Asian American Pacific Islander Month DDSD 2024.pptx
Presenting objective and subjective uncertainty information for spatial system-based models
1. Presenting objective and subjective uncertainty
information for spatial system-based models.
Kim Lowell
2. Presenting Objective and Subjective
Uncertainty Information for Spatial
System-based Models
Kim Lowell1,2, Brendan Christy1, Greg Day1
1Department of Primary Industries
2CRC for Spatial Information, University of Melbourne
3. The Rise of Models
Land management increasingly holistic
Multiple outcome questions
Systems-science
More reliance on models for Public Policy
Increased model use demands increased model
meta-data
Uncertainty especially
4. Project Context
Victorian Government Water White Paper
Action 2.20 – Water and forest plantations
Modelling to identify best” locations
9. Implications for Uncertainty
Objective estimation not possible for all
parameters
Truly independent validation not possible
Model complexity limits numerical evaluation
Size of error
Form of error distribution
10. Policy Makers . . . .
“I understand all that, but all I want to know is if the
model estimates are „good‟.”
11. Solution
Spatial and statistical uncertainty information
Statistical on stream gauges
Coefficient of efficiency:
– CE = 1 - Σ(Oi - Pi)2/Σ(Oi – OBar)2 (1)
– where Oi and Pi are Observed and Predicted
– OBar is the average observed value over entire
period
12. Interpretation of CE
>0.6 “Satisfactory”; > 0.8 “Good”
B a s e flo w (C o r a n g a m ite ) Q u ic k flo w (C o r a n g a m ite )
C o e ffic ie n t o f E ffic ie n c y
C o e ffic ie n t o f E ffic ie n c y
1
1
0. 8
0.8
0. 6
0.6
0. 4 0.4
0. 2 0.2
0 0
0 5 10 15 20 25 30 35 0 5 10 15 20 25 30
S tre a m G a u g e ID S tr e a m G a u g e ID
In -s tr e a m S a lt (C o r a n g a m ite ) S tr e a m flo w (C o r a n g a m ite )
C o e ffic ie n t o f E ffic ie n c y
C o e ffic ie n t o f E ffic ie n c y
1 1
0.8 0. 8
0.6 0. 6
0.4 0. 4
0.2 0. 2
0 0
0 5 10 15 20 25 30 0 5 10 15 20 25 30
S tr e a m G a u g e ID S tr e a m G a u g e ID
13. Spatial Uncertainty
Reflects the calibration data and method
For example….
Stream gauges + Limits to numerical
flow directions evaluation
15. For Users of Model Outputs
S tr e a m flo w (C o r a n g a m ite )
C o e ffic ie n t o f E ffic ie n c y
1
0. 8
0. 6
0. 4
0. 2
0
0 5 10 15 20 25 30
S tr e a m G a u g e ID
16. Conclusions
Model uncertainty can be communicated
without hard statistics.
Combining numerical/objective and
qualitative/subjective information is useful.
Uncertainty representation must reflect model
fundamentals.