GIS applications for measuring

and simulating human impacts on

the environment.

Isaac Ullah
www.public.asu.edu/~iullah
 Summary
1. What is a “GIS”?
1. Data types
2. Projections
2. Monitoring modern human impacts with GIS.
1. Remotely Sensed Data
2. Applications
3. Simulating ancient human impacts with GIS.
1. Landuse/Landcover modeling
2. Erosion/Deposition modeling
4. Questions?
 What is a GIS?
• “GIS” stands for “Geographic Information
Systems”
• A GIS is a software platform for storing,
organizing, viewing, querying, and transforming
spatial data.
• Two most prevalent software platforms are the
expensive commercially licensed ESRI
ArcGIS©® and the free and open-source
TM

GRASS GIS platform. (Guess which one I use?)

• Data in a GIS is stored in either Vector or in
Raster formats.
 Vector Data
• Discreet geometrical objects which are either points,
lines, or polygons
• Vertices are placed by X and Y location for all vector
types.
• For line and polygons, the vertices are joined by lines
according to geometrical functions
• Attributes are associated with each discreet vector
shape
• Attributes are stored in a database; and therefore, each
object can have multiple dimensions of data associated
with it
• Easy database editing with your favorite spreadsheet
software (most are in “.dbf” format)
• Data can be displayed thematically for easy visual
analysis
 Vector Lines

Data associated with each

individual vector line
 Vector Points

Multiple dimensions of
data associated with
each vector point
Thematic Vector Points Overlain on
Raster Density Surface
Map of Sherd Density Index
Overlayed on Wall Height Contour
1
2

$
3 12 13
21 10
Legend 27 11
36 15
30 31
43
Density Index 42
46 40 57
54
0 68
73 66
55
78 56
95 72 19
1 79 74
120 113 99 96 107 83 81
85
2 108
82
137140
141 155 109
84 71 100
3 119 130 142 90 70
102 116 104 87 53
139 149 156 160 145
150
94
4 110 134 93 105
136147 164 153 64
161 126 111 92
162 121 69
177 152
165 175 172 146 114 97 77 62
5 131 132 122
80
166 176 112 75 50 37
185 178 128
167 173 180 174 127 115 117
Wall Height in Courses 189
183 143
65
48
39
28
187 106
184 91 76 29
0-1 191 144 138 118 44
88 51 41
194 129 25
1.1 - 2 227
220
9
213 157 123
47 32
2.1 - 3 226 188 24
244
38
236 158
3.1 - 4 230
241 193 8
23
243 169
4.1 - 5 7
231 271
35
5.1 - 6 276
240
242 22 6
6.1 - 7
7.1 - 8
267 186
8.1 - 9 182
257 289 234 198
238 282 179
250
256 291 292 254 239
171 245
283 196
301 296
253 251 262 199
306 223 197
278
237 307
309 293 285 268 259 207
303
273 280 304 295
246 288 286 224
247 255 299 308
270
300 260 200
252 281
275 249 214
264 298 287
248 297
219 294 201
272 269 263 233 215
266 258
225
202
208
209

205
204 203

218

221

0 4.5 9 18 27 36
Meters
 Raster Data
• Continuous data (a matrix of values)
• Each layer has a maximum of 3.5 dimensions of data (X,
Y, Z, Label)
• Multiple layers can be stacked to represent many
dimensions of data
• Display of data can be adjusted by ranges for heuristic
analysis
• Raster surfaces can be interpolated from discreet data
(ie. vector points)
• Complex statistics and matrix math can be calculated at
each pixel or between pixels of single or multiple layers.
• Can be viewed in simulated 3-D
• This allows for complex data transformation and
simulation of phenomena that cannot be practically
measured/observed in real life
Graphic Display of Raster Matrix
 Output: Continuous Data
Input: Discreet Data Density Probability Surface

Artifact Point Densities

Various 3-D displays of a Raster DEM

Original Raster File

 Which is Better?
• It depends on your needs
• Vectors are better for associating many data types with
one spatial object (ie. site point) in one file
• Vectors can only be used to represent discreet
phenomena
• Easy to display for interesting thematic maps
• Raster's are better at representing massive amounts of
spatially differing data
• They are also better for doing mathematical operations
on that data
• They can represent both continuous or discreet data, but
only in one dimension per layer
• Can display in 3-D!
• You can use raster data in complex simulation modeling
 Soils Data as Vector

Soils Data as Raster

 Raster
Elevation Data
(Digital
Elevation
Model, or DEM)

Vector Elevation
Data (Contour
Map)
 A Quick Note on Projections
• All maps are flat representations of a round
world, and GIS data is no different
• Projections are different ways to mathematically
“unbend” curvilinear distances into flat distances
• Any projection needs to reference a datum point
from which all mapped measurements can be
tied back to the Earth
• There are many types of map projection
systems, but only two you are likely to deal with
on a regular basis: Lat/Lon and UTM
 Two Major Projection Types
• Latitude/Longitude (Lat/Lon) projection
• Works worldwide
• However, all distances in this type of projection
are measured as fractions of the Earths
diameter (degrees, minutes, and seconds or
decimal degrees)
• Universal Transverse Mercator (UTM) projection
• Broken up into a series of zones across the
world
• It’s units are meters, but you must stay within
only the correct zone, or your data will become
distorted
Lat/Lon Projection
UTM Projection
UTM Zones
Monitoring Modern Human Impacts
• Essentially all from remotely-sensed data
• Earliest data sources are aerial photographs beginning in
the early the 20th century. They are “Panchromatic”
(black and white) images, sometimes in stereo 3-D
• Oldest wide-coverage data come from 1960’ spy
satellites, especially the CORONA missions (high
resolution, stereo, panchromatic)
• Starting the late 1970’s, LandSat data are available for
the whole world (multiband, medium resolution)
• From the 1990’s onward we also have:
– Space Born Radar: TerraASTER, SRTM
– High-altitude imagery and laser topography: QuickBird, Lidar
– Full coverage Satellite-born sensors: AVHRR, MODIS, IKONOS
• Many of these data are archived and made accessible
through the web by the Global Landcover Facility
Monitoring Modern Human Impacts
• Typical applications:
– Hazard identification
– Landcover and Landscape mapping
– Tracking changes through time (time series)
• GIS operations:
– Rectification and Georeferencing
– Band manipulation of multiband imagery
– Classification
– Feature identifcation
– Mapping and Digitization
– Quantification
Human-induced Landscape Changes

High resolution imagery

Vegetation Monitoring and Comparison
Evidence for Climate Change
Urban Heat
Island
Monitoring
Northern Jordan Landsat
Band Manipulation

Red, Green, and Blue Spectra

Northern Jordan Landsat
Band Manipulation

Near Infrared, Red, and Green Spectra

Northern Jordan Landsat
Band Manipulation

Far Infrared, Near Infrared, and Red Spectra

Northern Jordan Landsat
Classification

Unsupervised (automatic) landcover classification

Northern Jordan Landsat
Classification

Supervised (with user input) landcover

classification
Northern Jordan Landsat

Feature Identification (farm fields) and Mapping

 Modeling Ancient Human Impacts
• What affect, if any did ancient human landuse
have on the environment?
– Little direct archaeological evidence of human
impacts on the environment
– Even less direct evidence of the processes that
created them
• We must simulate ancient landuse through
spatially explicit process-based models
– Human processes (farming, herding, deforestation)
– Natural processes (climate, vegetation, geological)
• Compare the results with proxy data (pollen
records, archaeological evidence, sedimentary
record)
 Mediterranean Landscape
Dynamics Project
• The Medland project aims to
understand the long-term
effects of ancient landuse
practices on the environment.
• GIS-based surface process
simulation coupled with semi-
dynamic stochastic landuse
models (eventually with
Agent-Based landuse model)
• Track the effects of landuse
on landcover and
subsequently on the spatial
extent and severity of erosion
and deposition through time
 1. Potential landscape model
Terrain
Terrain
modeling:
modeling:
multi-yr.
multi-yr. steps
steps

Climate
Climate
model
model
Potential
Potential
landscape
(natural processes only—
no human impacts!))
landscape
model
model

Vegetation
Vegetation
modeling:
modeling:
multi-yr.
multi-yr. steps
steps

Paleobotanical
Initial state
Paleobotanical
data
data Paleo-
Paleo-

2. Reference landscape time

vegetation
vegetation
Veg.
Veg. edaphic
edaphic
parameters
parameters

Modern
series (Built from available
proxy data: Geology,
Modern
DEM
DEM Paleo-
Paleo- Reference
Reference
terrains
terrains landscape
landscape

paleoecology,
(DEM’s)
(DEM’s) chronoseq.
chronoseq.
Geological
Geological
data
data

Prehistoric
paleogeography,
archaeology)
Archeological Prehistoric
Archeological settlement
data settlement
data && landuse
landuse

Initial state
& validation
at various stages

3. Agropastoral socioecology
Agent
Agent
Modeling
Modeling
Vegetation
Vegetation
modeling:
modeling:
multi-yr.
multi-yr. steps
steps
model (Semi-Stochastic
Settlement
Settlement
&
& landuse
landuse
modeling
Agropastoral
Agropastoral
socioecology
socioecology
model
model
and Agent-based human
landuse models coupled
modeling
Climate
Climate Terrain
Terrain
model
model modeling:
modeling:

with natural process

multi-yr.
multi-yr. steps
steps

models)
Topography
• Terra ASTER DEM
• Re-interpolated to
15m resolution
• Ultra-high resolution
topography from
aerial photograph
stereo pairs (near
future)
• Study areas defined
as watersheds
using hydrologic
modeling
 Paleoclimatological Modeling
Annual Precipitation 8000-2000 BC
Wadi Ziqlab Area Weather Stations

• Weather station data 2500

retrodicted for 14ky at 200 yr 2000

Baqura

intervals to produce Shuneh-North

Beit Qad "Jenin"

mm of precipitation
1500 Irbid Nursery

sequences for annual and Ramtha

Wadi Yabis
Ras Muneef

monthly precipitation, 1000 Mafraq

Deir Alla
Wadi Faria

temperature (mean, days>40°, 500

Tulkarm

days <0°), and storms. 0

• Monthly and annual climate

-10000 -9000 -8000 -7000 -6000 -5000 -4000
years BP

sequence models interpolated

Annual
to create paleoprecipitation Precipitation
at 7000 BP
surfaces using multiple
regression (topograpy,
distance from sea, latitude,
etc)
 Landcover Modeling
• Potential natural vegetation
model based on climate
and topography
• Patch models incorporating
successional dynamics
• Feeds directly into erosion
model
Impact Areas: Initial Conditions
Catchment modeling
Different landuse models
Resultant Vegetation Profiles

Grass Shrubs/Maquis Woodland

Erosion/Deposition Modeling
 Model the effect of the resulting
landcover on erosion
40 years of
fallow
agriculture with
grazing

Control Model (no landuse)

 3-D results with human landuse

Control
model (no
human
landuse)

Reality Check. There IS a deep

canyon in this location!!!
Thank You!