Evaluation of global fire detection algorithms using simulated AVHRR infrared data

This paper provides a comparison of selected algorithms that have been proposed for global active fire monitoring using data from the NOAA Advanced Very High Resolution Radiometer (AVHRR). A simple theoretical model was used to generate scenes of AVHRR infrared channel 3 and channel 4 data containing fires of various sizes and temperatures in a wide range of terrestrial biomes and climates. Three active fire detection algorithms were applied to the simulated AVHRR images and their performance was characterized in terms of probability of fire detection and false alarm as functions of fire temperature and area, solar and viewing geometry, visibility, season and biome. Additional comparisons were made using AVHRR imagery. Results indicate that while each algorithm has a comparable probability of detecting large (1000m2) fires in most biomes, substantial differences exist in their ability to detect small fires, their tolerance of smoke and neighbouring fires, the number of false alarms, and their overall suitability for global application. An improved automatic algorithm is finally presented. It offers enhanced active fire detection with comparable or reduced false alarm rates in most biomes.     

On the suitability of the use of normalized difference vegetation index for forest fire risk assessment

The NOAA‐AVHRR normalized difference vegetation index (NDVI) has often been used for forestry applications, including forest fire risk estimation. The decrement of NDVI values over a particular area has been considered an indicator of vegetation stress and linked to high fire risk. In the Mediterranean region, a large number of fires occurred in areas where the NDVI values were low. Consequently, the link between low NDVI values and fire occurrences was established. However, studies supporting this hypothesis were based only on analysis in areas that suffered fires and information over similar areas where fires did not occur was not considered. This study investigates the ability of NDVI to discriminate levels of fire risk in Spain using a very large dataset of satellite sensor images and fire events. The relative frequency distribution of NDVI values in both areas that suffered fires and those where fires did not occur, was compared in a 10 year period. The results highlight that NDVI values in areas where fires took place were similar to NDVI values in areas in which fire did not occur, showing the limitations of using the NDVI as an index of fire risk.     

A BIM-based visualization and warning system for fire rescue

Structural fires are common disasters. In Taiwan, about 100 firefighters die during fire rescues each year, primarily because they are unaware of the causes of the fire and unfamiliar with the location’s environment. Meanwhile, evacuees often die in the panic of evacuation. To solve these problems, this research proposes a Building Information Modeling (BIM)-based visualization and warning system for fire rescue. A fire dynamics simulator (FDS) simulates various conditions of structural fires in conjunction with the visualization and integration properties of BIM, and the simulation results for temperature, carbon monoxide, and visibility can be integrated and presented in the BIM model for briefing purposes before rescue operations begin. In addition, this research integrates Internet of Things (IoT) technology, which allows real-time situation monitoring. In the event of a fire, the BIM model will immediately display the situation of the fire scene and control LED escape route pointers according to the actual situation. The primary objective of this system is to provide useful information to firefighters such that they can be aware of the fire’s environment and create an effective rescue plan. Moreover, the automated LED escape route pointer may assist the building’s occupants to escape, provide the firefighters with valuable information, and allow them quickly to discover hazards so that the number of casualties can be minimized.     

Integrating fire spread patterns in fire modelling at landscape scale

Fire spread modelling in landscape fire succession models needs to improve to handle uncertainty under global change processes and the resulting impact on forest systems. Linking fire spread patterns to synoptic-scale weather situations are a promising approach to simulating fire spread without fine-grained weather data. Here we present MedSpread—a model that evaluates the weights of five landscape factors in fire spread performance. We readjusted the factor weights for convective, topography-driven and wind-driven fires (n = 123) and re-assessed each fire spread group's performance against seven other control simulations. Results show that for each of the three fire spread patterns, some landscape factors exert a higher influence on fire spread simulation than others. We also found strong evidence that separating fires by fire spread pattern improves model performances. This study shows a promising link between relevant fire weather information, fire spread and fire regime simulation under global change processes.     

Detecting unknown coal fires: synergy of automated coal fire risk area delineation and improved thermal anomaly extraction

This paper presents two complementing algorithms for remote sensing based coal fire research and the results derived thereof. Both are applicable on Landsat, ASTER and MODIS data. The first algorithm automatically delineates coal fire risk areas from multispectral satellite data. The second automatically extracts local coal fire related thermal anomalies from thermal data. The presented methods aim at the automated, unbiased retrieval of coal fire related information. The delineation of coal fire risk areas is based on land cover extraction through a knowledge based spectral test sequence. This sequence has been proven to extract coal fire risk areas not only in time series of the investigated study areas in China, but also in transfer regions of India and Australia. The algorithm for the extraction of thermal anomalies is based on a moving window approach analysing sub‐window histograms. It allows the extraction of thermally anomalous pixels with regard to their surrounding background and therefore supports the extraction of very subtle, local thermal anomalies of different temperature. It thus shows clear advantages to anomaly extraction via simple thresholding techniques. Since the thermal algorithm also does extract thermal anomalies, which are not related to coal fires, the derived risk areas can help to eliminate false alarms. Overall, 50% of anomalies derived from night‐time data can be rejected, while even 80% of all anomalies extracted from daytime data are likely to be false alarms. However, detection rates are very good. Over 80% of existing coal fires in our first study area were extracted correctly and all fires (100%) in study area two were extracted from Landsat data. In MODIS data extraction depends on coal fire types and reaches 80% of all fires in our study area with hot coal fires of large spatial extent, while in another region with smaller and ‘colder’ coal fires only the hottest ones (below 20%) can be extracted correctly. The success of the synergetic application of the two methods has been proven through our detection of so far unknown coal fires in Landsat 7 ETM+ remote sensing data. This is the first time in coal fire research that unknown coal fires were detected in satellite remote sensing data exclusively and were validated later subsequently during in situ field checks.     

A forest fire risk assessment using NOAA AVHRR images in the Valencia area,eastern Spain

The risk of widespread forest fire has been assessed from information supplied by the AVHRR sensor onboard NOAA satellites, for the area of the Autonomous Community of Valencia in eastern Spain, where several major forest fires occurred in the summer of 1994. The burnt surface data were obtained through unsupervised classification of the spectral information of the forest areas, first, from a date previous to the forest fire; and second, from a date following the fire. The methodology for the forest fire risk evaluation is based on the temporal evolution of the NDVI weekly maximum value. Actual forest fires appear to be statistically correlated with the deduced high risk forest fire areas.     

Satellite-based forest fire detection for fire control in boreal forests

Forest fires in large sparsely populated areas in the boreal forest zone are difficult to detect by ground based means. Satellites can be a viable source of information to augment air-borne reconnaissance. The Advanced Very High Resolution Radiometer (AVHRR) sensor aboard the National Oceanic and Atmospheric Administration (NOAA) satellites has been used to detect and map fires in the past mainly in the tropics and mainly for environmental monitoring purposes. This article describes real-time forest fire detection where the aim is to inform local fire authorities on the fire. The fire detection is based on the 3.7 mu m channel of the NOAA AVHRR sensor. In the fire detection algorithm, imaging geometry is taken into account in addition to the data from the near-infrared and thermal infrared channels. In an experiment in summer 1995, 16 fires were detected in Finland. One was a forest fire, 11 were prescribed burnings and 4 false alarms. Three of the false alarms were due to steel factories. We conclude that satellite-based fire detection for fire control is feasible in the boreal forest zone if the continuous supply of frequent middle-infrared data can be guaranteed in the future.     

Evaluation of novel thermally enhanced spectral indices for mapping fire perimeters and comparisons with fire atlas data

We evaluated the potential of two novel thermally enhanced Landsat Thematic Mapper (TM)‐derived spectral indices for discriminating burned areas and for producing fire perimeter data (as a potential surrogate to digital fire atlas data) within two wildland fires (1985 and 1993) in ponderosa pine (Pinus ponderosa) forests of the Gila Wilderness, New Mexico, USA. Image‐derived perimeters (manually produced and classified from an index image) were compared to fire perimeters recorded within a digitized fire atlas. For each fire, the highest spectral separability was achieved using the newly proposed Normalized Burn Ratio‐Thermal (NBRT₁) index (M = 1.18, 1.76, for the two fires respectively). Correspondence between fire atlas and manually digitized fire perimeters was high. Landsat imagery may be a useful supplement to existing historical fire perimeters mapping methods, but the timing of the post‐fire image will strongly influence the separability of burned and unburned areas.     

Evaluation of remote sensing-based active fire datasets in Indonesia

Eight different fire datasets for Indonesia were compared with each other and to fine spatial resolution burnscar maps. Results show that each dataset detects different fires. More than two-thirds of the fires detected by one dataset are not detected by any other dataset. None of the datasets detect fires in all test areas. Fire regime, satellite sensor characteristics and fire detection algorithms all influence which fires are detected. Fire datasets were not complementing each other as they all had commission as well as omission errors.     

Neural network and GA approaches for dwelling fire occurrence prediction

This paper describes three approaches for the prediction of dwelling fire occurrences in Derbyshire, a region in the United Kingdom. The system has been designed to calculate the number of fire occurrences for each of the 189 wards in the Derbyshire. In terms of the results from statistical analysis, eight factors are initially selected as the inputs of the neural network. Principal Component Analysis (PCA) is employed for pre-processing the input data set to reduce the number of the inputs. The first three principal components of the available data set are chosen as the inputs, the number of the fires as the output. The first approach is a logistic regression model, which has been widely used in the forest fire prediction. The prediction results of the logistic regression model are not acceptable. The second approach uses a feed-forward neural network to model the relationship between the number of fires and the factors that influence fire occurrence. The model of the neural network gives a prediction with an acceptable accuracy for the fires in dwelling areas. Genetic algorithms (GAs) are the third approach discussed in this study. The first three principle components of the available data set are classified into the different groups according to their number of fires. An iterative GA is proposed and applied to extract features for each data group. Once the features for all the groups have been identified the test data set can be easily clustered into one of the groups based on the group features. The number of fires for the group, which the test data belongs to, is the prediction of the fire occurrence for the test data. The three approaches have been compared. Our results indicate that the neural network based and the GA based approaches perform satisfactorily, with MSEs of 2.375 and 2.875, respectively, but the GA approach is much better understood and more transparent.     

基于贝叶斯网络的林火概率预测系统设计与实现

针对林火预测具有影响因素多、机制复杂、难以结构化等特点,设计并实现了一个基于贝叶斯网络的实用林火概率预测系统。该系统以气象、植被、地理、人类活动等数据作为输入,综合林火历史数据建立贝叶斯网络模型,并应用联合树算法进行概率推理,进而预测出林火发生概率。在某省实际林火历史数据上对系统进行了测试,比较了所设计系统与加拿大火险天气指标系统（FWI）的预测性能,验证了系统的可行性和实用性。     

Using multiple endmember spectral mixture analysis to retrieve subpixel fire properties from MODIS

The Moderate-Resolution Imaging Spectroradiometer (MODIS) sensors on NASA's Terra and Aqua satellites image most of the Earth multiple times each day, providing useful data on fires that cannot be practically acquired using other means. Unfortunately, current fire products from MODIS and other sensors leave large uncertainties in measurements of fire sizes and temperatures, which strongly influence how fires spread, the amount and chemistry of their gas and aerosol emissions, and their impacts on ecosystems. In this study, we use multiple endmember spectral mixture analysis (MESMA) to retrieve subpixel fire sizes and temperatures from MODIS, possibly overcoming some limitations of existing methods for characterizing fire intensities such as estimating the fire radiative power (FRP). MESMA is evaluated using data from the Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) to assess the performance of FRP and MESMA retrievals of fire properties from a simultaneously acquired MODIS image, for a complex of fires in Ukraine from August 21, 2002. The MESMA retrievals of fire size described in this paper show a slightly stronger correlation than FRP does to fire pixel counts from the coincident ASTER image. Prior to this work, few studies, if any, had used MESMA for retrieving fire properties from a broad-band sensor like MODIS, or compared MESMA to higher-resolution fire data or other measures of fire properties like FRP. In the future, MESMA retrievals could be useful for fire spread modeling and forecasting, reducing hazards that fires pose to property and health, and enhancing scientific understanding of fires and their effects on ecosystems and atmospheric composition.     

Detection rates of the MODIS active fire product in the United States

MODIS active fire data offer new information about global fire patterns. However, uncertainties in detection rates can render satellite-derived fire statistics difficult to interpret. We evaluated the MODIS 1 km daily active fire product to quantify detection rates for both Terra and Aqua MODIS sensors, examined how cloud cover and fire size affected detection rates, and estimated how detection rates varied across the United States. MODIS active fire detections were compared to 361 reference fires (≥ 18 ha) that had been delineated using pre- and post-fire Landsat imagery. Reference fires were considered detected if at least one MODIS active fire pixel occurred within 1 km of the edge of the fire. When active fire data from both Aqua and Terra were combined, 82% of all reference fires were found, but detection rates were less for Aqua and Terra individually (73% and 66% respectively). Fires not detected generally had more cloudy days, but not when the Aqua data were considered exclusively. MODIS detection rates decreased with fire size, and the size at which 50% of all fires were detected was 105 ha when combining Aqua and Terra (195 ha for Aqua and 334 ha for Terra alone). Across the United States, detection rates were greatest in the West, lower in the Great Plains, and lowest in the East. The MODIS active fire product captures large fires in the U.S. well, but may under-represent fires in areas with frequent cloud cover or rapidly burning, small, and low-intensity fires. We recommend that users of the MODIS active fire data perform individual validations to ensure that all relevant fires are included.     

Volume rendering of pool fire data

In a pool fire, an ignited puddle or pool of liquid fuel burns in the atmosphere. Understanding pool fires is important to devising methods to control the hazards resulting from spilled fuels. In this case study, we consider techniques for visualizing the data measured in pool fires and for computing the radiative transfer from pool fires. We used basic tools to image fire data and to compute irradiation-efficient ray casting and radiometrically accurate line integrations. Visualization techniques help answer a number of questions about pool fires     

Use of NOAA-AVHRR NDVI images for the estimation of dynamic fire risk in Mediterranean areas

Wildfires are a major cause of land degradation in the Mediterranean region due to their frequent recurrence in the same areas. The evaluation of fire risk is therefore of high practical importance, particularly during the summer arid season, when fires are most frequent and harmful. Recent studies have demonstrated that the evaluation of dynamic fire risk can be carried out by the use of remotely sensed images, and specifically of NOAA-AVHRR Normalized Difference Vegetation Index (NDVI) data. This use relies on the sensitivity of the index to vegetation dryness, which is a major predisposing factor for fire occurrence. Several problems, however, remain linked to the spatial variability of the risk in environmentally heterogeneous areas, which requires the application of suitable processing techniques to the low-resolution imagery.The current work reports on the development and testing of different methodologies for estimating dynamic fire risk by the use of NOAA-AVHRR data. The investigation was conducted in Tuscany (Central Italy) using a large archive of fires that occurred in the region and NOAA-AVHRR NDVI data of 16 years (1985-2000). Relying on previous methodological achievements of our group and other research groups, several procedures were tested to extract information related to fire risk from the remotely sensed images. These trials led to define an optimum method which is based on the identification of pixels where the accordance between interyear variations in fire probabilities and NDVI values is maximum. The accuracy of the risk estimates from this optimum method was finally evaluated by a leave-one-out cross-validation strategy. In this way, the performance of the methodology was assessed, together with its potential for operational fire risk monitoring and forecasting in Mediterranean areas.     

On the use of fuzzy synthetic evaluation and optimal classification for computing fire risk ranking of buildings

Fire risk ranking is particularly useful for building designers to compare two different solutions to assess if the safety is similar. However, the multi-criteria and imprecise nature of the fire safety attributes in buildings has caused difficulties in quantifying the fire safety level. Further to the previous works, the author presented a fuzzy synthetic evaluation system for computing the fire risk ranking of buildings. It could serve for multi-level fire safety assessment framework. Linguistic terms were adopted instead of subjective numerical values. A two-level evaluation model, including fuzzy optimal classification model and linear weighted mean model, was developed to facilitate the synthetic process. A case of fire safety ranking in high-rise residential buildings in Hong Kong was presented for illustration purpose. The evaluation result was analyzed by the maximum membership degree principle method.     

A BIM-based simulation framework for fire safety management and investigation of the critical factors affecting human evacuation performance

Fire hazards are a big threat to human life and property safety. The U.S. fire statistics reveal that, in 2017 alone, 1,319,500 fires caused 3400 deaths and 14,670 injuries, which resulted in a loss of $23 billion [1]. Effective evacuation planning in densely occupied buildings should be primarily put in place if both the number of injuries/fatalities and the level of property loss are to be minimized. However, it is not realistic, and is unethical to study human evacuation performance under a burning building. For this reason, computational tools tend to be the best approach for simulating fire growth as well as human response to fire hazards. This study aims to develop a BIM-based simulation framework that implements the Fire Dynamic Simulator (FDS) and agent-based modeling (ABM) for simulating fire growth and evacuation performance for different building layout scenarios. An experimental implementation is conducted to validate the proposed framework, which verified the benefits of (1) using BIM to offer a platform for conducting simulation design and visualizing the simulation results of (a) hazardous fire zones and (b) effective escape routes; (2) simulating fire growth using the FDS tool; (3) developing an agent-based model that accounts for the critical factors affecting evacuation performance; and (4) applying a statistical analysis for investigating the effects of influential parameters from the proposed model. As a result, the simulation outputs can be used to optimize the building design and to investigate the influential factors on human evacuation efficiency. The proposed framework contributes to building fire safety management by enabling to minimize both injuries/fatalities and property loss.     

Combining AVHRR and ATSR satellite sensor data for operational boreal forest fire detection

The potential to combine data from two different satellite systems was studied to increase fire detection sensitivity and image acquisition frequency in real-time fire detection and fire control. A fully automatic fire detection algorithm was applied to all scenes that were acquired using both satellite systems. Local fire authorities were notified about each detected fire in their territory using real-time fire reports that were sent by telefax. The average time from the start of National Oceanic and Atmospheric Administration (NOAA), Advanced Very High Resolution Radiometer (AVHRR) image acquisition until the sending of a telefax fire report was 25 min. During the straw-burning season in April 2000, the Along Track Scanning Radiometer (ATSR) instrument detected twice as many fires as the AVHRR per unit image area. The main reason for this may be the average resolution cell of the ATSR, which is half the size of that of the AVHRR in terms of area. The response from fire authorities was used to estimate the number of correct alerts and false alarms. A false alarm rate of 12% and 7% was obtained in the fire seasons of 1999 and 2000, respectively.     

The use of spatial analytical techniques to explore patterns of fire incidence: A South Wales case study

The application of mapping and spatial analytical techniques to explore geographical patterns of crime incidence is well established. In contrast, the analysis of operational incident data routinely collected by fire brigades has received relatively less research attention, certainly in the UK academic literature. The aim of this paper is to redress this balance through the application of spatial analytical techniques that permit an exploration of the spatial dynamics of fire incidents and their relationships with socio-economic variables. By examining patterns for different fire incident types, including household fires, vehicle fires, secondary fires and malicious false alarms in relation to 2001 Census of Population data for an area of South Wales, we demonstrate the potential of such techniques to reveal spatial patterns that may be worthy of further contextual study. Further research is needed to establish how transferable these findings are to other geographical settings and how replicable the findings are at different geographical scales. The paper concludes by drawing attention to the current gaps in knowledge in analysing trends in fire incidence and proposes an agenda to advance such research using spatial analytical techniques.     

An ontology to represent firefighters data requirements during building fire emergencies

Firefighters require accurate and timely information regarding a building and its environment to perform their duty safely and successfully during a fire emergency. However, due to the chaotic nature of building fires, firefighters often receive erroneous, conflicting, or delayed information that can affect the outcome of an emergency. In this paper, we propose a solution in the form of an ontology that defines building and environmental data that is needed by firefighters during a building fire emergency. The ontology development followed the METHONTOLOGY method and was implemented using the web ontology language (OWL) in Protégé 5.5.0. Built-in reasoners in Protégé and an ontology pitfall scanning tool were employed to verify the structure and consistency of the new ontology. To validate the ontology efficacy, we developed a prototype web application to represent and visualise relevant information based on the ontology and used that as a basis for conducting interviews with firefighters. Finally, a specification document that describes the ontology was created and published online. The proposed ontology can be a basis for developing intelligent tools and systems that support firefighters.