Gemma Sanjuan

Wind Field Uncertainty in Forest Fire Propagation Prediction.

Abstract 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.

Determining map partitioning to accelerate wind field calculation

Wind speed and direction are parameters that affect forest fire propagation dramatically. So, an accurate estimation of such parameters is crucial to predict the fire propagation precisely. WindNInja is a wind field simulator that can easily be coupled to a forest fire propagation simulator such as FARSITE. However, wind field simulators present to main drawbacks: They take too much time to compute the wind field and they require a lot of memory. So, a map partitioning strategy has been developed to compute partial wind field maps that can be aggregated afterwards. Each map part can be computed in parallel and the amount of memory required is available in a single node. In this work a methodology to determine the most adequate map partitioning is presented. The map part shape, map part size, amount of overlapping and number of parts have been studied considering execution time and effects on wind field estimation. The results are based on a wide experimentation and are validated with real case scenarios.

Applying vectorization of diagonal sparse matrix to accelerate wind field calculation

Wind field calculation is a critical issue in reaching accurate forest fire propagation predictions. However, when the involved terrain map is large, the amount of memory and the execution time can prevent them from being useful in an operational environment. Wind field calculation involves sparse matrices that are usually stored in CSR storage format. This storage format can cause sparse matrix-vector multiplications to create a bottleneck due to the number of cache misses involved. Moreover, the matrices involved are extremely sparse and follow a very well-defined pattern. Therefore, a new storage system has been designed to reduce memory requirements and cache misses in this particular sparse matrix-vector multiplication. Sparse matrix-vector multiplication has been implemented using this new storage format and taking advantage of the inherent parallelism of the operation. The new method has been implemented in OpenMP, MPI and CUDA and has been tested on different hardware configurations. The results are very promising and the execution time and memory requirements are significantly reduced.

Applying domain decomposition to wind field calculation

Predicting forest fire propagation is a crucial issue to mitigate fire damages.Wind significantly affects forest fire propagation.Wind field calculation implies solving extremely large systems of equations.The Schur method has been applied to accelerate wind field calculation.Execution time is reduced to accomplish operational requirements. Forest fire are natural hazards that every year cause significant looses. Predicting the evolution of a forest fire is a critical issue in mitigating its effects. Such predictions must accomplish strict real time constraints to be effective. Wind field calculation is a key issue in providing accurate forest fire propagation predictions. However, it implies solving large linear systems with 105 to 108 variables that takes too long using conventional methods. Therefore, the domain decomposition Schur method has been applied to accelerate wind field calculation. Using the Schur method, the linear system is significantly reduced and several phases can be parallelised exploiting cluster computing capabilities. Results show that the execution time for the wind field calculation of a map of 800 × 800 cells has been reduced from 400 s to 90 s using 10 nodes.

Wind field parallelization based on Schwarz alternating domain decomposition method

Abstract 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.

Applying domain decomposition Schwarz method to accelerate wind field calculation

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. Solving such systems takes too much time and makes the approach unfeasible in real time operation. To overcome this problem the Schwarz alternating domain decomposition can be applied. 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 and the wind field calculation time can be significantly reduced.

Accelerating preconditioned conjugate gradient solver in wind field calculation

Wind field calculation is a common problem in different environmental applications from design of wind farms to forest fire propagation prediction. Calculating the wind field is a complex problem that involves solving huge linear systems. Solving such systems requires the use of iterative methods, such as Preconditioned Conjugate Gradient (PCG) that in most cases take long execution time. The PCG solver with different preconditioners has been analyzed and the performance and scalability of this solver has been determined. The most time consuming operations have been identified and a new method has been developed to improve the parallelization reducing the execution time and increasing the scalability. The new method has been applied on a wind field simulator, called WindNinja, usually coupled to forest fire propagation models. The results are very promising and the new parallelization method appears as a key point to be integrated in other approaches.

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.

Effect of Wind Field Parallelization on Forest Fire Spread Prediction

Forest fire spread prediction is a crucial issue to mitigate forest fire effects. Forest fire propagation models require several input parameters describing the conditions where the fire is taking place. However, some parameters, such as wind, present a different value on each point of the terrain due to topography. So, it is necessary to couple a wind field model that evaluates the wind on each terrain point. However, calculating the wind for each point on large maps is a time consuming task that can make the prediction unfeasible. So, it is necessary to parallelize the wind field computation. One approach is to apply a map partitioning technique, so that the wind field is calculated for each map part. The wind field obtained is lightly different from the one obtained with a single global map, and it is necessary to evaluate the effect of such difference on forest fire spread prediction.