Spatial Model & Simulations
The following lecture is in reference of “Spatial Models: What? Why? How?” by Dan O Sullivan.
What is a Model?
Models are everywhere. Models are representations of the world that allows us to describe, understand, and explore the real world. We create and employ models in our daily lives constantly, often unknowingly. For example, language is a model of the real world because we learn about the meaning behind specific words and is able to recreate a mental image and understanding of words when we hear them.
Much like an architecture model, the model serves as a to-scale representation of “reality”, or a “version of reality”. Today, design models also exist digitally through algorithmic representation of geometry. “Computer-Aided Design” (CAD) models allows us to quickly explore options and new ideas.
A scientific model, or data models, follow similar structures. Base on O. Sullivan, scientific models can be defined as:
“a simplified representation of a system under study, which can be used to explore, to understand better or to predict the behavior of the system it represents.”
Types of Models
Conceptual Model
The first step to create a model is to start with a Conceptual Model, which helps us explain complex real world systems into simpler elements by studying it through a theoretical perspective. Real-world phenomenon are usually expressed in complex conceptual models, but in essence, a simple conceptual model can be expressed in words such as:
If X is true, then Y.
Mathematical Model
Mathematical Model is the most common type of model and covers all types of data models. In a mathematical mode, components are represented by values. Equations are simple mathematical model that describes the relationship (operations) between each components. For example:
Pedestrian Flow = Speed x (Pedestrian / Sidewalk Space)
Empirical Model
Empirical Model, is a model based on empirical observations to study complex relationships between components within a phenomenon. These models are generally statistical, where variables are observed and analyzed for correlation or causation. In a linear-regression model , a line is drawn to define the relationship between independent and dependent variables. This line therefore predicts the outcome of future events when one variable changes.
Finally, like any statistical model, it is important to note that “correlation is not causation”. There are many examples of how the misinterpretation of empirical models can lead to detrimental policy decisions.
See “Field Guide to Life in Urban Plazas” by SWA Group
Simulation Model
- System Dynamics Models (SDMs) are computer simulations of complex mathematical models consisting of multiple equations in relationship to one another. Usually consisting of three types of variables: stocks, flows, and parameters, system dynamics model typically describes how resources (stocks) is moved or distributed (flows) across situations (parameters) These relationships are often derived from empirical models or conceptual models in addition to quantitative rules.
- Spatially Explicit Models (SEMs) represent simulation models where the spatial relationships (location, spatial and environmental attributes) are crucial to understand the system studied. Recall that most models are simplifications of reality, and therefore many scientific models are spatially implicit (where spatial information is not included). As designers and planners, we understand how powerful spatial relationships are, and work spatially explicitly both manually and computationally.
- Agent Based Models (ABMs) is a simulation system is modeled as a collection of autonomous decision-making entities called agents. Each agent individually assesses its situation and makes decisions on the basis of a set of rules. At the simplest level, an agent-based model consists of a system of agents and the relationships between them. Even a simple agent-based model can exhibit complex behavior patterns and provide valuable information about the dynamics of the real-world system that it emulates. In addition, agents may be capable of evolving, allowing unanticipated emergent behaviors.
Reality ←→ Generality
It is important to remember that models are simplified representations of the real world. As models become more detailed and more specific, it loses its precision and generality to inform larger patterns. For example, a model that is specifically tailored the demographic character of one community is likely to produce inaccurate results elsewhere; a highly generalized model of people’s commute pattern within in a city may also fail to recognize the spatial conditions across different neighborhoods.
However, it doesn’t mean that you couldn’t learn general patterns from a specific model and vice versa. You would agree that there are similarities in the patterns between any neighborhood and community. A robust model should 1) evolve and grow in specificity as more detailed data is introduced, and 2) scale and adapt when placed in a different environment.
Why Use Simulation Models?
“the simplest way to think of the relationship between models and simulations is that simulations are implementations, usually in a computational setting.”
The author makes a case for simulations because its often unethical to experiment on real world populations, such as where a failed policy experiment could impact entire generations. Architecture is often unsimulated and un-tested for many of the behavioral aspects we as designers claim it will change. Is it unethical to deploy architecture that has not been tested for its use?
