Deck 4/4 Flashcards
Guest lecture
how can we manage finite water resources
integrated water resources management
systems approach to water management
- circular
- interdisciplinary
- participatory
- practical
circular
feedbacks (closed loops), nonlinearities and time delays are considered only
interdisciplinary
socio-economic and natural systems are linked in one consistent framework
participatory
modelers interact directly with stakeholders at all stages of the project
practical
provides quantitative decision-support tools for researchers, policy-makers, and potentially the public
systems
networks of positive and negative feedbacks
all dynamics arise from?
the interaction of feedback loops
modes of behaviour in a dynamic system (multiple feedbacks all interacting)
- exponential growth
- goal seeking (down to a threshold line)
- s shaped growth
- oscillation
- growth with overshoot (exponential an dthen wavy at the top )
- overshoot and collapse (up and then peaks and comes down)
exponential growth
- caused by positive feedback
- state of the system doubles in a fixed period of time
- doubling time can be determined by the Rule of 70
positive feedback are powerful because?
their rate of increase grows as the state of the system grows
nonlinearities ____ positive feedbacks and ____ negative feedbacks until exponential growth ____
weaken, strengthen, stops
when are errors large in exponential growth
when growth rate increases or horizon lengthens (time scale increases)
model integration
two possible approaches combined into one
“ugly” constructs
constructs that are perfectly valid as software products but ugly and useless as models
challenges to model integration
mismatched scales, skewed geometry, overwhelming complexity, confusion of tongues (different disciplines use different terms)
why is it important to have valid simulation models
conclusions are no longer valid, misleading, and potentially dangerous
calibration
range in model parameters sampled until differences between the observed and simulated are minimized
evaluating performance of the model
pairwise comparisons of simulated and observed values to see how well the model represents past observed data
why evaluate performance
- provide quantitative estimate of ability to reproduce and predict
- provide a means for evaluating improvements to modelling approach
- compare current modelling with previous study results
white box models
causal-descriptive
- represents the theory of the real system
- must reproduce the behaviour and explain how the behaviour is generated
black box models
correlational
- data driven
- don’t necessarily have physical significance
philosophical classifications of validation
- reductionist: valid model as an objective representation of a real system, can be corret or incorrect, validity as accuracy not usefulness
- holistic: valid model as one of many possible ways of describing a real situation, models are not correct or incorrect but lie on continuum of usefulness
structure validity
direct structure tests and structure-oriented behaviour tests
behaviour validity
emphasis on pattern prediction rather than point (event) prediction
what is the goal of modelling
simplification
all models are wrong
t/f
true, but some are useful
linking pre-existing models can easily lead to ____ models
more complex