Multi-criteria modeling 1

Model

A representation of a phenomenon or a system

Model simplification

Models can be simplified by focusing on key variables

Digital twin

A very realistic model of a system

No model is perfect

Models are not exact representations of reality

Why models are useful

They help us understand complex systems, solve problems with multiple data inputs, provide dynamic simulations, and support decisions or design processes

Garbage in = garbage out

Poor input data or assumptions will produce poor model outputs

“All models are wrong, but some are useful”

A quote by George Box about the usefulness of imperfect models

Descriptive model

A model used to describe current or historical conditions

Main question of a descriptive model

What is happening

Example of a descriptive model

Global average sea surface temperature

Prescriptive model

A model used to predict future condition

Main question of a prescriptive model

What should we do

Example of a prescriptive model

Projected global mean sea level rise under different SSP scenarios

Deterministic model

A model where output is determined by the input variables

Main property of a deterministic model

The same input variables produce the same output

Example of a deterministic model

IDW

Another deterministic model example

Least-cost path

Stochastic model

A model that includes randomness and provides uncertainty

Main property of a stochastic model

It incorporates uncertainty in model prediction

Shaded colors in stochastic model output

Represent the uncertainty of model prediction

Dynamic model

A model that treats time as a variable

Example of a dynamic model

Global projections of future urban land expansion under SSPs

Static model

A model that treats time as a constant

Main property of a static model

Represents conditions at one point in time

Species Distribution Model (SDM)

A model for predicting suitable habitats for a species based on environmental variables and species occurrence records

Main purpose of an SDM

To predict suitable habitat for a species

Pseudo-absence points

Artificially generated non-presence points used in species distribution modeling

Biophysical data

Environmental variables used in the species distribution model

Forecasted biophysical data

Future environmental variables used to project future habitat suitability

Predicted habitat

The output of the species distribution model showing suitable habitat

Presence-only data in SDM

Species distribution modeling may require only presence data

Maxent

A method for estimating the probability of species presence in a given area

maxent() function

The R function mentioned for running Maxent

Species extinction risk analysis

Analysis of species health and extinction risk over time

Annual land cover mapping

An example of a dynamic model plus descriptive model

Land cover projection

An example of a dynamic model plus prescriptive model

SDM as a stochastic model

An example of a stochastic model used in the lecture