Elsevier

Fire Safety Journal

Volume 71, January 2015, Pages 100-109
Fire Safety Journal

Optimizing watchtower locations for forest fire monitoring using location models

https://doi.org/10.1016/j.firesaf.2014.11.016Get rights and content

Highlights

  • Optimization models were developed to find optimal watchtower locations.

  • Explored tradeoff between coverage and cost using abiobjective optimization model.

  • Tested using the data from a forest part in Guangzhou, China.

Abstract

Automated forest fire monitoring systems can be constructed using forest fire watchtowers equipped with laser night vision cameras or high-definition video cameras. In order to minimize the construction cost and to maximize the monitoring coverage of forest fires, efficiently placing the watchtowers is critical. This paper examines efficient watchtower locations by integrating visibility analysis and location-allocation models. Specifically, based on the classical location set covering problem and maximum covering location problem, three optimization models are developed to satisfy three kinds of requirements of forest fire monitoring in practice: minimizing cost with full coverage, maximizing coverage with a fixed budget, and maximizing coverage while minimizing the cost. The models are tested using integer programming and a multi-objective genetic algorithm, with an application in a forest park in Guangzhou, China. The results suggest that this model-based optimization approach to watchtower location can be used to improve the efficiency of forest fire alarm systems.

Introduction

Forest fire is a severe natural disaster and public emergency of the world. Fire incidents are often abrupt, spread rapidly, difficult to control, and highly disastrous, and have become a serious threat to forest resources globally as they affect forest ecosystem succession and global climate change. According to a report of Greenpeace Research Laboratories and climate change research of United States Environmental Protection Agency, there has been an increase in the number of forest fires because of global warming and intense human activities [1], [2]. In recent years, forest fires, known as wildfires, consumed more than 6.25 million acres of forest in Alaska (roughly equal to the area of Massachusetts) [2].Climate change is projected to increase the extent, intensity, and frequency of wildfires in certain areas of our earth. Forest fires have also become a major concern in China in recent decades because of the increasingly serious damage they have done to the environment and the loss of societal wealth incurred. According to a report on Chinese forestry development, a total of 3966 forest fire incidences were identified in 2012, and China spent ¥342,000,000 fighting forest fire, and the government invested more than 2 billion Yuan on 190 construction projects to prevent forest fire [3].

Among many preventive measures, early detection and suppression of forest fires are the main ways to minimizing damage. The critical issue in forest fire monitoring systems is the immediate response in order to minimize the scale of destructions. Many countries that have recognized the significant importance of forest fire monitoring have developed effective technologies, including monitoring via observation towers, cruising aircrafts, remote sensing using meteorological satellites, and sensor networks, to improve their response ability [4], [5], [6], [7], [8]. Forest fire monitoring technology was implemented belatedly in China but has experienced a rapid growth, with watchtowers being the first selection for forest fire monitoring in the country. In the light of the 2013 report of Chinese forestry development, for example, various preventive and monitoring measures have been implemented, and the coverage rate of forest fire monitoring has increased from 45.3 to 63.1 percent in China [3]. Since December 2013, the Administration of Forestry and Gardening of Guangzhou Municipality launched a project of 74 million Yuan to monitor fire of key forest zones and green parks by video cameras, which will equip 941 cameras to cover 8 parks, 3 forest farms, and 2 forest and wild animal protection areas [9].

Protecting wild animals, forests, and the environment from forest fires has long been a major concern in environment and natural resources management [10]. In the past, forest personnel monitored fires from watchtowers that were located on hilltops in forests so that forest fires could be discovered and alarmed as soon as possible [6], [7]. However, living conditions are often difficult at lookout towers for human observations who may also lack the consistency and reliability required for constant monitoring. As a result, vision techniques such as automatic video surveillance systems (AVSS) were proposed to monitor small forests [8]. Many medium and large-scale fire surveillance systems currently do not accomplish timely detection because of low resolution or long periods between scans [6], [7]. With developments in technology, high definition video cameras and sensor networks are now being used to equip watchtowers and automate forest fire monitoring. Further, watchtowers can be equipped with solar panels that supply power to support the compression of images and real-time transmissions to command centers via wireless networks. Forest fire watchtowers equipped with laser night vision cameras or high-definition video cameras can constitute an automated forest fire monitoring system that has a wide coverage of monitoring viewsheds and can quickly respond to forest fire alarms (Fig. 1).

Determination of the optimum location of permanent fire watchtowers in a given forest area has been, and continues to be, of significant interest to both the practitioners and research communities [11], [12]. The efficient location of watchtowers equipped with cameras has become increasingly important as it can directly influence the construction cost of watchtowers and the monitoring coverage of forest fires [13]. It is a combinatorial optimization problem and, consequently, is difficult to obtain optimal location solutions using simple enumeration and search methods or viewshed analysis based on geographic information systems (GIS). Further, watchtowers need to be sited optimally to meet forest fire monitoring requirements such as full coverage, maximal coverage, minimal cost, and/or minimum overlap for forest fire monitoring. There are also specific constraints such as terrain limitations and the effective detection range of cameras installed on watchtowers of various heights.

This paper proposes a modeling approach to optimizing the spatial coverage of watchtowers equipped with cameras in forest zones. This approach integrates coverage models and visibility analysis into a spatial optimization framework and applies the result to forest fire monitoring. The aim of the approach is to develop a procedure for finding optimal solutions of locating watchtower that satisfy a set of objectives and specific constraints of forest fire monitoring in practice. In the remainder of this paper, Section 2 analyzes technical problems associated with forest fire electronic monitoring, Section 3 presents the three proposed optimization models, Section 4 outlines the implementation procedure for the three models and discusses the application results obtained, and Section 5 summarizes and concludes this paper.

Section snippets

Problem analysis

When a watchtower is planned for construction on an undulating terrain, a major concern is the tower's viewshed, meaning a set of locations on the terrain that are visible from the watchtower extending out to the maximum visibility distance of the camera. Viewshed analysis is an important function of GIS as a method of visibility analysis based on the terrain and has been successfully applied in many applications [14], [15], [16], [17]. The viewshed of a watchtower is computed and analyzed at

Watchtower location models

Based on the LSCP and MCLP models, we developed three application models specifically for locating watchtowers in a context of forest fire monitoring. We incorporate multiple types of watchtowers that can be equipped with cameras with different specifications. The indices and constants used in our models are listed below:

  • t=index of a watchtower type (0tT1),

  • i=index of a potential location for building a tower (0iM1),

  • j=index of a demand cell in the grid that needs to be monitored (0jN1),

  • C

Computational experiments

The study area is located at Longdong Forest Park that is part of the southern end of the Dayu Mountains, in the northeast of the city of Guangzhou, China. Forest covers 96 percent of the park, causing the area to be at high risk of forest fires during the dry seasons. The fire risk period in the Guangzhou region is from mid-September to the end of April next year, with November to March as a critical period. The size of the Forest Park is nearly 10 square kilometers, including more than 50

Conclusions and discussion

This paper suggested that integrating location models and visibility analysis can help efficiently place watchtowers for forest fire monitoring on the terrain. We discussed the procedure of data preparation using viewshed analysis in a GIS and optimization models based the location set covering problem and the maximum covering location problem. We demonstrated how multiple types of watchtower can be incorporated in the optimization models and how a bi-objective optimization model can be used to

Acknowledgments

This work was supported by Guangdong Sci-Tech Project award No. 8451064201000943 and Guangzhou Sci-Tech project award No. 2010Z1-E031. The authors thank the anonymous reviewers for their insightful comments that greatly helped improve the manuscript.

References (37)

  • N. Xiao et al.

    Interactive evolutionary approaches to multi-objective spatial decision making: a synthetic review

    Comput. Environ. Urban Syst.

    (2007)
  • M.P. Scaparra et al.

    A bilevel mixed-integer program for critical infrastructure protection planning,

    Comput. Oper. Res.

    (2008)

    .

    (2008)
  • J. Cotter

    Forest Fires: Influences of Climate Change and Human Activity,

    (2009)
  • S.H. Julius et al.

    Preliminary Review of Adaptation Options for Climate-Sensitive Ecosystems and Resources

    (2008)
  • Chinese State Forestry Administration, Annual Report of Chinese Forestry development in 2013,...
  • J.L. Casanova, A. Calle, A. Romo, et al., Forest fire detection and monitoring by means of an integrated MODIS-MSG...
  • A. Xu et al.

    Distribution of forest fire prevention resources based on GIS, remote sensing of the environment: 16th national symposium on remote sensing of China, 2008, China

    Proc. SPIE – Int. Soc. Opt. Eng.

    (2008)
  • F.C. Rego et al.

    Modeling the effects of distance on the probability of fire detection from lookouts

    Int. J. Wildland Fire

    (2006)
  • Cited by (0)

    View full text