Last updated November 08, 2009 01:20, by Bruce Schubert
Feedicon  

Campbell Prediction System

Vision

Version 1.2

Bruce Schubert

Revision History

Name Date Reason for Changes Version
Bruce Schubert May 4, 2008 Created. 1.0
Bruce Schubert October 24, 2009 Updated constraints. 1.1
Bruce Schubert November 7, 2009 Added context diagram. 1.2

Project Description

The Campbell Prediction System (CPS) [1] method is a practical way to use on-scene observations in order to determine fire behavior strategies and tactics. The observed fire behavior becomes the baseline for fire behavior predictions. A special logic replaces intuition allowing an explanation of how tactics are developed. Developing a strong case for acting on the fire’s potential rather than waiting for the fire to make the change would save many of the lives that were lost as a result of firefighters who had reacted too late. If people could explain what the potential of the fire is in their situation, few accidents would happen. The Campbell Prediction System provides the logic and language to do so.

This project will codify fire behavior observations as GIS data elements and analyze them using data mining and machine learning techniques to determine the potential fire behaviors. The goal is to graphically render the potential fire behaviors relative to time and space on a 3D terrain viewer and to generate Incident Action Plan (IAP) maps that highlight that both dangerous and opportunistic potential fire behaviors situations.

References

  1. The Campbell Prediction System, A Wild Land Fire Prediction and Communication System, 1991, Doug Campbell

Problem Statement

The current state-of-the-art fire modeling software does little to assist in tactical suppression efforts. Computer-aided wildland fire predictions can be improved. Fireground operations can benefit from state-of-the-art GIS enabled application. The most prominent fire modeling applications in use include the following (overviews from FireModels.org, see http://www.firemodels.org for complete descriptions)

Overview of Existing Systems

BehavePlus – Fire Modeling System

The BehavePlus fire modeling system is a PC-based program that is a collection of models that describe fire behavior, fire effects, and the fire environment. It is a flexible system that produces tables, graphs, and simple diagrams and can be used for a multitude of fire management applications. BehavePlus is the successor to the BEHAVE fire behavior prediction and fuel modeling system (Andrews 1986, Andrews and Chase 1989, Burgan and Rothermel 1984, Andrews and Bradshaw 1990). It is called the BehavePlus fire modeling system to reflect its expanded scope. Development continues with the addition of fire modeling capabilities and features to facilitate application.

  • Can be considered a 'point' system.
  • Each calculation is for a set of uniform conditions.
  • Rarely is a single calculation done.
  • The user looks at the effect of a range of values on the results.
  • Input is entered by the user. GIS data are not used.
  • Results are in the form of tables, graphs, and simple diagrams.

FARSITE – Fire Area Simulator

FARSITE is a fire behavior and growth simulator for use on Windows computers. It is used by Fire Behavior Analysts from the USDA FS, USDI NPS, USDI BLM, and USDI BIA, and is taught in the S493 course. FARSITE is designed for use by trained, professional wildland fire planners and managers familiar with fuels, weather, topography, wildfire situations, and the associated concepts and terminology.

  • Automatically computes wildfire growth and behavior for long time periods under heterogeneous conditions of terrain, fuels, and weather.
  • Uses existing fire behavior models for surface and crown fires, post-frontal combustion, and fuel moisture.
  • It is a deterministic model, meaning that you can relate simulation results directly to your inputs.
  • Produces outputs that are compatible with PC and Workstation graphics and GIS software for later analysis and display.
  • Can simulate air and ground suppression actions.
  • Can be used for fire gaming, asking multiple "what-if" questions and comparing the results.
  • Accepts both GRASS and ARC/INFO GIS raster data themes.

FlamMap – Fire Mapping and Analysis System

FlamMap is a fire behavior mapping and analysis program that computes potential fire behavior characteristics (spread rate, flame length, fireline intensity, etc.) over an entire FARSITE landscape for constant weather and fuel moisture conditions. FlamMap software creates raster maps of potential fire behavior characteristics (spread rate, flame length, crown fire activity, etc.) and environmental conditions (dead fuel moistures, mid-flame wind speeds, & solar irradiance) over an entire FARSITE landscape. These raster maps can be viewed in FlamMap or exported for use in a GIS, image, or word processor. FlamMap is not a replacement for FARSITE or a complete fire growth simulation model. There is no temporal component in FlamMap. It uses spatial information on topography and fuels to calculate fire behavior characteristics at one instant.

  • It uses the same spatial and tabular data as FARSITE;
  • It incorporates the following fire behavior models;
  • FlamMap runs under Microsoft Windows operating systems (Windows 95, 98, me, NT, 2000, and XP) and features a graphical user interface.
  • Users may need the support of a geographic information system (GIS) analyst to use FlamMap because it requires spatial coincident landscape raster information to run.

FSPro – Fire Spread Probability

  • Probability of fire spread from a known perimeter or point.
  • Not a fire perimeter like FARSITE.
  • Not a projection of fire size.
  • Results are based on thousands of FARSITE simulations for simulated weather sequences.
  • FSPro modeling requires computing power beyond that available on a personal computer.

Stakeholders

A stakeholder is a person, group or entity with an interest in or concern about the realization of the system. Stakeholder roles include:

Communicator
Explains the system to other stakeholders via documentation and training materials.
Developer
Constructs and deploys the system from specifications.
Maintainer
Manage the evolution of the system once it is operational.
User
Define the system's functionality and ultimately make use of it.
StakeholderRoleMajor BenefitsAttitudesWin ConditionsConstraints
Bruce Schubert -- System Architect Communicator, Developer, Maintainer Continuing development of software architecture and programming skills; explore new technologies. Strong affinity for, and history, with the fire service; desire to succeed. High number of downloads; requests for training and materials. Time
Doug Campbell -- CPS Author and Domain Expert Communicator, User Increased visibility and acceptance of CPS training and methods. Champion Reduce burnover accidents that lead to injury or loss-of-life.
CPS Instructor User
CPS Student User
Incident Command Personnel (e.g., Fire Behavior Analyst) User


Vision Statement

For wildland firefighters and incident command personnel engaged in the suppression of wildland fires who need tactical decision support tools and training systems to ensure the safety of firefighters and the effective use of firefighting resources the Campbell Prediction System software is a decision support system and visualization tool that predicts the potential fire behavior on the fireground based on the CPS method; unlike BehavePlus and FARSITE et al this product uses on-scene fire behavior observations in addition to fire behavior calculations to identify both when and where trigger points and opportunities for control exist on the fireground.

Project Scope

The following context diagram depicts the boundary and connections of the system being developed and everything else in its universe.

Figure 1. CPS Context Diagram.

The goal is to develop a computer system, “… that can use a wildland fire’s ground truths to project the thresholds of control as well as the trigger points of change and the alignment runs.”

The system will learn to classify fire behavior based on actual fire behavior observations (ground truths) collected from the fireground. The system will blend the technologies of GIS and data mining (machine learning) to predict potential fire behavior in the unburned landscape. This information will be used in the planning of safety and suppression efforts on the active fire.

This system will allow the user to visualize the fireground and recognize a fire’s potential via graphical output including:

  • Fire progression maps – the fire’s history
  • Actual fire behavior observations – these are the ground truths
  • Potential fire behavior in unburned areas – this is what the system is to discover
  • Projected trigger points – the points at which fire behavior has good potential to change
  • Tracks of the alignment of forces – these lines symbolize where the fire has high potential to make head-fire runs

To accomplish the above, the system will need several inputs. First and foremost are the fire behavior observations. These observations are both spatial and temporal. They describe when and where the observed fire behavior occurred. Each observation includes flame length, flame type (head, flanking, or heel), weather conditions and plume effects. Conditions defined by the landscape will be extracted by using GIS technologies.

For inputs, the system will consume standard GIS raster data layers such as elevation, slope, aspect, and fuels. Raster layers may also include fuel temperatures in the form of georeferenced aerial IR imagery. Additional GIS layers may include raster datasets describing fuel treatments and suppression efforts. The fire perimeter and the fire behavior observations will be represented in vector data layers.

Global data inputs will include wind, weather, and fire danger information. This data may be temporal in nature, varying from day to day or from time of day.

The intersection of a fire behavior observation with the other spatial and temporal input layers produces a list of attributes which become an example, or instance, of fire behavior to the machine learning scheme. All of the instances collected over the life of a fire become the training and test data used in the learning scheme. From these instances, the system will learn a way to classify potential fire behavior in the unburned landscape.

The system will retain the fire behavior observations and fire history in a persistent database. Over the life of the system(s), the history of several fires can be accumulated and mined for information. During initial attack, or when fire behavior observations are difficult to obtain, a model fire can be selected from the database and used instead.

Major Features

  • FE-1: GIS features include importing raster and vector data such as fire progression shapefiles, FARSITE landscape files, LANDFIRE datasets and georeferenced imagery (e.g., maps, aerial photography, and IR imagery).
  • FE-2: Editing features include the ability place and annotate fire behavior observations on the landscape via drawing tools.
  • FE-3: Analysis features include the ability to discover the potential fire behavior across the landscape by examining the past and current fire behavior characteristics and the conditions in which they occurred. The system can determine where and when the trigger points occur and where the forces are in alignment.
  • FE-4: Rendering features include the ability to display the potential fire behaviors across the landscape; this display can be varied by time-of-day or it can set to display the maximum potential fire behaviors. Trigger points and aligned forces can be visualized.
  • FE-5: Output capabilities include producing fire progression maps and IAP maps which include standard ICS symbology and CPS symbology including Trigger Points, Alignment Runs and Time Tags.
  • FE-6: Retention capabilities include uploaded and storing the fire history to a data warehouse for use in data mining and machine learning.

Constraints

DimensionConstraint (state limits)Driver (state objective)Degree of Freedom (state allowable range)
FeaturesComplete freedom to explore technologies and methods to best implement the feature set.
QualityProduct is designed to showcase capabilities and to promote CPS method to end-users. The objective to prevent injury and/or loss-of-life on the fireground requires quality and correctness and ease-of-use.
CostOpen source project developed out of personal budget; budget limited to domain support and development tools; funds for supporting development of internet database support is limited.
ScheduleMilestones are arbitrary set to coincide with opportunities to demonstrate functionality to interested parties.
Staff1 architect/developer initially; open source community support anticipated after project develops momentum

Copyright © 2008, 2009 Bruce Schubert, http://www.emxsys.com.

  • Mysql
  • Glassfish
  • Jruby
  • Rails
  • Nblogo
Terms of Use; Privacy Policy;
© 2014, Oracle Corporation and/or its affiliates
(revision 20160708.bf2ac18)
 
 
Close
loading
Please Confirm
Close