Skip to content

Latest commit

 

History

History
19 lines (7 loc) · 4.07 KB

How To Use CHIME.md

File metadata and controls

19 lines (7 loc) · 4.07 KB

Summary

This document describes how to use the CHIME ABM, and how to determine if it is working 'correctly.' The CHIME model was designed to function as a virtual laboratory to test hypotheses about information flow and decision making during weather-related threats, specifically, for this modeling setup, hurricanes. Users may modify different model parameters to explore the relative effect of those changes on patterns of household evacuation and other protective decisions. For example, users can adjust the importance of factors that influence citizen-agent decision making (e.g., the weighting of different types of information, inclusion of census information), study different hurricanes affecting different regions, and compare the effects of different hurricane forecasts (e.g., perfect versus imperfect forecasts). Because of its focus on functioning as a hypothesis testing tool, the CHIME ABM does not produce a single verifiable outcome. However, the model produces output of the timing and location of household evacuations, which can be compared to historical records of evacuation.

How to Run the Model

The CHIME ABM is written for the Netlogo ABM platform. To open CHIME ABM, one must download Netlogo, a freely available modeling environment available at https://ccl.northwestern.edu/netlogo/ . The current version of CHIME ABM uses version 6 of the Netlogo environment. The CHIME.nlogo file can be opened with the Netlogo program. Do not move any of the accompanying folders, such as STORMS or REGION. These folders contain data used by the model and their path is assumed to begin with the same directory as the CHIME.nlogo file. Once the file is loaded, the user is presented with a column of buttons to run the model, along with sliders and switches to adjust model parameters and output. The simulation can be viewed in a display at the center of the interface.

First, choose the storm of interest under the 'which-storm?' button. Next, click the 'Setup Simulation' button. The display will load an image of the modeled domain for that storm and the agents. A gray line, which traces the actual path of the storm, is also shown across the screen. Once the model is set up, click the button 'Run Simulation' to begin a model simulation. As the simulation runs, agents will collect, share, and interpret information about the hurricane, and citizen agents will make decisions about whether to evacuate, take another protective action, or adjust their frequency of data collection and decision making. The simulation will continue until the hurricane has passed through the model domain. Information about the agents in the simulation can be saved in several different files, which can be chosen from the switches labeled 'Model Output Controls.'

Example Simulation

The Netlogo interface allows a user to design and run experiments using an interface called BehaviorSpace. It can be accessed by selecting 'Tools' followed by 'BehaviorSpace'. From the list of experiments, choose 'experiment_example' (the first experiment in the list) and then select the 'Run' button. After clicking 'Run', do not select 'Spreadsheet output', 'Table output', 'Update view', and 'Update plots and monitors', as the simulation will finish faster if these options are not checked. The output from the simulations will be saved to a folder called 'output' located in the same directory as the CHIME model. Each example run will create a global_output.csv file with information about the parameters used and citizen evacuations. In this simple example, we have modified the weighting of forecast information in citizen-agents' risk assessments and decisions (the parameter forc-weight). With a low value of forc-weight of 0.2, citizens do not evacuate at all, while for a value of 1, 15-20% of the citizens evacuate. The number of citizens that chose to evacuate is recorded in the column entitled 'percent_all_evacuated.' A variety of information, including when citizens choose to evacuate as well as the location of citizens in the domain, can also be outputed and analyzed from each simulation to answer specific questions or hypotheses.