domain is built by considering a fictitious triangle
containing all points of V in its interior

Dataset V
 Watson: process

• After building the initial triangle, all points of V are

added one at a time

• Finally the initial triangle and all edges incident at its

vertices are deleted

• The main step in this algorithm is the insertion of a

new point in the current Delaunay triangulation

 Watson: insertion of a new point

• We call the influence polygon RP of a point P in a

triagulation T the union of all triangles of T whose
circumscribing circle contain P
• After inserting P in T, we update T by deleting all
edges internal to RP and by joining P with all the
vertices of RP

RP
P

 Watson: insertion of a new point

• We call the influence polygon RP of a point P in a

triagulation T the union of all triangles of T whose
circumscribing circle contains P
• After inserting P in T, we update T by deleting all
edges internal to RP and by joining P with all the
vertices of RP

RP
P

 Watson: insertion step

• Points in V are added one at a time in the current

Delaunay triangulation
 Watson: insertion step (cont.d)

• First vertex P1: the fictitious triangle is its influence

polygon RP1; join P1 with its three vertices
 Watson: insertion step (cont.d)
• Inserting the second vertex P2: calculation of RP1
 Watson: insertion step (cont.d)
• Inserting the second vertex P2: updating the current
triangulation
 Watson: insertion step (cont.d)
• Inserting new vertices
 Watson: insertion step (cont.d)
• Inserting new vertices
 Watson: insertion step (cont.d)
• Inserting new vertices
 Watson: insertion step (cont.d)
• Inserting new vertices
 Watson: insertion step (cont.d)
• Inserting new vertices
 Watson: insertion step (cont.d)
• Inserting new vertices
 Watson: insertion step (cont.d)
• Inserting new vertices
 Watson: insertion step (cont.d)
• Inserting new vertices
 Watson: insertion step (cont.d)
• Inserting new vertices
 Watson: insertion step (cont.d)
• Inserting new vertices
 Watson: insertion step (cont.d)
• Inserting new vertices
 Watson: final step
• Delete the fictitious triangle and all edges incident at
its vertices
 Watson: final step
• Delete the fictitious triangle and all edges incident at
its vertices
 Watson: final step
• Given V, Watson’s algorithm calculates the Delaunay
triangulation with vertices at points of of V
The structure of a TIN. The TIN is a topological data model. The data are stored in
a set of tables that retain the coordinate values as well as the spatial relations of
the facets.
 TIN Applications

Slope and aspect

Contouring
Finding drainage networks
Creating cross-sections
Visualization of a 3-dimensional surface

 Advantages/disadvantages

DEMs:
accept data direct from digital altitude matrices
must be resampled if irregular data used
may miss complex topographic features
may include redundant data in low relief areas
less complex and CPU intensive
TINs:
accept randomly sampled data without resampling
accept linear features such as contours and breaklines (ridges
and troughs)
accept point features (spot heights and peaks)
vary density of sample points according to terrain complexity



‘Much of the life of the mind consists in applying concepts to things’
(Fodor 1998:24)