The Aboriginal Sites Decision Support Tool (ASDST)

Visualising spatial archaeological data
The Aboriginal Sites Decision Support Tool (ASDST)
Mal Ridges
Aboriginal sites decision support tool ASDST:
http://www.environment.nsw.gov.au/licences/AboriginalSitesDecisionSupportTool.htm
AHIMS
GRASP
Pre1750
model
All modelled
Feature types
Feature
Data
Muti-resolution
Segmentation
ASLUs
Spatial
variables
SGA
algorithm
Model
reliability
All modelled
Feature types
Survey
Priority
calculator
All modelled
Feature types
Reliability
model
Land-Use
Layer
Survey priority
model
Vulnerability
parameters
Current
Condition
calculator
KEY:
Corp.
database
GIS layer
Data
Tool or
algorithm
Current
model
Accum.
impact
calculator
All modelled
Feature types
Accum. Impact
model
Model testing using withheld data:
• Models were built with 2007 site data
• Since then ~10,000 sites have been added to the database
• Independent data, but also suffers the same inherent recording biases
Scarred trees:
Stone artefacts:
Comparison with Pardoe / Martin model:
Predictions for AHIMS sites
Predictive accuracy
comparison:
NOTE: Measured against current
AHIMS data (including RERP data
not used to build either model)
Low likelihood
High likelihood
CONCLUSION: Both models perform well, predicting high likelihood for the
majority of sites. The ASDST model performs slightly better
(fewer sites with low likelihood)
Survey priority
Survey priority
M
SP   M f ( pre1750)  I f  R f
Mf
Survey priority combines:
• large data gap
• low impact
• high current likelihood
Mined areas that have
been the focus of much
assessment are LOW
survey priority
Data gaps and low
impact areas are HIGH
survey priority
Scarred
Scarredtree
treemodel
modelpre1750
current
Impacts in the
Hunter Valley
Site feature
Artefacts
Rock art
Burials
Grinding grooves
Hearths
Shell middens
Stone quarries
Scarred trees
SA
If 
% Impacted
26.7
18.6
63.5
24.0
70.5
55.0
20.3
69.9
SA
 M f ( pre1750)   M f (current )
x
x
SA
 M f ( pre1750)
 100
x
Site feature
% Protected
Artefacts
3.3
Rock art
5.9
Burials
3.0
Grinding grooves
4.7
Hearths
2.3
Shell middens
4.0
Stone quarries
4.4
Scarred trees
3.6
C
Pf 
M
x
SA
f ( pre1750 )
 M f ( pre1750)
x
 100
http://www.gbif.org/
These tools can be
automated
Providing users with
custom results for
defined regions
Result: better contextual
data for decision making
Accumulated
impacts
Accumulated impacts
 I f ( c ) RSR  rx ( c ) 

I   f ( x) 


RSR 
Mf
I c  2rx ( c )
M
I SR
• Using the difference between
the pre1750 and current
version of each model, the
level of impact can be
mapped
• This maps some impacts
across all feature types
modelled
• Highest impacts occur in
mined areas, irrigated
cropping and urban extent
Segmentation
for CMA
• Multi-resolution
segmentation on
current models
• 53,386 segments
mapped for the
CMA
• Average area is
190 Ha
• Information can be
tagged to each
segment
• The information can
be used for
property scale
planning
Take home
message
• All this is possible just from
using site type + location
• BUT…its focused on
generating landscape scale
data that supports decision
making
• It also provides simple
information that can be used by
non-archaeologists in order to
trigger appropriate assessment
• Derived information from the
ASDST is easy to deliver
through on-line tools: eg
regional profiling