We describe the development and implementation of an operational human-caused wildland fire occurrence prediction (FOP) system in the province of Ontario, Canada. A suite of supervised statistical learning models was developed using more than 50 years of high-resolution data over a 73.8 million ha study area, partitioned into Ontario’s Northwest and Northeast Fire Management Regions. A stratified modelling approach accounts for different seasonal baselines regionally and for a set of communities in the Far North. Response-dependent sampling and modelling techniques using logistic generalized additive models are used to develop a fine-scale, spatiotemporal FOP system with models that include nonlinear relationships with key predictors. These predictors include inter- and intra-annual temporal trends, spatial trends, ecological variables, fuel moisture measures, human land-use characteristics, and a novel measure of human activity. The system produces fine-scale, spatially explicit maps of daily probabilistic human-caused FOP based on locally observed conditions along with point and interval predictions for the expected number of fires in each region. A simulation-based approach for generating the prediction intervals is described. Daily predictions were made available to fire management practitioners through a custom dashboard and integrated into daily regional planning to support detection and fire suppression preparedness needs.

Fire danger rating has become the cornerstone of national fire management programs, and operational systems have been available for over 40 years. Fire danger information is used across a broad spectrum of fire management decision making including daily operations, seasonal strategic planning, and long-term fire and land management planning under future climate change. There are many different national fire danger rating systems in use worldwide. Early warning of extreme fire danger is critical for fire managers to mitigate or prevent wildfire disaster. Early warning is provided using forecasted fire weather, which is further enhanced with remotely sensed fire activity and fuels information in fire early warning systems. Fire danger and early warning systems can operate at global to local levels, depending on fire management requirements. Current operational systems and applications are reviewed.

Field experiments are one way to develop or validate wildland fire-behavior models. It is important to consider the implications of assumptions relating to the locality of measurements with respect to the fire, the temporal frequency of the measured data, and the changes to local winds that might be caused by the experimental configuration. Twenty FIRETEC simulations of International Crown Fire Modeling Experiment (ICFME) plot 1 and plot 6 fires were performed using horizontally homogenized fuels. These simulations enable exploration of the sensitivity of model results to specific aspects of the interpretation and use of the locally measured wind data from this experiment. By shifting ignition times with respect to dynamic measured tower wind data by up to 2 min, FIRETEC simulations are used to examine possible ramifications of treating the measured tower winds as if they were precisely the same as those present at the location of the fire, as well as possible implications of temporal averaging of winds or undersampling. Model results suggest that careful consideration should be paid to the relative time scales of the wind fluctuations, duration of the fires, and data collection rates when using experimentally derived winds as inputs for fire models.

Using anomalies calculated from General Circulation Model (GCM) climate predictions we developed scenarios of future fire weather, fuel moisture and fire occurrence and used these as the inputs to a fire growth and suppression simulation model for the province of Ontario, Canada. The goal of this study was to combine GCM predictions with the fire growth and suppression model to examine potential changes in area burned in Ontario due to climate change, while accounting for the large fire suppression activities of the Ontario Ministry of Natural Resources (OMNR). Results indicate a doubling of area burned in the Intensive and Measured fire management zones of Ontario by the decade of 2040 and an eightfold increase in area burned by the end of the 21st century in the Intergovernmental Panel on Climate Change Special Report on Emissions Scenarios (IPCC SRES) A2 scenario; smaller increases were found for the A1b and B1 scenarios. These changes are driven by increased fire weather conducive to large fire growth, and increases in the number of fires escaping initial attack: for the Canadian GCM's business-as-usual (A2) scenario, escaped fire frequency increased by 34% by 2040 and 92% by the end of the 21st century. Incorporating more detail on large fire growth than previous studies, our model predicts higher area burned under climate change than do these previous studies, as large numbers of high-intensity fires overwhelm suppression capacity.

For Canada's boreal forest region, the accurate modelling of the timing of the appearance of aspen leaves is important to forest fire management, as it signifies the end of the spring fire season that occurs after snowmelt. This article compares two methods, a midpoint rule and a conditional expectation method used to estimate the true flush date for interval-censored data from a large set of fire-weather stations in Alberta, Canada. The conditional expectation method uses the interval censored kernel density estimator of Braun et al. (2005). The methods are compared via simulation, where true flush dates were generated from a normal distribution and then converted into intervals by adding and subtracting exponential random variables. The simulation parameters were estimated from the data set and several scenarios were considered. The study reveals that the conditional expectation method is never worse than the midpoint method, and that there is a significant advantage to this method when the intervals are large. An illustration of the methodology applied to the Alberta data set is also provided.

Canadian forest fire management agencies use the Canadian Fire Weather Index (FWI) system (Van Wagner, 1987) to estimate fuel moisture and generate a series of relative fire behaviour indices based on weather observations. The current day’s indices and future values generated from forecasted weather aid in deploying suppression resources to areas of potentially high fire activity. FWI system indices are relative indicators of potential fire behaviour, however and do not, by themselves, predict differences in expected behaviour between substantially different forest types. While over the years operational personnel have developed experience in understanding the fuel-type specific fire behaviour expected under given values of the FWI system indices, the Canadian Fire Behaviour Prediction (FBP) system (Forestry Canada 1992) is often used to predict fire behaviour (fuel consumption, rate of spread and head fire intensity (HFI)) for a range of fuel types found in the Canadian forest.

After a decade of speculation and debate, there is now a general scientific consensus that rising greenhouse gas levels in the earth's atmosphere will result in significant climate change over the next century. The recent statement by the Intergovernmental Panel on Climate Change (Watson et al. 1995) that "the observed increase in global mean temperature over the last century (0.3–0.6°C) is unlikely to be entirely due to natural causes, and that a pattern of climate response to human activities is identifiable in the climatological record" is a strong endorsement of this conclusion. The recently negotiated Kyoto Protocol to the United Nations Framework Convention on Climate Change recognizes the influence of greenhouse gas concentrations on global warming and requires signatory countries to commit to significant reductions in emissions in the near future, further evidence of a growing acknowledgment that climate change is a reality.

The predicted increase in climate warming will have profound impacts on forest ecosystems and landscapes in Canada because of increased temperature, and altered disturbance regimes. Climate change is predicted to be variable within Canada, and to cause considerable weather variability among years. Under a 2 × CO 2 scenario, fire weather index (FWI) is predicted to rise over much of Ontario by 1.5 to 2 times. FWI may actually fall slightly, compared to current values, in central eastern Ontario (Abitibi), but for central-south Ontario it is expected to rise sharply by as much as 5 times current values. We predict that the combination of temperature rise and greater than average fire occurrence will result in a shrinkage of area covered by boreal forest towards the north and east; that some form of Great Lakes forest type will occupy most of central Ontario following the 5 C isotherm north; that pyrophilic species will become most common, especially jack pine and aspen; that patch sizes will initially decrease then expand resulting in considerable homogenization of forest landscapes; that there will be little 'old-growth' forest; and that landscape disequilibrium will be enhanced. If climate change occurs as rapidly as is predicted, then some species particularly those with heavy seeds may not be able to respond to the rapid changes and local extinctions are expected. Anthropogenically-altered species compositions in current forests, coupled with fire suppression over the past 50 years, may lead to forest landscapes that are different then were seen in the Holocene period, as described by paleoecological reconstructions. In particular, forests dominated by white pine in the south and black spruce in the middle north may not be common. Wildlife species that respond at the landscape level, i.e., those with body sizes >1 kg, will be most affected by changes in landscape structure. In particular we expect moose and caribou populations to decline significantly, while white-tailed deer will likely become abundant across Ontario and Quebec.

pcraturc and moisture changes and to disperse, but they have ipored the effects of disturbances caused by cliniate change (e.g., Ojinia et al. 1991). Yet niodeling studies indicate the ini-portance of climate effects on disturbance regimes (He et al. 1999). ImA, regional, and global changes in teniperature and precipitation can influence the OCCLII-i-cncc', timing, frc-quency, duration, extent, and intensity ofdisturba~ices (Raker 1995, Turner et 31. 19%). Because trees can survive from decades to centuries and take years to bcconic established, climate-change inipacts are expressed in forests, in part, through alterations in disturbance regimes (Franklin et al. 1992, lhle et al. 2000). Disturbances, both hLunan-induced and natural, shape fol-est systems by influencing their composition, structure, and functional processes. Indeed, the forests of the United States are molded by their la~dusc and disturbance history. Within the United States, natural disturbances having the greatest effects on forests in&de fire, drought, introduced species, insect and pdhogen outbreaks, hurric:uies, windstorms, ice storms, :tnd landslides (Fipre 1). Each disturbance :tffects forests tlifferently. Sonic ca~lse large-scale tree inortality, whcre;is others affect coniinunity structure and orgnnization CLIMATE CHANGE CAN AFFECT FORESTS LANDSLIDES without causing massive mortality (e.g., ground fires). Forest disturbances influence how much carbon is stored in trees or dead wood. All these natural disturbances interact with hunian-induced effects on the environment, such as ail pollution and land-use change resulting from resource extraction , agriculture, urban and suburban expansion, and recreation. Sonic disturbances can be functions of both natural and human conditions (e.g., forest fire ignition and spread) (Figure 2). and Neilson are committee members of the Forest Sector of the National Assessiiient on Climate Change W/IO are focusing on disturbances. The following scientists who have special knowledge of tllese disturbances assisted: Ayres (insects and patliogens); Flannigan, Stocks, and Wotton (fires); Hanson (drought); lrland (ice storms); Lugo (hurricanes); Peterson (windstorms); Sitmberloff (introduced species): and Swanson (landslides). The submitted manuscript has been authored by a coi?tractor of the US govern,i?ent under contract IIO. DE-AC05 96OR22464. Accordingly, the US government retains a nonexclusive, royalty-free license to publish or reproduce tile published form of this col?tribution, or allow others to do so, for US government purposes.