Final Fire Size Research Articles

Saharan Hot and Dry Sirocco Winds Drive Extreme Fire Events in Mediterranean Tunisia (North Africa)

With hot and dry summers, the Mediterranean basin is affected by recurrent fires. While drought is the major driver of the seasonal and inter-annual fire distribution in its northern and mildest climate conditions, some extreme fire events are also linked to extreme winds or heat waves. The southern part of the Mediterranean basin is located at the driest range of the Mediterranean bioclimate and is influenced by Saharan atmospheric circulations, leading to extreme hot and dry episodes, called Sirocco, and potentially acting as a major contributor to fire hazard. The recently created fire database for Tunisia was used to investigate the ±10-day pre- and post-fire timeframe of daily weather conditions associated with fire events over the 1985–2006 period. Positive anomalies in minimum and maximum temperatures, negative anomalies in air relative humidity, and a preferential south-eastern wind during fire events were identified, which were characteristic of Sirocco winds. +7 °C anomalies in air temperature and −30% in relative air humidity were the critical thresholds for the most extreme fire conditions. In addition, meteorological anomalies started two days before fire events and lasted for three days after for large fires >400 ha, which suggests that the duration of the Sirocco event is linked with fire duration and final fire size. Lastly, the yearly number of intense Sirocco events better explained the inter-annual variability of burned area over the 1950–2006 period than summer drought based on Standardized Precipitation Evaporation Index (SPEI) indices.

Factors influencing fire suppression success in the province of Quebec (Canada)

In the managed forest of Canada, forest fires are actively suppressed through efficient initial attack capability; however, the impact of different factors on the suppression success remains to be understood. The aim of this paper was to analyze the influence of operational suppression objectives (fire detection, initial attack, and fire control) along with fire intensity, fuel type, fire ignition cause, year, workload, and homogeneous fire regime zones on the achievement of the fire suppression objective (fire < 3 ha) using the Forest Fire Protection Agency of Quebec (SOPFEU) as a case study. The overall success of the suppression objective was very high (88%) over the study period (1994–2015). Both detection and control had significant effects on the suppression success through their interaction with fuel type, ignition cause, fire intensity, and zone variables. When the suppression objective was not achieved, final fire size was influenced by control, fuel type, fire intensity, and zone. The paper highlights the importance of the operational objectives and of regional differences for both fire suppression success and final fire size. Our results can help forest fire protection agencies to better understand their wildland fire suppression systems for a better adaptation to the upcoming fire regime changes.

Machine learning to predict final fire size at the time of ignition.

Fires in boreal forests of Alaska are changing, threatening human health and ecosystems. Given expected increases in fire activity with climate warming, insight into the controls on fire size from the time of ignition is necessary. Such insight may be increasingly useful for fire management, especially in cases where many ignitions occur in a short time period. Here we investigated the controls and predictability of final fire size at the time of ignition. Using decision trees, we show that ignitions can be classified as leading to small, medium or large fires with 50.4 ± 5.2% accuracy. This was accomplished using two variables: vapour pressure deficit and the fraction of spruce cover near the ignition point. The model predicted that 40% of ignitions would lead to large fires, and those ultimately accounted for 75% of the total burned area. Other machine learning classification algorithms, including random forests and multi-layer perceptrons, were tested but did not outperform the simpler decision tree model. Applying the model to areas with intensive human management resulted in overprediction of large fires, as expected. This type of simple classification system could offer insight into optimal resource allocation, helping to maintain a historical fire regime and protect Alaskan ecosystems.

A comparison of the US National Fire Danger Rating System (NFDRS) with recorded fire occurrence and final fire size

Most previous research has assessed the ability of the National Fire Danger Rating System (NFDRS) to portray fire activity at either single sites or on small spatial scales, despite it being a nation-wide system. This study seeks to examine the relationships between a set of NFDRS fire danger indices (Fire Danger Ratings, Staffing Level and the Ignition Component) and measures of fire activity (fire occurrence and final fire size) across the entire conterminous US over an 8-year period. We reveal that different regions of the US display different levels of correspondence between each of the fire danger indices and recorded fire activity. Areas in the Southern and Eastern Geographic Area Coordination Centers (GACCs) exhibit weaker correlations than those in the Northwest, Northern Rockies, Great Basin and Northern California GACCs. Peaks in fire occurrence are shown to occur at mid–low values of fire danger whereas final fire sizes increase monotonically with each fire danger index. Our findings appear to align with perceived shifts in management practices currently employed across the US and indicate that the ability of the NFDRS to apportion the resources required to combat large fires is in general well developed.

The drivers of wildfire enlargement do not exhibit scale thresholds in southeastern Australian forests

Wildfires are complex adaptive systems, and have been hypothesized to exhibit scale-dependent transitions in the drivers of fire spread. Among other things, this makes the prediction of final fire size from conditions at the ignition difficult. We test this hypothesis by conducting a multi-scale statistical modelling of the factors determining whether fires reached 10 ha, then 100 ha then 1000 ha and the final size of fires >1000 ha. At each stage, the predictors were measures of weather, fuels, topography and fire suppression. The objectives were to identify differences among the models indicative of scale transitions, assess the accuracy of the multi-step method for predicting fire size (compared to predicting final size from initial conditions) and to quantify the importance of the predictors. The data were 1116 fires that occurred in the eucalypt forests of New South Wales between 1985 and 2010.The models were similar at the different scales, though there were subtle differences. For example, the presence of roads affected whether fires reached 10 ha but not larger scales. Weather was the most important predictor overall, though fuel load, topography and ease of suppression all showed effects. Overall, there was no evidence that fires have scale-dependent transitions in behaviour. The models had a predictive accuracy of 73%, 66%, 72% and 53% accuracy at 10 ha, 100 ha, 1000 ha and final size scales. When these steps were combined, the overall accuracy for predicting the size of fires was 62%, while the accuracy of the one step model was only 20%. Thus, the multi-scale approach was an improvement on the single scale approach, even though the predictive accuracy was probably insufficient for use as an operational tool. The analysis has also provided further evidence of the important role of weather, compared to fuel, suppression and topography in driving fire behaviour.

A spatial database of wildfires in the United States, 1992-2011

Abstract. The statistical analysis of wildfire activity is a critical component of national wildfire planning, operations, and research in the United States (US). However, there are multiple federal, state, and local entities with wildfire protection and reporting responsibilities in the US, and no single, unified system of wildfire record keeping exists. To conduct even the most rudimentary interagency analyses of wildfire numbers and area burned from the authoritative systems of record, one must harvest records from dozens of disparate databases with inconsistent information content. The onus is then on the user to check for and purge redundant records of the same fire (i.e., multijurisdictional incidents with responses reported by several agencies or departments) after pooling data from different sources. Here we describe our efforts to acquire, standardize, error-check, compile, scrub, and evaluate the completeness of US federal, state, and local wildfire records from 1992–2011 for the national, interagency Fire Program Analysis (FPA) application. The resulting FPA Fire-Occurrence Database (FPA FOD) includes nearly 1.6 million records from the 20 yr period, with values for at least the following core data elements: location, at least as precise as a Public Land Survey System section (2.6 km2 grid), discovery date, and final fire size. The FPA FOD is publicly available from the Research Data Archive of the US Department of Agriculture, Forest Service (doi:10.2737/RDS-2013-0009). While necessarily incomplete in some aspects, the database is intended to facilitate fairly high-resolution geospatial analysis of US wildfire activity over the past two decades, based on available information from the authoritative systems of record.

An econometric model for ex ante prediction of wildfire suppression costs

This paper develops an econometric model that can provide predictions of fire suppression costs (per acre and in total) for a given large fire before final fire acreage is known. The model jointly estimates cost per acre and acreage equations via Maximum Likelihood, accounting for sample truncation based on final fire size. Formulas and results are shown for predictions of costs and fire size for wildfires in general, and for large fires in particular. Marginal effects of explanatory variables on cost and acreage are discussed. The distribution of these model predictions illustrates the importance of accounting for sample truncation when generating predicted outcomes based on ex ante information.

Fuel treatment effects on modeled landscape-level fire behavior in the northern Sierra Nevada

Across the western United States, decades of fire exclusion combined with past management history have contributed to the current condition of extensive areas of high-density, shade-tolerant coniferous stands that are increasingly prone to high-severity fires. Here, we report the modeled effects of constructed defensible fuel profile zones and group selection treatments on crown fire potential, flame length, and conditional burn probabilities across 11 land allocation types for an 18 600 ha study area within the northern Sierra Nevada, California. Fire modeling was completed using FlamMap and FARSITE based on landscape files developed with high-resolution aerial (IKONOS) imagery, ground-based plot data, and integrated data from ARCFUELS and the Forest Vegetation Simulator. Under modeled 97th percentile weather conditions, average conditional burn probability was reduced between pre- and post-treatment landscapes. A more detailed simulation of a hypothetical fire burning under fairly severe fire weather, or “problem fire”, revealed a 39% reduction in final fire size for the treated landscape relative to the pre-treatment condition. To modify fire behavior at a landscape level, a combination of fuel treatment strategies that address topographic location, land use allocations, vegetation types, and fire regimes is needed.

Assessing the Accuracy of Wildland Fire Situation Analysis (WFSA) Fire Size and Suppression Cost Estimates

Abstract To determine the optimal suppression strategy for escaped wildfires, federal land managers are required to conduct a wildland fire situation analysis (WFSA). As part of the WFSA process, fire managers estimate final fire size and suppression costs. Estimates from 58 WFSAs conducted during the 2002 fire season are compared to actual outcomes. Results indicate that estimates of fire size and suppression costs are systematically biased. Modifications to the WFSA process are suggested to address these problems.

Evaluation of fire danger rating indexes using logistic regression and percentile analysis

Methods are presented for analysing the relationship between fire danger rating indexes and fire activity as a means of evaluating the performance of fire danger rating systems. Percentile analysis is used to examine the data itself; logistic regression provides a means for statistical analysis. Ranking of selected items indicates indexes that deserve further assessment using subjective considerations. Methods rely on generally available data: the fire danger index for every day in the fire season, fire discovery date, and final fire size. For logistic regression analysis, the independent variable is the index, and dependent variables are fire-day, large-fire-day, and multiple-fire-day. Data analysis methods have been incorporated into the FireFamily Plus computer program for easy application. Potential uses of the analysis include choosing the most appropriate fire danger index and fuel model for an area, evaluating proposed changes to a fire danger rating system, and assessing the performance of a system in a location other than that for which it was designed. As a demonstration, this technique was applied to evaluation of several indexes and fuel models of the U.S. National Fire Danger Rating System on the Tonto National Forest in Arizona, USA, using fire and weather data for 1974–2001.

Simulating Wildfire Containment with Realistic Tactics

Abstract Existing simulation models for fire protection planning rely on a containment algorithm which fails to account for the interaction between the production of containment line and a fire's capacity to spread. This paper describes a technique for simulating wildland fire containment which explicitly accounts for this interaction by extending, generalizing, and, in some cases, simplifying methods reported earlier (Anderson 1983, Albini et al. 1978, Mees 1985, Anderson 1989). Representing the containment boundary in parametric form [i.e., expressing its Cartesian (x, y) or polar (r, ) coordinates as functions of a dimensionless parameter instead of specifying y as a function of x or r as a function of ], leads to a formalism with two significant advantages: (1) it allows a free choice for the shape of the free burning fire boundary, and (2) it is appropriate for simulations of parallel (indirect) attack as well as direct head and tail attack. In general, the technique requires solution of a first-order, nonlinear, differential equation, a task easily accomplished using numerical methods. The dependence of final fire size and containment time on the line-building rate and eccentricity of the free burning fire boundary are illustrated with the special case of an expanding ellipse and constant line-building rate, for which the containment boundary can be obtained by a simple quadrature. Our General Formulation algorithm is demonstrated with a series of simulations that span the range of likely parameters for fire spread and fireline production. Comparisons made on representative fires between predictions of the General Formulation and models currently in use indicate that current models almost always overestimate fire size, sometimes by as much as an order of magnitude. For. Sci. 42(3):267-281.

Sensitivity of fire size to fireline construction rates in a simulation model

An initial attack model was sensitive to changes in production rate of suppression forces, as shown by changes in final fire size.