Sarah A. Lewis

Remote sensing techniques to assess active fire characteristics and post-fire effects

Space and airborne sensors have been used to map area burned, assess characteristics of active fires, and characterize post-fire ecological effects. Confusion about fire intensity, fire severity, burn severity, and related terms can result in the potential misuse of the inferred information by land managers and remote sensing practitioners who require unambiguous remote sensing products for fire management. The objective of the present paper is to provide a comprehensive review of current and potential remote sensing methods used to assess fire behavior and effects and ecological responses to fire. We clarify the terminology to facilitate development and interpretation of comprehensible and defensible remote sensing products, present the potential and limitations of a variety of approaches for remotely measuring active fires and their post-fire ecological effects, and discuss challenges and future directions of fire-related remote sensing research.

Remote sensing for prediction of 1-year post-fire ecosystem condition

Appropriate use of satellite data in predicting >1 year post-fire effects requires remote measurement of sur- face properties that can be mechanistically related to ground measures of post-fire condition. The present study of burned ponderosa pine (Pinus ponderosa) forests in the Black Hills of South Dakota evaluates whether immediate fractional cover estimates of char, green vegetation and brown (non-photosynthetic) vegetation within a pixel are improved predictors of 1-year post-fire field measures, when compared with single-date and differenced Normalized Burn Ratio (NBR and dNBR) indices. The modeled estimate of immediate char fraction either equaled or outperformed all other immediate metrics in predicting 1-year post-fire effects. Brown cover fraction was a poor predictor of all effects (r 2 < 0.30), and each remote measure produced only poor predictions of crown scorch (r 2 < 0.20). Application of dNBR (1 year post) provided a consid- erable increase in regression performance for predicting tree survival. Immediate post-fire NBR or dNBR produced only marginal differences in predictions of all the 1-year post-fire effects, perhaps limiting the need for prefire imagery.Although further research is clearly warranted to evaluate fire effects data available 2-20 years after fire, char and green vegetation fractions may be viable alternatives to dNBR and similar indices to predict longer-term post-fire ecological effects.

Review of fuel treatment effectiveness in forests and rangelands and a case study from the 2007 megafires in central, Idaho, USA

This report provides managers with the current state of knowledge regarding the effectiveness of fuel treatments for mitigating severe wildfire effects. A literature review examines the effectiveness of fuel treatments that had been previously applied and were subsequently burned through by wildfire in forests and rangelands. A case study focuses on WUI fuel treatments that were burned in the 2007 East Zone and Cascade megafires in central Idaho. Both the literature review and case study results support a manager consensus that forest thinning followed by some form of slash removal is most effective for reducing subsequent wildfire severity.

Using hyperspectral imagery to estimate forest floor consumption from wildfire in boreal forests of Alaska, USA

Wildfire is a major forest disturbance in interior Alaska that can both directly and indirectly alter ecological processes. We used a combination of pre- and post-fire forest floor depths and post-fire ground cover assessments measured in the field, and high-resolution airborne hyperspectral imagery, to map forest floor conditions after the 2004 Taylor Complex in Alaskas boreal forest. We applied a linear spectral unmixing model with five endmembers representing green moss, non-photosynthetic moss, charred moss, ash and soil to reflectance data to produce fractional cover maps. Our study sites spanned low to moderately high burn severity, and both black and white spruce forest types; high cover of green or non-photosyntheticmoss in the remotely sensed imagery indicated low consumption,whereas high coverofcharredmoss,ashorsoilindicatedhigherconsumption.Strongrelationships(R 2 ¼0.5to0.6)betweengreenmoss estimated from the imagery and bothpost-fire depth andpercentage consumption suggest that potentialburn severity may be predicted by a map of green (live) moss. Given that the depth of the post-fire forest floor is ecologically significant, the methodofmappingtheconditionoftheorganicforestfloorwithhyperspectralimagerypresentedheremaybeausefultool to assess the effect of future fires in the boreal region.

A synthesis of post-fire Burned Area Reports from 1972 to 2009 for western US Forest Service lands: trends in wildfire characteristics and post-fire stabilisation treatments and expenditures

Over 1200 post-fire assessment and treatment implementation reports from four decades (1970s–2000s) of western US forest fires have been examined to identify decadal patterns in fire characteristics and the justifications and expenditures for the post-fire treatments. The main trends found were: (1) the area burned by wildfire increased over time and the rate of increase accelerated after 1990; (2) the proportions of burned area assessed as low, moderate and high burn severity likely have remained fairly constant over time, but the use of satellite imagery that began c. 2000 increased the resolution of burn severity assessments leading to an apparent decreased proportion of high burn severity during the 2000s; (3) treatment justifications reflected regional concerns (e.g. soil productivity in areas of timber harvest) and generally reflected increased human encroachment in the wildland–urban interface; (4) modifications to roads were the most frequently recommended post-fire treatment type; (5) seeding was the most frequently used land treatment, but declined in use over time; (6) use of post-fire agricultural straw mulch has steadily increased because of proven success; and (7) the greatest post-fire expenditures have been for land treatments applied over large areas to protect important resources (e.g. municipal water sources).

Validation of a probabilistic post-fire erosion model

Post-fire increases of runoff and erosion often occur and land managers need tools to be able to project the increased risk. The Erosion Risk Management Tool (ERMiT) uses the Water Erosion Prediction Project (WEPP) model as the underlying processor. ERMiT predicts the probability of a given amount of hillslope sediment delivery from a single rainfall or snowmelt event on unburned, burned and recovering forest, range and chaparral hillslopes and the effectiveness of selected mitigation treatments. Eight published field study sites were used to compare ERMiT predictions with observed sediment deliveries. Most sites experienced only a few rainfall events that produced runoff and sediment (1.3–9.2%) except for a California site with a Mediterranean climate (45.6%). When sediment delivery occurred, pooled Spearman rank correlations indicated significant correlations between the observed sediment delivery and those predicted by ERMiT. Correlations were ρ = 0.65 for the controls, ρ = 0.59 for the log erosion barriers and ρ = 0.27 (not significant) for the mulch treatments. Half of the individual sites also had significant correlations, as did 6 of 7 compared post-fire years. These model validation results suggest reasonable estimates of probabilistic post-fire hillslope sediment delivery when compared with observations from eight field sites.

Value and challenges of conducting rapid response research on wildland fires

Rapid Response Research is conducted during and immediately after wildland fires, in coordination with fire management teams, in order to collect information that can best be garnered in situ and in real-time. This information often includes fire behavior and fire effects data, which can be used to generate practical tools such as predictive fire models for managers. Drawing upon lessons learned from fire managers and researchers working on active wildland fires, we identify challenges including high costs, logistics, and safety; understanding and fitting into the fire management organization; building relationships with managers and other researchers; and science delivery. Our recommendations for safer and more effective Rapid Response Research are that researchers must understand the fire organizations and their objectives because a fire manager’s primary responsibility is to manage the fire safely, not support research. In addition, researchers must be prepared with equipment, a “red card” signifying sufficient training and fitness, and appropriate knowledge when arriving to do research on a fire. Further, researchers must have and follow an operations plan. We recommend using a liaison to build strong relationships with managers and sharing what was learned. Science guided by questions that are important to managers is essential to improving both the understanding of wildland fire dynamics and developing strategies to address fire risk, rehabilitation, and restoration, yet researchers must be aware of the challenges of conducting research on active wildland fires.

Indicators of burn severity at extended temporal scales: a decade of ecosystem response in mixed-conifer forests of western Montana

We collected field and remotely sensed data spanning 10 years after three 2003 Montana wildfires to monitor ecological change across multiple temporal and spatial scales. Multiple endmember spectral mixture analysis was used to create post-fire maps of: char, soil, green (GV) and non-photosynthetic (NPV) vegetation from high-resolution 2003 hyperspectral (HS) and 2007 QuickBird (QB) imagery, and from Landsat 5 and 8 imagery collected on anniversary dates in 2002, 2003 (post fire), 2004, 2007 and 2013. Initial estimates of char and NPV from the HS images were significantly correlated with their ground-measured counterparts (ρ=0.60 (P=0.03) and 0.68 (P=0.01) respectively), whereas HS GV and Landsat GV were correlated with canopy GV (ρ=0.75 and 0.70 (P=0.003) respectively). HS imagery had stronger direct correlations with all classes of fine-scale ground data than Landsat and also had stronger predictive correlations with 10-year canopy data (ρ=0.65 (P=0.02) to 0.84 (P=0.0003)). There was less than 5% understorey GV cover on the sites initially, but by 2013, it had increased to nearly 60% regardless of initial condition. The data suggest it took twice as long for understorey GV and NPV to replace char and soil as primary ground cover components on the high-burn-severity sites compared with other sites.

Utility of Remotely Sensed Imagery for Assessing the Impact of Salvage Logging after Forest Fires

Remotely sensed imagery provides a useful tool for land managers to assess the extent and severity of post-wildfire salvage logging disturbance. This investigation uses high resolution QuickBird and National Agricultural Imagery Program (NAIP) imagery to map soil exposure after ground-based salvage operations. Three wildfires with varying post-fire salvage activities and variable ground truth data were used to evaluate the utility of remotely sensed imagery for disturbance classification. The Red Eagle Fire in northwestern Montana had intensive ground truthing with GPS-equipment logging equipment to map their travel paths, the Tripod Fire in north central Washington had ground truthed disturbance transects, and the School Fire in southeastern Washington had no salvage-specific ground truthing but pre-and post-salvage images were available. Spectral mixture analysis (SMA) and principle component analysis (PCA) were used to evaluate the imagery. Our results showed that soil exposure (disturbance) was measureable when pre-and post-salvage QuickBird images were compared at one site. At two of the sites, only post-salvage imagery was available, and the soil exposure correlated well to salvage logging equipment disturbance at one site. When ground disturbance transects were compared to NAIP imagery two years after the salvage operation, it was difficult to identify disturbance due to vegetation regrowth. These results indicate that soil exposure

Using QuickBird imagery to detect cover and spread of post-fire straw mulch after the 2006 Tripod Fire, Washington, USA

Agricultural straw mulch is a commonly applied treatment for protecting resources at risk from runoff and erosion events after wildfires. High-resolution QuickBird satellite imagery was acquired after straw mulch was applied on the 2006 Tripod Fire in Washington. We tested whether the imagery was suitable for remotely assessing the areal coverage of the straw mulch treatment. Straw mulch was easily identified in the imagery because of the distinct spectral signature of the mulch against the burned background. The measured straw cover on the ground was correlated to the modeled cover in the imagery with a correlation coefficient of r = 0.47, and a rank analysis indicated the ability to predict relative straw cover amounts on the plots with a rank correlation of ? = 0.40. Better correlations may be possible if the time between the image acquisition and field validation was shorter. Our results encourage further exploration of the use of high-resolution imagery for research applications and postfire management.