Fire Propagation Prediction Research Articles

Prediction of Fire Propagation in Forest Areas using Genetic Algorithm

Forest fire causes significant damages to environment and living organisms. It becomes troublesome to recover from the consequences of forest fire. The early and accurate prediction of forest fire propagation in forest area will help in fire management efforts. The models required for prediction of forest fire spread requires many parameters and which are not known precisely during the emergency. So there is a need of intelligent system to model and predict the propagation of forest fire to reduce the consequences of forest fire and minimize the management efforts. In this work, we have discussed different techniques available in literature for forest fire prediction. This paper also describes the implementation and analysis of prediction of forest fire propagation using Genetic Algorithm and Geometric Semantic Genetic Programming.

Dynamic Wildfire Navigation System

Wildfire, a natural part of many ecosystems, has also resulted in significant disasters impacting ecology and human life in Australia. This study proposes a prototype of fire propagation prediction as an extension of preceding research; this system is called “Cloud computing based bushfire prediction”, the computational performance of which is expected to be about twice that of the traditional client-server (CS) model. As the first step in the modelling approach, this prototype focuses on the prediction of fire propagation. The direction of fire is limited in regular grid approaches, such as cellular automata, due to the shape of the uniformed grid, while irregular grids are freed from this constraint. In this prototype, fire propagation is computed from a centroid regardless of grid shape to remove the above constraint. Additionally, the prototype employs existing fire indices, including the Grassland Fire Danger Index (GFDI), Forest Fire Danger Index (FFDI) and Button Grass Moorland Fire Index (BGML). A number of parameters, such as Digital Elevation Model (DEM) and forecast weather data, are prepared for use in the calculation of the indices above. The fire study area is located around Lake Mackenzie in the central north of Tasmania where a fire burnt approximately 247.11 km 2 in January 2016. The prototype produces nine different prediction results with three polygon configurations, including Delaunay Triangulation, Square and Voronoi, using three different resolutions: fine, medium and coarse. The Delaunay Triangulation, which has the greatest number of adjacent grids among three shapes of polygon, shows the shortest elapsed time for spread of fire compared to other shapes. The medium grid performs the best trade-off between cost and time among the three grain sizes of prediction polygons, and the coarse size shows the best cost-effectiveness. A staging approach where coarse size prediction is released initially, followed by a medium size one, can be a pragmatic solution for the purpose of providing timely evacuation guidance.

Fire propagation visualization in real time

Our motivation comes from the need of a tailored computational tool for simulation and prediction of forest fire propagation, to be used by firefighters in Patagonia, Argentina. Based on previous works on Graphic Processing Units (GPU) for fitting and simulating fires in our region, we developed a visualization interface for real time computing, simulation and prediction of fire propagation. We have the possibility of changing the ensemble of raster maps layers to change the region in which fire will propagate.The visualization platform runs on GPUs and the user can rotate and zoom the landscape to select the optimal view of fire propagation. Opacity of different layers can be regulated by the user, allowing to see fire propagation at the same time that underlying vegetation, wind direction and intensity. The ignition point can also be selected by the user, and firebreaks can be plotted while simulation is going on.After the performance of a high number of stochastic simulations in parallel in GPUs, the application shows a map of the final fire surface colored according to the probability that a given cell burns. In this way the user can visually identify the most critical direction for fire propagation, a useful information to stop fire optimizing resources, which is specially important when they are scarce like is the case of our Patagonia region.

Assessing improvements in models used to operationally predict wildland fire rate of spread

The prediction of fire propagation across landscapes is necessary for safe and effective fire management. We analyzed the predictive accuracy of models currently used operationally in Australia for predicting fire spread rates in five different fuel types (grasslands, temperate and semi-arid shrublands, dry eucalypt and conifer forests) compared to their previous counterparts. We calculated error statistics and contrasted model predictions against observed spread rates of field observations of wildfires and prescribed fires. We then compared the changes in error metrics of older models to newer ones. Evaluation results show newer models to have improved prediction accuracy. Mean absolute errors were reduced by 56%, 68% and 70% in dry eucalypt forests, grasslands and crown fires in conifer forests, respectively. The most significant improvement was the reversion of under-prediction bias achieved with newer models. This study has highlighted the value of continuous improvement when it comes to developing operational wildland fire spread models.

Wind field parallelization based on Schwarz alternating domain decomposition method

Wind field is a critical issue in forest fire propagation prediction. However, wind field calculation is a complex problem that, for large terrains, involves solving huge linear systems. To solve such systems, the Preconditioned Conjugate Gradient (PCG) solver is applied. SSOR and Jacobi preconditioners are usually used, but solving such systems takes too much time and makes the approach unfeasible in real time operation. Parallelization appears as a way to make the approach operational in real time. The PCG with both preconditioners has been parallelized to accelerate the execution. However, the improvement in execution time is not enough, and the Schwarz alternating domain decomposition has been applied to exploit a second level of parallelism. Using this method, the linear system is decomposed in a set of overlapped subdomains that can be solved in parallel using a Master/Worker paradigm, where each worker exploits the PCG solver parallelism. As a result, the wind field calculation time is significantly reduced; for example, a large map of 1200×1200 cells, whose solution took more than 2000 seconds in the original WindNinja, can now be solved in less than 240 seconds using 4 subdomain and 4 cores per subdomain.

Data resolution effects on a coupled data driven system for forest fire propagation prediction

Every year, thousands of forest hectares are burned worldwide, causing important consequences on the atmosphere, biodiversity and economy. Proper prediction of the fire evolution allows to manage the fire fighting equipment properly. Therefore, it is crucial to use reliable and fast simulations in order to predict the evolution of the fire. WRF-SFIRE is a wildland fire simulator, which couples a meteorological model called Weather Research and Forecasting Model (WRF) and a forest fire simulator, SFIRE reproducing the interaction between the propagation of the fire and the atmosphere evolution. The mesh resolution used to solve the atmosphere evolution has a deep impact in the prediction of small scale meteorological effects. At the same time, the ability of introducing these small scale meteorological events into the forest fire simulation implies enhancements in the quality of the data that drives the simulation. Therefore, better fire propagation predictions are obtained. However, this improvement can be affected by the instability of the problem to be solved. This paper states an accuracy study changing the mesh resolution when running WRF-SFIRE using as example a real case that took place in Catalonia (northeast of Spain) in 2005.

A comparative study of evolutionary statistical methods for uncertainty reduction in forest fire propagation prediction

Predicting the propagation of forest fires is a crucial point to mitigate their effects. Therefore, several computational tools or simulators have been developed to predict the fire propagation. Such tools consider the scenario (topography, vegetation types, fire front situation), and the particular conditions where the fire is evolving (vegetation conditions, meteorological conditions) to predict the fire propagation. However, these parameters are usually difficult to measure or estimate precisely, and there is a high degree of uncertainty in many of them. This uncertainty provokes a certain lack of accuracy in the predictions with the consequent risks. So, it is necessary to apply methods to reduce the uncertainty in the input parameters. This work presents a comparison of ESSIM-EA and ESSIM-DE: two methods to reduce the uncertainty in the input parameters. These methods combine Evolutionary Algorithms, Parallelism and Statistical Analysis to improve the propagation prediction.

Hybrid application to accelerate wind field calculation

Wind field calculation is a crucial point in predicting forest fire evolution in real emergencies. Such wind field calculation involves complex models that must be solved in a very short time to accomplish the real-time constraints of forest fire propagation prediction. However, when the terrain map under consideration is large (for example, 30km×30km) and the resolution is high (for example, 30m), the time incurred by wind field calculation makes its use unfeasible in real emergencies. To accelerate the wind field calculation, several approaches have been studied (map partitioning, domain decomposition and matrix storage). The results are quite significant, but in large cases, the execution time does not meet the expected objectives. Therefore, a new approach, integrating all the previous work, has been designed, and a hybrid MPI-OpenMP version of WindNinja has been implemented. This hybrid application exploits different sources of parallelism to reduce execution time without introducing a significant loss of accuracy in wind field calculation. The results obtained with different terrain maps show that the execution time is reduced to below an established limit of 100s.

Time aware genetic algorithm for forest fire propagation prediction: exploiting multi‐core platforms

SummaryForest fire propagation prediction is a crucial issue when fighting these hazards as efficiently as possible. Several propagation models have been developed and integrated in computer simulators. Such models require a set of input parameters that, in some cases, are difficult to know or even estimate precisely beforehand. Therefore, a calibration technique based on genetic algorithm (GA) was introduced to reduce the uncertainty in input parameters values and improve the accuracy of the predictions. Such a technique requires the execution of a set of simulations and several iterations of the process to calibrate the values of the input parameters. To reduce the execution time of this calibration stage, an Message Passing Interface master/worker scheme was developed to distribute the simulations of one iteration among the worker processes. However, the execution time of each simulation varies drastically depending on the particular input parameters used, provoking a significant load imbalance. To overcome this imbalance and reduce execution time to operational requirements, core allocation policies have been developed. These policies are based on execution time estimation and classification of simulations according to the estimated execution time. Then, multicore capabilities of the current systems are applied to devote more resources (cores) to the longest simulations reducing the load imbalance. These simulations that are estimated as taking too long, even when many resources are devoted to them, require especial consideration. So, a generation time limit has been introduced, and three different strategies have been designed considering individuals that exceed the generation execution time limit. In the first one, the longest individuals are replaced before starting the execution with shorter individuals (Time Aware Core allocation with replacement). In the second one, these individuals are executed, but when the generation limit is reached, the individuals still executing are killed (Time Aware Core allocation without replacement). In the third one, all the individuals are executed normally, and when the generation time limit is reached, the GA is applied considering the individuals that have finished their executions, while the individuals still executing are allowed to continue running and are considered by the GA when they finish. The three strategies have been tested in real scenarios, and the results show these policies significantly improve the calibration accuracy within the superimposed deadlines. © 2016 The Authors. Concurrency and Computation: Practice and Experience Published by John Wiley & Sons Ltd.

Determining map partitioning to minimize wind field uncertainty in forest fire propagation prediction

Wind speed and direction are the parameters that most significantly affect forest fire propagation. The accurate estimation of such parameters is critical in providing precise predictions of forest fire propagation. Wind speed and direction can be measured in meteorological stations or estimated from meteorological models, but, in both cases, they are obtained at a very low resolution. Moreover, the meteorological wind is modified by the terrain topography and a complete wind field is generated with different wind speed and direction at any point of the terrain. So, wind field models providing wind speed and direction at a very high resolution (30m) are crucial in forest fire propagation prediction. WindNinja is one of the most widely used wind field simulators in this area. However, when the terrain map under consideration is very large (30km×30km or more), the execution time becomes unaffordable and the simulator cannot be used in real operation. A map partitioning strategy was developed to parallelize the wind field calculation and reduce the execution time to make it affordable. Map partitioning introduces a certain error in wind field that is propagated to forest fire propagation prediction. So, a complete methodology has been developed to determine the map partitioning that accomplishes real operation execution time constraints and keeps the forest fire propagation error below feasible limits. The experimental results show the execution time reduction accomplished and the accuracy of the wind field generated and forest fire propagation prediction.

Large Forest Fire Spread Prediction: Data and Computational Science

The accurate prediction of forest fire propagation is a crucial issue to minimize its effects. So, several models have been developed to determine the forest fire propagation beforehand. Such models require several input parameters that, in some cases, cannot be known precisely in a real emergency. So, a Two-Stage methodology was developed to calibrate the input parameters to improve the quality of the prediction. This methodology was based on Genetic Algorithms which require the execution of many simulations. Moreover, when the fire is large some input parameters cannot be considered uniform among the whole fire and extra models must be introduced. One of these non-uniform parameters is wind. So, in this work a wind field model is introduced. This model implies more computation time and response time is the main constraint. The prediction must be provided as fast as possible to be useful, thus it is necessary to exploit all available computing resources. So a Hybrid MPI-OpenMP application has been developed to reach a response in the shortest possible time. This work focuses on reducing the execution time of a worker in a MPI Master/Worker structure analyzing the simulation software parts which compose the Fire Simulator System.

Adapting Map Resolution to Accomplish Execution Time Constraints in Wind Field Calculation

Abstract Forest fire propagation prediction is a key point to fight against such hazards. Several models and simulators have been developed to predict forest fire propagation. These models require input parameters such as digital elevation map, vegetation map, and other parameters describing the vegetation and meteorological conditions. Coupling wind field model and forest fire propagation model improves accuracy prediction, but increases significantly prediction time. This fact is critical since propagation prediction must be provided in advance to allow the control centers to manage firefighters in the best possible way. This work analyses WindNinja execution time, describes a WindNinja parallelisation based on map partitioning, determines the limitations of such methodology for large maps and presents an improvement based on adapting map resolution to accomplish execution time limitations.

Forest Fire Propagation Prediction Based on Overlapping DDDAS Forecasts

Forest fire devastate every year thousand of hectares of forest around the world. Fire behavior prediction is a useful tool to aid coordination and management of human and mitigation resources when fighting against these kind of hazards. Any fire spread forecast system requires to be fitted with different kind of data with a high degree of uncertainty, such as for example, me- teorological data and vegetation map among others. The dynamics of this kind of phenomena requires to develop a forecast system with the ability to adapt to changing conditions. In this work two different fire spread forecast systems based on the Dynamic Data Driven Application paradigm are applied and an alternative approach based on the combination of both predictions is presented. This new method uses the computational power provided by high performance computing systems to deliver the predictions under strict real time constraints.

Wind Field Uncertainty in Forest Fire Propagation Prediction

Forests fires are a significant problem especially in countries of the Mediterranean basin. To fight against these disasters, the accurate prediction of forest fire propagation is a crucial issue. Propagation models try to describe the future evolution of the forest fire given an initial scenario and certain input parameters. However, the data describing the real fire scenario are usually subject to high levels of uncertainty. Moreover, there are input parameters that present spatial and temporal variation that make the prediction less accurate. Therefore, to overcome such uncertainty and improve accuracy it is necessary to couple complementary models such as the case of the wind field model. Such models use the meteorological forecasted wind to provide the wind direction and speed depending on the topography of the terrain. We use WindNinja as wind field simulator. This simulator takes a lot of time to deliver the predictions and it is a serious problem because fire propagation prediction must accomplish strict time constraints. To solve this problem, we propose map partitioning and solving independently for each one of the parts. However, the model has problems concerning boundary effects which is an additional source of uncertainty. Therefore, it is necessary to apply certain degree of overlapping among parts to reach a stable wind field without inconsistencies and a minimum uncertainty.

Relieving the Effects of Uncertainty in Forest Fire Spread Prediction by Hybrid MPI-OpenMP Parallel Strategies

The accurate prediction of forest fire propagation is a crucial issue to minimize its effects. Several models have been developed to determine the forest fire propagation. Simulators implementing such models require diverse input parameters to deliver predictions about fire propagation. However, the data describing the actual scenario where the fire is taking place are usually subject to high levels of uncertainty.The input-data uncertainty represents a serious drawback for the correctness of the prediction. So, a two-stage methodol- ogy was developed to calibrate the input parameters in an adjustment stage so that the calibrated parameters are used in the prediction stage to improve the quality of the predictions. This way, we relieve the effects of such uncertainty. In this work, we take advantage of this two stage methodology applying Genetic Algorithms as the calibration technique. However, the use of Genetic Algorithms require the execution of many simulations. This fact, added to the eventual long executions of the underlying simulator (due to its inherent complexity), implies to deal with another serious problem: the time needed to deliver the predictions. To be useful, the prediction must be provided much faster than real time. So, it is necessary to exploit all available computing resources.In this work, we present a two-stage forest fire spread prediction hybrid MPI-OpenMP application based on the Master- Worker paradigm and the parallelization of the FARSITE simulator in order to minimize the response time. The results as regards the enhancement in the quality of the predictions are reported, as well as the results regarding the time saving obtained by this hybrid application.

Coupling Diagnostic and Prognostic Models to a Dynamic Data Driven Forest Fire Spread Prediction System

Forest fires cause important losses around the world every year. A good prediction of fire propagation is a crucial point to minimize the devastating effects of these hazards. Several models that represent this phenomenon and provide a prediction of its spread have been developed. These models need input parameters which are usually difficult to know or even estimate. A two- stage prediction methodology was proposed to improve the quality of these parameters. In this methodology, such parameters are calibrated according to real observations and then, used in the prediction step. However, there are several parameters, which are not uniform along the map, but vary according to the topography of the terrain. Besides, these parameters are not constant along time but they are strongly dynamic. In such cases, it is necessary to introduce complementary models that overcome both restrictions. In the former case, the need of a spatial distribution model of a given variable is needed to be able to provide a spatial distribution for a given variable along the whole terrain by starting from the measured values of that parameter in certain points of the terrain. In the case of time variability, a complementary model such as weather forecasting model, could enable the capability of dealing with dynamic behavior of these parameters along time. In this paper, we describe an enhanced two-stage prediction scheme, where both type of complementary models a wind field model and a weather prediction model are coupled to the prediction scheme by enabling the system to dynamically adapts to complex terrains and dynamic conditions.

Assessment of ForeFire/Meso-NH for wildland fire/atmosphere coupled simulation of the FireFlux experiment

Numerical simulations using a coupled approach between Meso-NH (Non-Hydrostatic) LES (Large Eddy Simulation) mesoscale atmospheric model and ForeFire wildland fire area simulator are compared to experimental data to assess the performance of the proposed coupled approach in predicting fine-scale properties of the dynamics of wildland fires. Meso-NH is a non-hydrostatic, large eddy simulation capable, atmospheric research model. ForeFire insures a front tracking of the fire front by means of Lagrangian markers evolving on the earth’s surface according to a physical rate-of-spread model. The atmospheric model forces the fire behavior through the surface wind field, whereas the fire forces the atmosphere simulation through surface boundary conditions of heat and vapor fluxes. The FireFlux experiment, an experimental 32Ha burn of tall grass instrumented with wind profilers and thermocouples, was designed specifically to estimate the atmospheric perturbation introduced by wildland fire. Comparisons of the simulations at different resolutions with the large-scale experiment validate the chosen coupling methodology and the choice of a coupled approach with a meso-scale atmospheric model for the prediction of wildland fire propagation. Distinct fire propagation behavior is simulated between coupled and non-coupled simulation. While the simulations did not reproduce high frequency perturbations, it is shown that the atmospheric model captures well atmospheric perturbations induced by combustion at the ground level in terms of behavior and amplitude.

Coupling Wind Dynamics into a DDDAS Forest Fire Propagation Prediction System

Natural hazards are significant problems that every year cause important loses around the world. A good prediction of the behavior of the hazards is a crucial issue to fight against them and to minimize the damages. The models that represent these phenomena need several input parameters and in many cases, such parameters are diffcult to know or even to estimate in a real scenario. So, a methodology based on the DDDAS paradigm was developed to calibrate the input parameters according to real observations of the behavior and evolution of the hazard. Such calibrated parameters are then used to provide an improved prediction for the next time interval. This methodology was tested on Forest Fire Propagation Prediction with significant results. The developed methodology takes the fire behavior and propagation during a time interval and then searches for the values of the input parameters that best reproduce the propagation of the fire during that interval. Several Artificial Intelligence (AI) methods were applied to carry out this search as fast as possible. The values of the parameters that best reproduce the behavior of the fire were then used as input parameters to predict the propagation during the next time interval. These parameters were considered constant during both time intervals and a single value for each parameter was used for the calibrating process and for the prediction stage. This methodology fits on the DDDAS paradigm since the prediction is dynamically driven by the system evolution. However, there are several parameters that are not constant through time, but they may vary dynamically. In the case of forest fires, a typical example is the wind. In some cases, when the time interval is short an average value for the wind can be a feasible value, but when the time interval is longer, in most cases, a single value cannot represent the variability of the wind. We can estimate wind behavior applying some complementary model. In this work, we are going a step further considering the dynamic behavior of such parameters. We propose an extension of the existing prediction scheme that takes into account the dynamically changing parameters by coupling a weather prediction system on a DDDAS Forest Fire Propagation Prediction system.

On the Way of Applying Urgent Computing Solutions to Forest Fire Propagation Prediction

A quick response becomes crucial in natural hazard management. When an emergency occurs, hazard evolution simulators are a very helpful tool for the teams in charge of making decisions. To perform the simulations, they rely on data which usually constitutes a big set of parameters, which have been previously recorded from observations, usually coming from remote sensors, images, etc. However, this data is frequently subject to a high degree of uncertainty. This data uncertainty also produces uncertainty in simulators’ results. To overcome this drawback, in previous works we developed a two-stage prediction method, which has been demonstrated to improve noticeably the quality of the predictions. The time incurred in performing this strategy, however, may vary significantly depending on different factors. As it is well known, the execution time of a particular simulator depends on the specific setting of the input parameters. Moreover, decision control centers in charge of making decisions to fight against the ongoing disaster require a certain degree of quality in the final prediction. Depending on how demanding are the quality and time constraints, the two-stage strategy may need the support of Urgent Computing solutions, in order to meet the requirements. In this work, we focus on forest fires spread prediction, as a real application case of study, and we expose our methodology to characterize both the time needed and the expected quality of the two-stage prediction method, so that we are able to determine the amount of computational resources needed to respond effciently to an eventual emergency. This way, we emphasize the need of Urgent Computing mechanisms to be able to put into practice this method at real time.

Prediction of fire and smoke propagation in an underwater tunnel

Different models and solvers are used to calculate the spread of fire and smoke in a tunnel. The methodology for obtaining the numerical solution of this fire dynamics problem involves commercial software and a research program. Both can handle geometries described in three dimensions. Particular emphasis was placed on road tunnels in which vehicles are present. The specific application of this work is a study of a fire scenario in the Louis-Hippolyte-Lafontaine Tunnel which runs under a river in the Montreal area. Besides standard representation, visualization is also used, with elements which consider the optical properties of the phenomenon for a realistic rendering of smoke and fire.