<%BANNER%>

UFIR



Modeling Relationships among 217 Fires Using Remote Sensing of Burn Severity in Southern Pine Forests
http://www.mdpi.com/ ( Publisher's URL )
CITATION PDF VIEWER
Full Citation
STANDARD VIEW MARC VIEW
Permanent Link: http://ufdc.ufl.edu/IR00001305/00001
 Material Information
Title: Modeling Relationships among 217 Fires Using Remote Sensing of Burn Severity in Southern Pine Forests
Series Title: Malone, S.L., Kobziar, L.N., Staudhammer, C.L., and Abd-Elrahman, A. 2011. Modeling relationships among 217 fires using remote sensing of burn severity in southern pine forests. Remote Sensing 3: 2005-2028.
Physical Description: Journal Article
Creator: Kobziar, Leda
Malone, Sparkle L.
Abd-Elrahman, Amr
Staudhammer, Christina L.
Publisher: MDPI
Publication Date: Sept. 7, 2011
 Subjects
Subjects / Keywords: burn severity
remote sensing
differenced normalized burn ratios
fire frequency
pine flatwoods forest
fire model
wildfire
prescribed fire
Genre:
Spatial Coverage:
 Notes
Abstract: Abstract: Pine flatwoods forests in the southeastern US have experienced severe wildfires over the past few decades, often attributed to fuel load build-up. These forest communities are fire dependent and require regular burning for ecosystem maintenance and health. Although prescribed fire has been used to reduce wildfire risk and maintain ecosystem integrity, managers are still working to reintroduce fire to long unburned areas. Common perception holds that reintroduction of fire in long unburned forests will produce severe fire effects, resulting in a reluctance to prescribe fire without first using expensive mechanical fuels reduction techniques. To inform prioritization and timing of future fire use, we apply remote sensing analysis to examine the set of conditions most likely to result in high burn severity effects, in relation to vegetation, years since the previous fire, and historical fire frequency. We analyze Landsat imagery-based differenced Normalized Burn Ratios (dNBR) to model the relationships between previous and future burn severity to better predict areas of potential high severity. Our results show that remote sensing techniques are useful for modeling the relationship between elevated risk of high burn severity and the amount of time between fires, the type of fire (wildfire or prescribed burn), and the historical frequency of fires in pine flatwoods forests.
Acquisition: Collected for University of Florida's Institutional Repository by the UFIR Self-Submittal tool. Submitted by Leda Kobziar.
Publication Status: Published
 Record Information
Source Institution: University of Florida Institutional Repository
Holding Location: University of Florida
Rights Management: All rights reserved by the submitter.
Resource Identifier: doi - 10.3390/rs3092005
System ID: IR00001305:00001

Downloads

This item is only available as the following downloads:

Malone_et_al_remotesensing-03-020051 ( PDF )


Full Text

PAGE 1

Remote Sens. 2011 3 2005-2028; doi:10.3390/rs3092005 Remote Sensing ISSN 2072-4292 www.mdpi.com/journal/remotesensing Article Modeling Relationships among 217 Fires Using Remote Sensing of Burn Severity in Southern Pine Forests Sparkle L. Malone 1, Leda N. Kobziar 1,*, Christina L. Staudhammer 1,2 and Amr Abd-Elrahman 3 1 School of Forest Resources and Conservation, Univer sity of Florida, Newins-Ziegler Hall, P.O. Box 110410, Gainesville, FL 32611, USA; E-Mail: slmalone@crimson.ua.edu 2 Department of Biological Sciences, Universi ty of Alabama, 407 Biology Bldg, P.O. Box 870344, Tuscaloosa, AL 35487, USA; E-Mail: cstaudhammer@ua.edu 3 School of Forest Resources and Conservation, Univer sity of Florida, 1200 N. Park Road, Plant City, FL 33563, USA; E-Mail: aamr@ufl.edu Author to whom correspondence should be addressed; E-Mail: lkobziar@ufl.edu; Tel.: +1-786-489-1090; Fax: +1-250-348-1786. Received: 20 July 2011; in revised form: 20 August 2011 / Accepted: 29 August 2011 / Published: 7 September 2011 Abstract: Pine flatwoods forests in the southeaste rn US have experienced severe wildfires over the past few decades, often attributed to fuel load build-up. These forest communities are fire dependent and require regular bur ning for ecosystem maintenance and health. Although prescribed fire has been used to re duce wildfire risk and maintain ecosystem integrity, managers are still working to re introduce fire to long unburned areas. Common perception holds that reintr oduction of fire in long unburne d forests will produce severe fire effects, resulting in a reluctance to prescribe fire without first using expensive mechanical fuels reduction techniques. To info rm prioritization and timing of future fire use, we apply remote sensing analysis to exam ine the set of conditions most likely to result in high burn severity effects, in relation to vegetation, year s since the previous fire, and historical fire freque ncy. We analyze Landsat imagery-ba sed differenced Normalized Burn Ratios (dNBR) to model the relationships betw een previous and future burn severity to better predict areas of potential high severi ty. Our results show that remote sensing techniques are useful for modeling the relatio nship between elevated risk of high burn severity and the amount of time between fires, th e type of fire (wildfire or prescribed burn), and the historical frequency of fi res in pine flatwoods forests. OPEN ACCESS

PAGE 2

Remote Sens. 2011 3 2006 Keywords: burn severity; remote sensing; difference d normalized burn ratios; fire frequency; pine flatwoods forest; fire model; wildfire; prescribed fire 1. Introduction In forests characterized by a historically frequent fire return interval, prescribed fire is often used as a tool to mimic the effects of natural fire The absence of fire in such forests would cause significant changes in vegetative species structure and composition, and could increase the threat of large-scale wildfires. In pine fl atwoods forests of the southern US, prescribed burns reduce fuel accumulations to minimize damage from potential wildfires [1,2], improve wildlife habitat, and conserve biodiversity [3-6]. However, implementing prescribed burns is increasingly difficult due to concerns related to the wild land urban interface (WUI). In particular, fire management decision-making in Florida has been shown to be dictated by urban en croachment, forest fragmentation, and the challenges associated with smoke management [7]. In WUI areas fire behavior must be carefully controlled to prevent escapes. Managers strive to implement burns where fuel and w eather conditions will minimize the potential for the high-severity fires that creat e challenges for smoke management and post-fire ecosystem recovery. Fire severity is a measure of ecological and physical ch ange attributable to fire [8,9], and is dictated by the intersection of fuels and weather conditions In addition to being associated with smoke production, severity is an important post-fire metric used to explai n fire effects on exotic species establishment, soil responses, and fore st recovery. To describe fire effects in the southeastern US, burn severity is classified in four categories: unburned, low, medium and high severity [10-13]. Low severity burns are characterized by lightly burned areas where only fine fuels are consumed with minor scorching of trees in the understory [14]. Areas of moderate severity retain some fuels on the forest floor and have crown scorching in mid-large trees with mortality of small trees [14]. High severity zones generally experience complete combustion of most of the litter layer, duff and small logs, mortality of small to medium trees, and consumption of large tree crowns [14]. Assessing burn severity across a frequently burn ed landscape can provide important information about both the immediate and longer-term consequences of fire use and manageme nt, as the severity of one fire likely influences the severity of the subse quent fire. The timespan betw een fire events can also have a significant effect on subsequent fire behavior and fire effects [1,2]. However, few studies have addressed the relationship between fire frequency, burn severity, and subsequent fire patterns in Florida’s fire-prone forests. Out calt and Wade [2] found a significant relationship between the amount of time since last fire and tree mortality following a wildfire in Florida pi ne flatwoods [2]; as time increased to two or more years, mortality also incr eased [2]. Also in pine flatwoods, Davis and Cooper [1] showed that fuel accumulations of three years or less supported fe wer fires, lower fire intensities, and lower burned acreage. These studies suggest that one fire affects the next, but do not address patterns across landscapes or among fires of differing severi ty, and are limited in temporal scope. Combined with existing information about fire lo cations and perimeters, bur n severity histories can be mapped to monitor trends in fire effects over time, in relation to frequency of fire, and as a function

PAGE 3

Remote Sens. 2011 3 2007 of time since last fire. Such data can then be used to make infere nces about future fires. Remote sensing techniques are often utilized to monitor ch anges in fire regimes over time and to map burn severity [11,15,16], but the techniqu e has been under-uti lized for burn severity analysis in southern forests [15]. Normalized burn ratios (NBR) use the difference between Landsat Thematic Mapper (TM) near and mid-infrared band reflectance values to quantify the severity level of a burned area [14]. The difference between bands 4 and 7 reflect ance values can be attributed to fire induced changes in soil moisture, canopy cover, biomass, ch arring, and exposed soil. Difference normalized burn ratios (dNBR) capture fire effects by differenc ing preand post-fire NBRs from ETM/TM images directly before and soon after a fire. Changes in green reflectance va lues are captured by band 4; while increases in charred fuels, exposed soil, and decr eases in vegetation density cause an increase in band 7. The dNBR technique is effective in repres enting burn severity because it captures relative changes in the preand post-fire normalized bur n ratio. Employed as a radiometric index, dNBRs can be directly related to burn severity [14,16] and, as l ong as the fire is within the resolution range of the satellite sensor (30 m), fires and their associat ed burn severity are of ten detectable [17,18]. Previous studies in other regions have used dNBRs to calibrate severity levels to specific forest types [10,15,16,19], compare severity leve ls between fire events [12,19], interpret the effects of fuel management techniques on severity levels [20], and to monitor changes in vegetation over time [17,21] and across topography [11,22]. The multi agency project, Monitoring Trends in Burn Severity (MTBS) is currently using dNBRs to map burn severity and th e perimeters of wildfires greater than 405 ha in the western US and 202 ha in th e eastern US [23]. The MTBS doe s not, however, compare a given fire’s severity to the severity of subsequent fires, and given that most fires in southern states are less than 202 ha in size, many fires are overlooked. Here, we analyze a decade of wildfire and prescr ibed burn severities in Florida flatwoods pine forests using dNBR, and examine whether landscape, vegetation, soils, and fire history, and fire frequency influence burn severity pa tterns. We then use these relationships to derive predictive models for high severity prescribed burns and wildfires. We hypothesize that time-sin ce-fire will influence the probability of high burn severity in pine flatwoods only so long as the vegetation recovery does not result in altered species composition and structure. Severity should increase as time and fuel loads increase up to that threshold. We also hypot hesize that mesic communities will have a higher probability of high burn severity than hydric comm unities during prescribed burns, while the opposite could be observed for wildfires. Prescribed burns are most fre quently administered when hydric communities (e.g., cypress domes) are moist, as these communities are fire-sensitive. Because of this, prescribed burns only partially consume these unders tory fuels, with little consumption of the duff layer [2] resulting in overall low severity. Conseque ntly, hydric areas accumulate heavy fuel loads that can only support combustion during extended drought periods, and most of ten burn under weather conditions conducive to large, highintensity wildfires [1]. Research has shown that the number and size of wildfires in Florida are positively correlated with drou ght conditions [24]. Optimally, prescribed burns in these forests are administered for minimal overstory mortality and partial consumption of understory surface fuels, with little consumption of the ground fuels including the duff layer. Predicting the probab ility of high severity wildfire ri sk would benefit land managers in their wildfire mitigation planning and tactics. Th ere exist few locations worldwide where multiple prescribed burns and wildfires occur and overlap from one year to the next. The convergence of

PAGE 4

Remote Sens. 2011 3 2008 frequent prescribed fire and fast vegetation recovery rates sets the stage for this unique opportunity. Understanding how and why one fire affects a subs equent fire can lend insight into the complex relationships between fire behavi or, fuels and vegetation recovery, and burn severity over time, and expand the utility of remotely sens ed imagery for ecological research. 2. Methods 2.1. Study Site The Osceola National forest is located in no rth central Florida (Lat itude: 30.34371, Longitude: 82.47322) about 40 miles west of the city of Jacks onville (Figure 1). The r oughly 93,000 ha forest is largely dominated by pine flatwoods with scatte red areas of cypress a nd bay swamps. With an overstory of pines on flat, sandy, acidic soils, pine flatwoods have an understory dominated by shrubs, interspersed with herbaceous plants and grasses. Flatwood communities are fire dependent and require regular burning to perpetuate pyrogenic species and ecosystem health. Forest communities include longleaf ( Pinus palustris Mill.) pine–wiregrass ( Aristida stricta Michx ), and slash pine ( Pinus elliotti Englem )–gallberry ( Illex glabra (L.) Gray.)–saw palmetto ( Serenoa repens (Bartram) Small.). Cypress ponds ( Taxodium distichum (L.) Rich ) are found scattered throughout the forest in low lying wet areas. In this fire maintained comm unity the lack of fire for prolonged periods allows broadleaf trees to increase their status and emerge into the midand overstory canopies, and reduces herbaceous plant cover and eventually pine regeneration. Figure 1. Location of the Osceola National Fore st and dominant overstory vegetation.

PAGE 5

Remote Sens. 2011 3 2009 Fire management and use in the Osceola National Fore st is quite active, with an annual prescribed burning quota of over 13,000 ha. The majority of the fo rest is prescribed burned at a frequency of every 2–5 years. There are also sensitive areas within the forest that are not currently and actively managed by fire. Fire prescriptions are determined on a compartment level based on the current forest type and the desired future condition of the compar tment. Osceola fire managers are challenged to meet federal quotas for prescribed burning each y ear, which require burning large areas within the handful of days that meet prescribed fire weathe r conditions. Sensitive areas near the forest like Lake City Municipal Airport, US Interstate Highway 10, and the nearby City of Jacksonville, provide additional challenges for fire managers (Figure 1). 2.2. Image Analysis NBR raster GIS layers extracted from geometrica lly and radiometrically corrected Landsat 7 ETM and Landsat 5 TM imagery were downloaded thr ough the United States Geological Survey (USGS) global visualization website [25]. Pre-processing steps for all imag es were conducted by the USGS EROS Data Center [23,26,27]. Scenes cl osest to the date of the fire (w ithin the same season) were used for pre-fire images. Post-fire images were closest to the one-year anniversary of the fire (within the same season of the fire). The NBRs (Equation (1)) we re computed as a normalized difference of the Landsat (ETM/TM) reflectance values from band 4 ( B4; near-infrared band) and 7 ( B7; mid-infrared band). The dNBR was computed using preand post-fire event NBR to quantify burn severity using Equation (2): (1) (2) The dNBR can range from 2,000 to 2,000 yet typically range from 500 to 1,200 [28]. Negative values represent vegetation regrowth whereas positive values represent burn severity. Values close to zero are unburned areas and burned areas are gene rally greater than 100. The threshold between burned and unburned areas can rang e anywhere from 80 to 100 [27] depending on forest type. 2.3. Severity Classification We used the 101 prescribed burns and 116 documented wildfires grea ter than 0.4 ha in size that occurred during the period from 19 98 to 2008 to create a complete fi re history for the entire forest using dNBRs for each fire event. General severity levels provided by the United States Geological Survey (USGS) [29] were reclassified into four severity levels to co mbine classes that result in similar fire effects; unburned (dNBR 100 to 99), low severity (dNBR 500 to 101, 100 to 269), moderate severity (dNBR 270 to 439), and high severity (d NBR 440–1,300) (Table 1). Burn severity level thresholds are within the thresholds for pine fl atwoods and depression swamps developed by Picotte and Robertson 2011 [19]. Picotte and Robertson us ed 731 ground-level CBI plot s for pine flatwoods forest and depression swamps on the Apalachicola National Forest, th e Okefenokee National Wildlife Refuge, and the Osceola National Forest to deve lop burn severity thresholds. Unlike Picotte and Robertson 2011, we combined the re-growth burn severity class (dNBR 500 to 101) with the low

PAGE 6

Remote Sens. 2011 3 2010 severity class to distinguish burne d from unburned areas in the model (T able 1). We also used a single set of severity thresholds for all forest types. A re lativized version of the dNBR is often used to adjust the dNBR value according to the pre-fire NBR to co rrect for the variance in the reflectance of the initial plant cover types. Howeve r, we did not use this method. Picotte and Robertson 2011 found no added advantage with using the RdNBR method over th e dNBR approach in pine flatwoods forest. To account for temporal variation in phenology and su rface moisture conditions in the preand post-fire images, the mean value of unchanged forested pixe ls outside the burned ar eas were subtracted from the dNBR [12,28]. DNBRs were then clipped using fire perimeter shape files provided by the USDA Forest Service Osceola National Forest. Ne xt, fires were merged to create a map that represented fire events for each year. Table 1. Description of burn severity thresholds us ed to define severity classes (unburned, low, moderate, and high). USGS Severity Class Reclassified Severity Class DNBR range USGS Description Description 100 to +99 Unburned within a fire perimeter Unburned 500 to 101 Re-growth Low Severity +100 to +269 Low Severity +270 to +439 Low-Moderate Severity Moderate Severity +440 to +659 Moderate-High Severity High Severity +660 to +1,300 High Severity Fire events were used to analyze the relationship between severity leve l of a previous fire and changes in the severity level of the next fire by time inte rval. An additional analysis quantified the relationship between fire history metrics for the entire forest a nd the likelihood of high burn severity in a given year. The layers created for each year (see figure in Sec tion 3.6) in addition to fore st community and forest type data were used to calculate covariates for use in the statistical model. Specifically, the fire layers were used to calculate: (1) latest severity level, wh ich defines the burn severity level of the last fire event; (2) frequency, which is the nu mber of times a pixel has burned within the dataset; and (3) time since last fire, which is the number of years since last fire. Calcula tions were made using the raster calculator in the ArcGIS spa tial analyst extension software (ESRI, Redlands, CA, USA). 2.4. Model Development Models were created to investigate the effects of fire history and environmental conditions on the probability of subsequent fire and its severity. T hus, models examined the occurrence of subsequent fires in a particular pixel (an initial or first fire, versus a second or subsequent fire). Four logistic models were estimated to evaluate the effects of the severity of the initial fire, the Palmer Drought Severity Index (PDSI) [30], fire type (wildfire or prescribed fire), forest type, community type and time since the initial fire on the probability of: (1 ) the occurrence of subsequent high burn severity;

PAGE 7

Remote Sens. 2011 3 2011 (2) burn severity increasing from the first to the s econd fire; (3) burn severity decreasing from the first to the second fire; and (4) repeated fire (of any se verity level) during the study period (Table 2). Forests types were classified as pine, hardwood, hard wood-pine, or pine-hardwood forest types (Figure 1), and hydric or mesic community types. Forest type a nd community type were ob tained from the Florida Geographic Data Library [31]. The forest type la yer was developed by the University of Florida Geoplan Center. Vegetative commun ities were distinguished based on Davis [32]. Swamps, marshes, and other areas classified by the National Hydraulic Dataset as having standing water were classified as hydric and the rest of the fore st was classified as mesic based on soil and forest types. Average PDSIs were computed for four calendar years: the year before and the year of the first fire, and the year before and the year of the second fire. Usi ng calendar years as breakpoi nts for PDSI means that the fall/winter dry period is included in th e PDSI average for the year before a fire. Table 2. Independent variable inputs to models of subsequent fire and severity (all first-order interactions were also considered). Models 1–4 Severity Level of the In itial Fire (Low, Moderate, or High Burn Severity) Palmer Drought Severity Index ( 4 to +4) Fire Type (Wildfire or Prescribed Burn) Forest Type (Pine, Hardwood, Pine-Hardwood, or Hardwood-Pine) Community Type (Hydric or Mesic) Time since the initial Fire (1 to 10 years) Model 5 Severity Level of the In itial Fire (Low, Moderate, or High Burn Severity) Palmer Drought Severity Index ( 4 to +4) Fire Type (Wildfire or Prescribed Burn) Forest Type (Pine, Hardwood, Pine-Hardwood, or Hardwood-Pine) Community Type (Hydric or Mesic) Time since the initial Fire (1 to 10 years) Fire Frequency (0.1 to 1) We also developed a fifth model to investigate the relationship between the probability of experiencing high severity prescribed burns a nd wildfires and fire frequency in addition to the above-mentioned factors, and to use as a basis for testing model pr edictions. Fire frequency wa s computed as the total number of fires divided by the nu mber of years of the entire study period. The first four logistic models were based on data from all wildfires and prescribed burns >0.4 ha in size from 1998 to 2008 on the Osceola National Forest. All areas that were within fire perimeters were included in this analysis. Each pixel was treated as one observa tion. Only the first and second fires during the

PAGE 8

Remote Sens. 2011 3 2012 observation period were used in this analysis to represent the time between fires. Approximately 40,469 ha were included in the dataset. For the fifth model, fire histor y was developed for pixels using data up to 2005, and this data was then used to estimate the probability of exhibiti ng high burn severity eff ects in year 2006. The model was then tested against actual fire patterns for a specific year when multiple fires occurred. Model predictions were validated by compar ing actual dNBRs for prescribed fi re and wildfire that occurred in 2007. Probability of experiencing high burn severity was computed and presented spatially using the parameters estimated from the fifth logistic regr ession model. Model input s were organized into individual layers and calc ulations were made using the raster cal culator in the ArcGIS spatial analyst extension (ESRI, Redlands, CA, USA). The spatial model output showed the probability of a given area in the landscape displaying hi gh burn severity for either pr escribed burns or wildfires. 2.5. Logistic Regression Analysis Logistic regression methods were utilized to dete rmine fire probabilities as described by models 1–5 for both prescribed burns and wildfires, on a pixel level. Responses are coded as 1 or 0 for “success” or “failure”. For model 1, this correspo nds to a pixel being burned or unburned, respectively, at the high severity level: (3) where is a realization of a random variable Yi that can take on the values of 1 and 0 with probabilities i and 1 i. The variable Yi has a Bernoulli distribution with mean and variance depending on the underlying probability i such that E(Yi) = i and var(Yi) = i(1 i). The probability, i, is a linear function of a ma trix of observed covariates, xi, transformed via the logit function to remove range restrictions: (4) where is a vector of unknown parameters to be esti mated. The logit maps probabilities from the range [0, 1] to the spac e of all real numbers [ ]. Negative logits represent probabilities below 50% and positive logits represent proba bilities above 50%. Solving for the probability of success requires exponentiating the logit and cal culating the odds of success: (5) Maximum likelihood methods were used for parameter estimation. With this approach, parameters are estimated iteratively until parameters that maximize the log of the likelihood are obtained. Goodness of fit statistic s, Akaike’s information cr iteria (AIC) and Bayesian in formation criteria (BIC), were used to compare competing m odels. AIC and BIC are model selection statistics that facilitate comparisons between models with different numbers of paramete rs. Both avoid increasing the likelihood by over fitting, measuring how close fitted values are to true values, with a penalty for the number of parameters in the model [33]. The ratio of the Pearson chi-square statistic to its degrees of freedom was used to determine if the model displayed l ack of fit [29]. Values cl oser to 1 indicate that

PAGE 9

Remote Sens. 2011 3 2013 models fit the data well [33]. Since raster data is comprised of adjacent pixels, the assumption of independence among observations wa s likely violated due to spatia l autocorrelation. Therefore, a generalized linear mixed model, whic h would model the spatial correla tion directly, would be the most appropriate model. In theory, th is could be accomplished using th e SAS procedure PROC GLIMMIX, adding an appropriate spatial correlation structure. However, due to the volume of data involved, it is not possible to estimate this model without the aid of a s uper computer. Instead, random residuals were modeled to account for overdispersion [34]. A modified backward selection method was used to determine the appropriate covariates for the final model. First, all parameters were included in the model, including all first-order interactions between parameters. Then, the least significant covariates based on the Wa ld chi-square statistics were dropped one at a time until all covariates remaining in the model were significant (P < 0.05). At each step in model selection, not only was the significan ce of each model parameter evaluated, but we also ensured that the final model had the lowest AIC and BIC values. To test for differences among categorical levels, least square means were produ ced and differences were tested via the post-hoc Tukey-Kramer method. 3. Results 3.1. Predicting the Occurrence of Subsequent High Burn Severity Based on the dNBR analysis of fire events over time, severity level of the first fire, PDSI for the year of the first and second fire, fire type for th e second fire, community type, and the time interval between the first fire and the sec ond fire were all signifi cant indicators of high burn severity occurring subsequently. The overall model (P < 0.05) and the pa rameters were signifi cant based on their Wald chi-square statistics (P < 0.05). The ratio of the Pearson chi-square statistic to its degrees of freedom was approximately 1.15, indicating good model fit (Table 3(a)). As the severity level of the first fire increased, th e probability of high burn severity in subsequent fire also increased, but the influence was weak. Ar eas previously burned at a high burn severity had the highest probabilities for high burn severity in subsequent fire while areas previously burned by low burn severity had the lowest probabilities (Table 3(a)). Areas previously burned by wildfires and hydric areas had higher probabilities of high burn seve rity in subsequent fire than areas that were prescribed burned and mesic areas (T able 3(a)). In hydric areas, the pr obability of high burn severity in areas burned previously by wildfires was 9.3% higher than in areas that were prescribed burned. In mesic areas, the probability of high burn severity in areas burned previously by wildfires was 13.72% higher than in areas that were pr escribed burned. The probability of high burn severity in subsequent fire increased with the time interval between the fi rst and second fire up to five to six years, then declined with time interval between fires seven to ei ght and nine to ten years (Table 3(a), Figure 2(a)).

PAGE 10

Remote Sens. 2011 3 2014 Table 3. Parameter estimates and their respective standard errors, and p-values for the regression mode ls predicting the probability of : ( a ) high burn severity occurrence; ( b ) increasing in burn severity from one fire to the next; ( c ) decreasing in burn severity from one fire to the next; and ( d ) subsequent fire occurrence based on satellite imagery severity classi fications from 1998 to 2008. Parameter Model (a) Model (b) Model (c) Model (d) Estimate Std. Error Estimate Std. Error Estimate Std. Error Estimate Std. Error P-value Intercept 5.34 0.12 1.31 0.03 1.15 0.03 1.22 0.02 <0.0001 Fire 1 Severity Unburned 2.97 0.02 0.73 0.02 <0.0001 Low 0.35 0.03 0.75 0.02 2.89 0.02 0.57 0.02 <0.0001 Moderate 0.20 0.03 Reference 1.19 0.03 0.35 0.02 <0.0001 High Reference Reference Reference <0.0001 Fire 1 Type Wildfire 1.19 0.02 0.51 0.01 0.29 0.01 <0.0001 Prescribed burn Reference Reference Reference Fire 2 Type Wildfire 0.09 0.01 0.09 0.01 1.02 0.03 <0.0001 Prescribed burn Reference Refere nce Reference Time Interval Between Fires (Years) 1–2 1.68 0.11 1.54 0.02 0.58 0.02 0.61 0.02 <0.0001 3–4 0.45 0.12 0.76 0.02 0.66 0.02 0.79 0.02 <0.0002 5–6 4.66 0.12 0.22 0.03 0.57 0.03 1.19 0.02 <0.0001 7–8 2.05 0.12 1.53 0.02 0.72 0.03 0.17 0.02 <0.0001 9–10 Reference Reference Refere nce Reference Forest Type Hydric 0.21 0.12 <0.0001 Mesic Reference <0.0001 Palmer Drought Severity Index Average for the year before the first fire 0.65 0.01 <0.0001 Average for the year of the first fire 0.23 0.01 0.07 0.00 -0.47 0.01 <0.0001 Average for the year before the second fire 0.48 0.00 -0.25 0.01 <0.0001 Average for the year of the second fire 0.09 0.01 -0.08 0.00 -0.03 0.00 <0.0001 Interaction Between Time Interval Between Fire 1 and Fire 2 (Years) and Fire Type 2 1–2 Wildfire 1.30 1.30 <0.0001 Prescribed burn Reference 3–4 Wildfire 0.13 0.13 <0.0005 Prescribed burn Reference 5–6 Wildfire 1.95 1.95 <0.0001 Prescribed burn Reference 7–8 Wildfire 1.27 1.27 <0.0001 Prescribed burn Reference 9–10 Wildfire Reference Prescribed burn Reference Residual 1.15 0.96 0.98 1.00

PAGE 11

Remote Sens. 2011 3 2015 3.2 Probability of Burn Severity Increasing from the First to the Second fire Severity level of the first fire, average PDSI valu es, fire type (prescribed or wildfire) of both the first and the second fire, and the time interval betw een the first and second fire were all significant indicators of the probability of increasing burn seve rity. The overall model was significant (P < 0.05) and the parameters were significant based on their Wald chi-square statistic (P < 0.05). The ratio of the Pearson chi-square stat istic to its degrees of freedom wa s approximately 0.96, indicating good model fit (Table 3(b)). The severity level of the first fire and the probability of increasing subsequent burn severity were inversely related. Areas previously unburned within fire perimeters had the highest probabilities of increasing burn seve rity (73–83%) followed by areas pr eviously burned at a low burn severity level (23–35%) and at a moderate burn severity level ( 12–20%) (Table 3(b)). The PDSI value for the year immediately prior to the second fire had the greatest influence on subsequent increasing burn severity, suggesting that fall/w inter drought conditions we re positively associated with increasing burn severity during the spring/summer season (Table 3( b)). Areas that burned as wildfires in the first fire had a lower probability of in creasing subsequent burn severity versus those that burned in prescribed fire in the first fire. Areas burned by wildfires in the sec ond fire had higher probabilities of increasing burn severity. On average, areas that were first burned by wildfires had a 7% lower probability of increasing subsequent burn severity versus that of prescribed burns. When the time interval was five to six years between fires, probabil ity of increasing burn seve rity in subsequent fire was highest (ranging from 91–96%, depe nding on fire type), while the lo west probability of increasing burn severity (73–83% depending on fire type) occurred when ther e was one to two years between fires (Table 3(b); Figure 2(b)). 3.3. Probability of Burn Severity Decreasing from the First to the Second Fire The regression model for the probab ility of decreasing burn severity in a subsequent fire included severity level of the first fire, average PDSI for the year before and the ye ar of both the first and subsequent fires, fire type for both the first and second fire, and ti me interval between the first and second fire (Table 3(c)). The overall model was significant (P < 0.05) and the parameters were significant based on their Wald chisquare statistic (P < 0.05). The ra tio of the Pearson chi-square statistic to its degrees of freedom was approxima tely 0.98, indicating good model fit (Table 3(c)). The probability of decreasing in severity level in subse quent fire was higher for areas that had previously exhibited high burn severity (93–94%) than those that had burned at moderate (80–85%) and low burn severity (42–52%) (Table 3(c)). Ar eas burned by wildfires in the first fire had higher probabilities of decreasing subsequent burn severity than areas that were prescribed burned in the first fire (Table 3(c); Figure 2(c)). On average, the probability of decreasi ng subsequent burn severity in areas that burned as wildfires was 5.5% lower than that of areas that were prescribed burned. The probability of decreasing in severity level in subsequent fire was the lowest for the time interval five to six years (ranging from 19–24%). All other time intervals had sim ilar probabilities (3 0–50%) (Figure 2(c)).

PAGE 12

Remote Sens. 2011 3 2016 Figure 2. The relationship between tim e since last fire and: ( a ) the probability of high burn severity; ( b ) the probability of increasing burn severity; ( c ) the probability of decreasing burn severity; and ( d ) the probability of fire occurrence in subsequent fire by fire type. 3.4. Probability of Repeated Fires during the Study Period The following indicators were significant in pred icting the probability of a second fire occurrence during the study period:, the severity level of the firs t fire, fire type of the second fire, time interval between fires, and the interaction of time between fi res and the second fire’s type (Table 3(d)). The overall model was significant (P < 0.05) and the pa rameters were significant based on their Wald chi-square statistic (P < 0.05). The ratio of the Pearson chi-square st atistic to its degrees of freedom was approximately 1.00, indicating good model fit (Table 3(d)). As the burn seve rity level of the first fire increased, the probability of subsequent burni ng increased as well. Areas previously showing high burn severity had a 83–94% chance of burning, comp ared to areas previously showing low burn severity, which had a 70–76% probability. There wa s a higher probability of subsequent burning by wildfire than by prescribed fire (Table 3(d); aver age difference 13%, depending on previous fire type). Areas that had a time interval of five to six year s between fires had the highe st probability of burning (93–97%) (Figure 2(d)). The probability of burni ng was between 34 and 83% for all other time intervals (depending on fire type ), generally increasing with ti me (Table 3(d), Figure 2(d)). 0 20 40 60 80 100 1-23-45-67-89-10Probability of High Burn Severity (%) Wildfire (Hydric Forest) Wildfire (Mesic Forest) Prescribed Burn (Hydric Forest) Prescribed Burn (Mesic Forest) 0 20 40 60 80 100 1-23-45-67-89-10Probability of Burn Severity Increasing (%)Time interval between fires (years) Wildfire (Fire 1) Wildfire (Fire 2) Prescribed Burn (Fire 1) Prescribed Burn (Fire 2) 0 20 40 60 80 100 1-23-45-67-89-10Probability of Burn Severity Decreasing (%) Wildfire (Fire 1) Wildfire (Fire 2) Prescribed Burn (Fire 1) Prescribed Burn (Fire 2) 0 20 40 60 80 100 1-23-45-67-89-10Probability of Fire Occurrence (%)Time interval between fires (years) Wildifire Prescribed Burna.c. b.d.

PAGE 13

Remote Sens. 2011 3 2017 3.5. Temporal Thresholds and th e Importance of Fire Type The four models described above identify important thresholds for fire return intervals in pine flatwood forests. The time interval of five to six years between fires em erged as the point where previous fires had limited mitigating effects on subsequent fires. The probability of subsequent high burn severity was highest when th e time interval was five to six years between fires (ranging 64–88%) for both prescribed burns and wildfires (Figure 2(a)). This interval was also associated with the highest probability of increasing in burn severity in subseque nt fires, the lowest probability of decreasing burn severity (19–25%), and the highest probability of s ubsequent fire occurrence (93–97%) (Figure 2(b–d)). The models also identified three to four years between fires as an interval where previous fire activity reduced the probability of subseq uent high burn severity and incr eased the probability of lower subsequent burn severity, but increased the severity of subsequent fires. Areas that remained unburned greater than six years also showed a mitigating eff ect on the probability of increasing burn severity in subsequent fires. Areas that burned as wildfires had higher proba bilities for high burn severity (9% higher on average) than did areas of prescr ibed burns. Yet, areas that were first prescribed burned had higher probabilities for the second fire increasing in severi ty level (27% higher on average) than wildfire burned areas. Areas previously burned by prescribed fire also had slightly higher probabilities for increasing burn severity level in subsequent fires than areas that were burned by wildfires. The probability of decreasing burn severity was slightly lo wer for areas that had been previously prescribed burned (6% lower on average) than for areas previous ly burned by wildfire. Hydric forests had higher probabilities of high burn severity (~1% highe r on average) than did mesic forests. 3.6. Creating the Predictive Model to Us e for Testing Known Fire Patterns Based on the time series analysis of satellite im agery from 1998 to 2006, fire frequency, time since last fire, the interaction between fr equency and time since last fire, an d fire type were all significant predictors in the model for both high burn severity in prescribed burns and wildfires (Table 4). The overall model was significant (P < 0.05) and the pa rameters were significant based on their Wald chi-square statistic (P < 0.05). The ratio of the Pearson chi-square st atistic to its degrees of freedom was approximately 0.94, indi cating good model fit. The model predicts that the majority of the fo rest (96%) in 2006 had a pr obability of high burn severity less than 75%. Time sin ce last fire had a negative relations hip with the probability of high burn severity for prescribed burns and wildfires; as time since last fire increased, the probability of high severity prescribed burns and wildfires decr eased (Figure 3(a,b)). Also exhibiting a negative relationship with the probability of high burn severity, as frequency of fi re increased, the probability of experiencing a high burn severi ty decreased (F igure 3(b)).

PAGE 14

R p f p m f s W y u c p e R emote Sen s Table regres b ased Fi g ur e high b The seve p robability o f ire or co m p robability m odel data f ires had be e s econd fire’ s W e also de v y ears since t u nder any c c ommunity p arameters e xceeding o n s 2011 3 4. Param e sion model on satellite T Fire F F Fire T e 3. Relatio b urn severit y rity level o f o f subsequ e m munity ty p of high bu r that had b u e n previou s s severity. A v eloped a w t he last fir e c ombinatio n type were n resulted i n n e, indicati n e ter estimat predicting imagery s e Param e Interc e Fire Freq u T ime Since L F requency Last F i F ire Type: W T ype: Pres c Resid u nship betw e y fo r : ( a ) p r f the last fi r e nt high b u r p e in the m r n severity u rned betw e s ly identifie A large por t w eighted bu e Neither a p n of fire h n ot include d n the ratio n g overdis p es and the i the probab i e verity clas s e ter e pt u ency L ast Fire Time Sin c i re W ildfire c ribed Burn u al e en time si n r escribed b u r e, forest ty p r n severity m odel dist o We look e e en 2002 a n d as a thre s t ion of the f u rn severit y p proach yi e istory para m d in the fin of the Pe a p ersion. i r respecti v i lity of hig h s ifications fr Esti m 1.9 6 2.1 4 0.2 7 c e 0.5 9 3.3 0 s Refer e 0.93 n ce last fire u rns; and ( b p e, an d co m (P > 0.05) o rted the r e d further i n n d 2006 on s hold wher e f ull-period d y index by e lded a sig n m eters. Se v al model be a rson chi-s q v e standard h severity p fr om 1998 t o m ate Std 6 76 0. 4 24 0. 7 96 0. 0 9 07 0. 0 71 0. 0 e nce 59 frequency ) wildfire. m munity ty p Including r elationshi p n to the lac k ly, given t h e the first fi r d ataset had dividing t h n ificant mo d v erity leve l e cause incl u q uare stati s errors, an d rescribed b o 2006. Error P 3210 < 1899 < 0 2556 < 1458 < 0 7704 < of fire, an d p e were no t either the s p between k of signif i h at greater t r e had little more than f h e severity d el. Forest t l of the la s u ding any c s tic to its d d p-values urns and w P -value 0.0001 0.0001 0.0001 0.0001 0.0001 d the proba b t strong ind i s everity le v fire freque i cance by a t han five y e mitigating f ive years b level by t h t ype was n o s t fire, for e c ombinatio n d egrees of 20 1 for the w ildfires b ility of i cators of t h v el of the l a ncy and t h a ttempting t e ars betwe e effect on t h b etween fir e h e number o o t significa n e st type, a n n of the thr e freedom fa 1 8 h e a st h e t o e n h e e s. o f n t n d e e fa r

PAGE 15

R 3 2 p f c r c m f s 8 ( s g R emote Sen s 3 .7. Testing In order t 2 007 was c p atterns acr o f or prescri b c lassificatio n eflects the c ould have b m ore land w f ires, the m a s everity (0. 5 8 49 ha) wit h 30%; 375 h Fi g ur e fires i n severi t Model p r s everity ba s g reater tha n s 2011 3 the Predic t t o examine c hosen, as o ss the lan d b ed burns n of unbur n burn sever i b een misse d w as burned b a jority of t h 5 %; 8 ha). h low level s h a) burned a e 4. The o b n the Osce o t y for presc r r edictions f s ed on fire n 75% were t ive Model a the accur a both presc r d scape wer e and 5,544 n ed ha is b a i ty for the e d If the u n b y prescrib e h e burned a r The majo r s of high b u a t the mode r b served bur n o la Nationa l r ibed burns f or 2007 id e history for selected a s a gainst Kn o a cy of the p r ibed and w e heterogen e unburned a sed on th e e ntire incl u n burned hec e d burns (~ r eas had lo w r ity of land u rn severit y r ate burn s e n severity l e l Forest, F L and wildfi r e ntified are a combined s those exp o wn Fire P a p redictive m w ildfire ac t e ous with 4 hectares w e assumptio n u ded acrea g tares are s u 1,771 ha) t h w burn sev e burned b y y in wildfir e e verity leve l e vels for ( a L versus ( b ) r es. a s that had prescribed ected to b u a tterns m odel again s t ivity was 4 72 ha of u n w ithin per i n that the L g e, and a s m u btracted fr o h an by wild e rity (84%; y wildfires e s (0.14%; 2 l compared ) the actua l ) Model de r increased p burns and u rn severel y s t actual fi r widesprea d n burned ar e i meters fo r L andsat ET M m aller burn e o m the per i fires (~1,2 2 1,497 ha) w was also l o 2 ha). A la r to prescrib e l area burn e r ived proba b p robability wildfires. A y if a fire e v r e data for a d In 2007, e a within fi r r wildfires. M/TM 30 e d area wit h i mete r b ase 2 5 ha). Of t h w ith very li t o w burn se r ger portio n e d burns (1 4 e d during t h b ility of hi g of exhibiti n A reas with v ent were t 20 1 a given ye a actual bu r r e perimete r The ima g 30 m pix h in the pix d total are a h e prescrib e t tle high b u r verity (69 % n of wildfir e 4 %; 265 ha ) h e 2007 g h burn n g high b u r probabiliti e o occur. T h 1 9 a r, r n r s g e el el a s, e d r n % ; e s ) rn e s h e

PAGE 16

Remote Sens. 2011 3 2020 model predicting high burn severity showed that the majority of the forest ( 62%) had a low probability of high burn severity (<6%) for the en tire forested area burned within the perimeters of the actual fires occurring in 2007 (Figure 4). Only a small percentage of the actual bu rned area (9%) wa s predicted to burn with high burn severity, wh ile 0.4% actually displayed hi gh burn severity in 2007. The 2007 model identified areas with lower time since the last fire as targets for high burn severity (Figure 4). Of the entire area burned in 2007, the majority of hi gh burn severity areas (84 %) had a high probability of high burn severity (>75%). Only 11 ha displaye d high burn severity eff ects while 301 ha were identified as having a high probability of high burn se verity. Hectares burned at low and moderate burn severity levels (2,986 ha) were mostly identified as having a low probability of high burn severity fire (2,695 ha). The model predicted different patte rns of high burn severity for pres cribed burns and wildfires that ranged from 0 to 88% for the entire forest. Pres cribed burns had higher probabilities of high burn severity than did wildfires (Figure 5(a,b)). Probab ilities of high burn severity ranged from 0 to 88% for prescribed burns and 0 to 17% for wildfires. Areas th at burned more recently had elevated probabilities of high burn severity for both fire types. Figure 5. Model derived probability of high burn severity for ( a ) prescribed burns; ( b ) wildfires; and ( c ) time since last fire across the entire Osceola National Forest, FL. 4. Discussion 4.1. Limitations Associated with dN BRs in the Southeastern US The Use of dNBR is very challenging in many regi ons including the southeas tern US Given that dNBR is sensitive to other sources of land-cover change, dNBR imagery will respond to any land cover change between the pre-fire and post-fire images [35]. Rapid vegetation re-growth following fire, fluctuating hydrology in hydric communities and frequent cloud cove r may cause errors in

PAGE 17

Remote Sens. 2011 3 2021 detecting burned severity [19]. Periodic variati ons in soil moisture and hydrology can influence the spectral reflectance values captured in dNBRs [36]. Standing water can also lead to the miss-classification of burned areas as unburned. The cy clical droughts experienced by the southeastern region of the US can cause a reduction in vegeta tion greenness that can mimic damage due to burn severity. Seasonal timing of fires and time lags betw een preand post-fire imagery also accounts for detected differences that are not associated with fi re effects. When change in dNBR is due to natural phenological differences between preand post-fire images, there is a reduced distinction between burned and unburned areas and this c ould cause false positives for fire effects near or within burn perimeters. Nonetheless, dNBR has been shown to be a good measure of burn severity in the pine flatwoods of the south and can be used to identify the drivers of high fire severity. 4.2. Time and Severity Thresholds for Pr eventing High Burn Severity Effects The probability of experiencing high burn severit y, increasing and decreasing burn severity, and burning in subsequent fires has impo rtant implications for fire effects and the degree to which high severity fire is being mitigated. Based on the severity level of the first fire, fire type, and community type, we have the capacity to identify target time intervals between fires. On the Osceola National Forest, a time between fires of less than five y ears has been shown to have a mitigating effect on subsequent burn severity. After five years there is a marked increase in the probability of high burn severity (Figure 2(a)), of subsequent fires increasing in burn se verity (Figure 2(b)), and fire risk. Previous studies conducted on the Os ceola National Forest indicated th at time between fires must be kept below three years to adequately reduce the o ccurrence of catastrophic wildfire [1]. Vegetation recovery following fire is expedient due to fast growing and re-sprouting species [37]. Lemon [38] used permanent plots on the Alapaha Experimental Range (Georgia) to monito r changes in vegetation following prescribed fire, and found that the maximum amount of litter is approached at five years’ post fire, while by eight years post fire vegetation re turned to pre-burn status Findings presented in Outcalt and Wade [2] and Lemon [38] support our results that identify target fire return intervals of less than five years to mitigate the effect of high burn severity in subsequent fires. The relationship between fire occurrence and time interval between fires greater than six years implies that vegetation that has re mained unburned for seven to 10 ye ars may be less available to burn. This is likely due to higher fuel moisture cont ents and changes in specie s composition. Areas burned by wildfires also have a higher prob ability of subsequent high burn seve rity than areas first burned by prescribed fire and, hydric communities have a hi gher probability of high bur n severity than mesic communities. Prescribed burns are pe rformed under conditions that facilit ate low severity fire effects. During prescribed burns, understory fuel in hydric communities is partially consumed with little consumption of the duff layer [2]. Therefore, hydric areas generally carry very heavy fuel volumes that become largely available during ex tended drought periods, making them capable of very large very intense wildfires [1]. Our results suggest that areas th at have previously burned at high burn se verity have a lower probability of decreasing in burn severity in subsequent fire. Although this result may seem counterintuitive, a study in New Mexico found that hi gh burn severity areas were more likely to burn again at a high burn severity level in pinyon ( Pinus L.) -juniper ( Juniperus L.) woodlands [39].

PAGE 18

Remote Sens. 2011 3 2022 Although the vegetation types differ significantl y, this phenomenon suggests that recovery time between fires where high burn severity occurred is long enough that the fuel structure and composition once again supports high severity fi re. While recovering, these areas ar e less likely to be significantly altered by fire occurrence, simply because there is less fuel to be altered: In other words, remote sensing may underestimate burn severity in areas wh ere fuels are limited as a result of high burn severity. Whether this phenomenon is a result of a positive relationship between fuels recovery and fire susceptibility, or the result of inherent limitatio ns of using remote se nsing analysis alone ( i.e. without field validation) remains to be determined. 4.3. The Influence of Fire Frequency and Time since Last Fire on Severity Fire frequency is one of the most important determinants for sustai ning flatwoods ecosystems [40]. The relationship between fire frequency and the proba bility of high burn severity in prescribed burns and wildfires were not surprising; as frequencies increased the pr obability of high burn severity decreased (Figure 3). Post fire vegetation recovery to pre-fire biomass levels is rapid (1–4 years) in pine flatwoods forests [41-43]. Regular fire can reduce the amount of shrubby understory vegetation and promote grassy and herbaceous species that facili tate fire spread. Regular fire also reduces the growth of less-flammable broadleaf understory species and increases biodivers ity [44]. Less frequent fire promotes aerial, surface, and ground fuel buil dup, which likely facilitates high burn severity at least until significant composition change decr eases flammability. The model developed here successfully captures the expected relationshi p between fire frequency and burn severity. That the results suggest a negative relationship be tween time since last fire and the probability of high burn severity was surprising. We would expect the probability of high burn severity to increase with increased time since last fire [45], although th e change in forest flammability associated with increased broadleaf abundance over time may help to explain this result. Prev ious studies of a single wildfire conducted in th e Osceola National Forest found that as time between fire events increased from 1.5 to 3 years, post-fire tree mortality also in creased [2]. Additional analysis of pine flatwoods elsewhere in the state showed simila r trends [2]. A shift in dominant species from coniferous species to less flammable deciduous species could be respons ible for the observed relationship between time since last fire and the probabil ity of high burn severity. Deciduous litter is less aera ted, and does not propagate fire spread and combustion as well as c oniferous litter [44]. As deciduous encroachment increases over time, the probability of high burn severity may be decreased. Expected changes in microclimate and fuel moistu re content may also help explain why the model did not reflect the expected relati onship between time since last fire and the probability of high burn severity. In the mesic-hydric pine flatwoods forests of the Osceola NF, fire initiation and spread are limited by the abundance of saw palmetto shrubs, which, although flammable, have higher moisture contents than most grassy or herbaceous species. High rates of decomposition result in minimal fine fuel accumulation on the forest floor. According to fuels management officers, lightning ignitions in the Osceola forest are far less frequent than th ose recorded in the nearby Appalachicola National Forest, where higher depth to the water table results in drier soils and a grea ter abundance of grassy and herbaceous fuels [46]. The negative relationship between time since last fire and the potential for high burn severity may be a result of higher fuel mo isture contents, higher relative humidity, and lower

PAGE 19

Remote Sens. 2011 3 2023 temperatures where saw palmetto cover has had a l onger time to recover from the previous fire. Where fire has more recently occurred, fine fuels from n eedle cast would contribute to the available fuel, temperatures and wind speed would be higher due to reduced canopy cover, an d fire effects may be more severe when compared with long-unburned areas. The negative relationship between time since last fire and the proba bility of high bur n severity may also be due to a bias attributable to the influence of prescribed fires, which comprised 43% of the total fires used in the model. Suitable pr escribed fire conditions are determ ined by time since last fire and fire frequency. Land managers may be willing to burn areas that were more recently burned under more extreme weather conditions due to lower fuel loads and lower potential for fire escape. Fire effects in these areas may then appear, at least tem porarily, more severe than in areas that were burned under less extreme weather conditions. This may also be attributable to the fact that prescribed burns on the Osceola National Forest are primarily conducte d in the winter, when canopies are less dense and fire severity at the ground level is more apparent in the imagery. There is some support for these explanations: hi gh pre-fire biomass may be the cause of the relationship between time since last fire and the probability of high burn se verity. Previous studies have identified a relationship betw een burn severity level and prefire tree density [10,13]. Burn severity was highest in dense areas with high basa l area and areas with high pre-fire litter and duff depths in the Coconino National Fore st in Arizona [10]. These areas have the potential to both have and to indicate greater change between preand pos t-fire imagery [10]. Other authors have suggested that additional biases might exist. Using a relative version of dNBR, Miller and Thode [13] discovered a misclassification of areas with low vegetation due to the minimal detectable changes in a California forest. There was also a bias detected in classifyi ng burn severity in multiple vegetation types using the same severity thresholds [13]. Alle n and Sorbel [26] detected a sim ilar bias in the low burn severity class for deciduous forest and tall shrubs in boreal forest and Tundra ecosystems in Alaska. In this study, severity values for deciduous forest were lower than their composite burn index (CBI). 4.4. Burn Severity in Relation to Fire Ty pe, Community Type, and Forest Moisture Overall, the probability of high burn severity in wild fires was less than what we would expect when compared to prescribed burns, whic h are commonly assu med to be lower in severity (Figure 5(a,b)). The low probability may have been caused by how wildfires were mapped or may indicate that active fire management on the Osceola is effectively suppressing high severity wildfire. Wildfire perimeters were mapped using Landsat imagery based on ocular estimates of where fires occurred. The perimeters were not always exact, resulting in unburned and low burn se verity pixels occurring within the perimeter of a given wildfire. In contrast, prescribed burns tend to be more homogenous within their perimeters, as most are conducted during the winter season when fu els are uniformly dry. Also, most (70%) wildfires within this dataset were less than 10 ha in size. Wildfire size is determined by both suppression efforts and fuel availability, and if suppression is su ccessful, fires are extinguished before fuels are extensively consumed, reducing the total area of high burn severity in wildfires. Successful fire suppression also implies that weat her conditions may not have been conducive to promoting high burn severity effects. There were few wildfires (the Oak Fire of 1998, Friendly fi re of 1999 Impassible Bay fire of 2004, and the Bugaboo fire of 2007) that were large (e.g., greater than 3,000 ha) in size and that

PAGE 20

Remote Sens. 2011 3 2024 required extensive suppression efforts. In this regi on, wildfires typically occu r in the early spring, and the weather systems responsible for the lightning ignition may be accompanied by significant precipitation, increasing the patchiness of burns a nd ultimately reducing the overall continuity of burn severity. Our data shows that the majority of wildfires occurred in the spring, wh ile most prescribed burns were conducted in the drier winter months. Fuel consumption may ther efore be greater during the winter burns. Along with fire season, biases introduced by perimeter estimates and the higher proportion of smaller, less severe wildfires were likely causes for the low observed probability of experiencing high burn se verity in wildfires. Forest and community types were not significant predictors for subsequent high burn severity. We expected hydric communities to burn less severe ly in prescribed burns and potentially more severely during wildfires, when the high fuel lo ads could be dry enough to ignite during prolonged drought periods [2]. Within this da taset, there was no significant di fference between fire effects in hydric and mesic forests. Th is lack of differen ce in fuel availabili ty may have been due to suppression efforts or climatic factors. S uppression efforts may have been fo cused on preventing wildfires from entering areas of heavy fuel loads, climatic conditi ons during the burning of hydric areas may not have facilitated high burn severity effect s. We also expected forest type s to influence wildfire severity levels. The lack of significance may be due to the limited representation of non-flatwoods forest types in the Osceola National Forest (24%). Additional investigation into the relationships between forest age, structure, and severity patterns would incr ease our understand ing of the importance of forest characteristics in affecting fire effects. The spatial model identified areas where high severi ty prescribed burns and wildfires were expected to occur in 2007 based on fire frequ ency, fire type, and the amount of time since the last fire. Areas more recently burned exhibited a higher likelihood of high burn severity (Figure 5(a–c)). Most of the area impacted in 2007 was burned by prescribed fire and burned at moderate (21%) and low (78%) severity levels. Sections of the prescribed burn s that actually displayed high burn severity had probabilities of high burn severity greater than 75% (Fig ure 4(a,b)). This suggests that the model adequately identified areas that had a high risk of high burn seve rity in a prescribed fire. The effectiveness of this model in predicting future burn severity should be further evaluated using remotely sensed burn severity data linked to ground-based evalua tions. Given that the Osceola National Forest usually reaches its annual targets of 13,000 prescribed burn ha, there will be ample opportunity to validate this model here and in other flatwoods forests across the southern US. 5. Conclusions and Recommendations Remote sensing techniques were successfully used to model nearly a decad e’s worth of fires to determine high burn severity risk and important time thresholds for pine flatwoods management. The models identified areas that require attention in orde r to reduce the risk of hi gh burn severity effects, especially for prescribed burns that are commonly assumed to exhibit low burn severity [47]. Our analysis indicates that time since last fire and fire frequency are major factors a ffecting the risk of high burn severity. A fire-free interval of less than five years is recommended to reduce the risk of high burn severity in pine flatwoods forests. Changes in vegetation and microclimate in response to less

PAGE 21

Remote Sens. 2011 3 2025 frequent fire may require more extreme weather conditions to burn [43], increasing the probability of high severity fires. Areas that burned more recently had an elevated ri sk of high burn severity for both prescribed burns and wildfires. This result may be attributable to a bias in the detection of high severity effects in relation to pre-fire biomass, if areas that were recently burned are burned under more extreme weather prescriptions, or if vegetation recovery corresponds to a shift in flammability and microclimate that reduces burn severity. Further research addressi ng the relationship between pre-fire biomass, vegetation type, and dNBRs is necessa ry to determine if there is a bi as occurring in the low severity class due to species composition, or a bias in the high se verity class that is asso ciated with high pre-fire biomass in pine flatwoods forests. In areas with short fire return interv als, it may also be useful to look at the effects of delayed mortality to identify if this would cause further error in detecting high burn severity effects. Directly following a fire, delayed mortality may cause a bias in the low burn severity class and, burn severity in subseque nt fire may exhibit a bias in the high burn severity class due to the detection of fire effects from the previous fire. The models created here can effectively identify tim e thresholds that facilitate increased risks of high burn severity and areas with an increased risk based on the hi story of fire. Additionally, the models are able to capture the relationship between fire frequency and high severity, and time between fires and increased risks of high bu rn severity. As fire frequency ha s been identified as an important indicator of ecosystem condition in flatwoods fore st [40], these models can be used to inform prioritization and timing of fire use to maintain the pine flat woods forests of the southern Coastal Plain. Acknowledgements This work was funded in part by the USDA Nationa l Needs Fellowship, and the Conserved Forest Ecosystems: Outreach and Research (CFEOR) cooperative at the University of Florida School of Forest Resources and Conservatio n. We thank Jason Drake at the US Forest Service Supervisors Office in Tallahassee, FL, USA, fo r providing data for this projec t, along with the personnel of the Osceola National Forest. We would also like to thank the Kobziar Fire Science Lab for their assistance. References 1. Davis, L.S.; Cooper, R.W. How prescrib ed burning affects wildfire occurrence. J. For. 1963 61 915-917. 2. Outcalt, K.W.; Wade, D.D. Fuels management reduces tree mortalit y from wildfires in southeastern United States. South. J. Appl. For. 2004 28 28-34. 3. Main, M.B.; Richardson, L.W. Response of wildlife to prescribed fire in southwest Florida pine flatwoods. Wildl. Soc. Bull. 2002 30 213-133. 4. Heuberger, K.A.; Putz, F.E. Fire in the suburbs: Ecological impacts of prescribed fire in small remnants of longleaf pine sandhill. Restor. Ecol. 2003 11 72-81.

PAGE 22

Remote Sens. 2011 3 2026 5. Van Lear, D.H.; Carroll, W.D.; Kapeluck, P.R.; Johnson, R. History and restoration of the longleaf pine-grassland ecosystem: Im plication for species at risk. For. Ecol. Manag. 2005 211 150-165. 6. Parr, C.L.; Andersen, A.N. Patch mosaic burning for biodiversity conservation: A critique of the pyro diversity paradigm. Conserv. Biol. 2006 20 1610-1619. 7. Wolcott, L.; O’Brien, J.J.; Mordecai, K. A su rvey of land managers on wildland hazardous fuels issues in Florida: A technical note. South. J. Appl. For. 2007 31 148-150. 8. Agee, J.K. Fire Ecology of Pacific Northwest Forest ; Island Press: Washington, DC, USA, 1993; p. 417. 9. Hardy, C.C. Wildland fire hazard and risk: Problems, definitions, and context. For. Ecol. Manag. 2005 211 73-82. 10. Cocke, A.E.; Fule, P.Z.; Crouse, J.E. Comparis on of burn severity assessment using difference Normalized Burn Ratio and ground data. Int. J. Wildl. Fire 2005 14 189-198. 11. Holden, Z.; Morgan, P.; Evans, J. A predictive model of burn severity based on 20-year satelliteinfrared burn severity data in a la rge southwestern US wilderness area. For. Ecol. Manag. 2009 258 2399-2406. 12. Collins, B.M.; Miller, J.D.; Thode, A.E.; Kelly, M.; Wagtendonk, J.W.; Stephens, S. Interactions among wildland fires in a long-establishe d sierra nevada natural fire area. Ecosystems 2009 12 114-128. 13. Miller, J.D.; Thode, A.E. Quantifying burn severity in a heterogeneous landscape with relative version of the delta Normalized Burn Ratio (dNBR). Remote Sens. Environ. 2007 109 66-80. 14. van Wagtendonk, J.W.; Root, R.R.; Key, C.H. Comparison of AVIRS and Landsat ETM+ detection capabilities for burn severity. Remote Sens. Environ. 2004 92 397-408. 15. Godwin, D.R. Burn severity in a centr al Florida sand pine scrub wilderness area M.Sc. Thesis, University of Florida Press, Gainesville, FL, USA, 2008. 16. Hoy, E.E.; French, N.H.; Turetsky, M.R.; Trigg, S. N.; Kasischke, E.S. Evaluating the potential of landsat TM/ETM+ imagery for assessing fire severity in Alaska black spruce forest. Int. J. Wildl. Fire 2008, 17 500-514. 17. White, J.D.; Ryan, K.C.; Key, C. C.; Runnig, S.W. Remote sensing of forest fire severity and vegetation recovery. Int. J. Wildl. Fire 1996 6 125-136. 18. French, N.H.F.; Kasischke, E.S.; Hall, R.J.; Mu rphy, K.A.; Verbyla, D.L.; Hoy, E.E.; Allen, J.L. Using Landsat data to assess fire and burn severity in th e North American bore al forest region: An overview and summary of results. Int. J. Wildl. Fire 2008 17 443-462. 19. Picotte, J.J.; Robertson, K.M. Validation of remo te sensing of burn severi ty in southeastern US ecosystems. Int. J. Wildl. Fire 2011 20 453-464. 20. Safford, H.D.; Schmidt, D.A.; Carson, C.H. Effects of fuel treatments on fire se verity in an area of wildland-urban interface, Angora Fi re, Lake Tahoe Basin, California. For. Ecol. Manag. 2009 258 773-787. 21. Kuenzi, A.M.; Fule, P.Z.; Sieg, C.H. Effects of fi re severity and pre-fire stand treatment on plant community recovery after a large wildfire. For. Ecol. Manag. 2008 255 855-865.

PAGE 23

Remote Sens. 2011 3 2027 22. Oliveras, I.; Gracia, M.; Mor, G.; Retana, J. Fact ors influencing the pattern s of fire severity in large wildland fire under extreme meteorologi cal conditions in the Mediterranean basin. Int. J. Wildl. Fire 2009 18 755-764. 23. Eidenshink, J.; Schwind, B.; Brewer, K.; Z hu1, Z.; Quayle, B; Howard, S. A project for monitoring trends in burn severity. Fire Ecol. 2007 3 3-20. 24. Mitchener, L.J.; Parker, A.J. Climate, lightni ng, and wildfire in the national forests of the southeastern United States: 1989–1998. Phys. Geogr. 2005 26 147-162. 25. Earth Resources Observation an d Science (EROS). Available on line: http://gl ovis.usgs.gov/ distribution/downloadnotices.shtml (accessed on 7 January 2010). 26. Allen, J.L.; Sorbel, B. Assessing the differenced Normalized Burn Ratio ’s ability to map burn severity in boreal forest and tundra ec osystems of Alaska’s national parks. Int. J. Wildl. Fire 2008 17 463-475. 27. Zhu, Z.; Key, C.H.; Ohlen, D.; Benson, N.C. Evaluate Sensitivities of Burn Severity Mapping Algorithms for Different Ecosystems and Fire Histories in the United States ; Technical Report; Joint Fire Science Program: Boise, ID, USA, October 2006. 28. Key, C.H. Ecological and sampling constrai nts on defining landscape fire severity. Fire Ecol. 2006 2 34-59. 29. United States Geologic Servi ce (USGS). Difference Normalized Burn Ratio. Available online: http://burnseverity.cr.usgs.gov/pdfs/LAv4_BR_ CheatSheet.pdf (accessed on 7 January 2010). 30. Alley, W.M. The palmer drought severity index: Limitations and assumptions. J. Clim. Appl. Meteorol. 1984 23 1100-1109. 31. Florida Natural Areas I nventory and Department of Natural Resources. Guide to the Natural Communities of Florida ; Florida Natural Areas Inventory a nd Department of Natural Resources: Tallahassee, FL, USA, 1990. 32. Davis, J.H. General Map of Natural Vegetation in Florida. Agricultura l Experimentations ; Institute of Food and Agricultural Sciences, Univer sity of Florida: Gainesville, FL, USA, 1967. 33. Littell, R.C.; Milliken, G.A.; Stroup, W. W.; Wolfinger, R.D.; Schabenberger, O. SAS for Mixed Models 2nd ed.; SAS Publishing: Caryn, NC, USA, 2006; p. 170. 34. SAS Institute Inc. SAS/STAT 9.2 User’s Guide ; SAS Institute Inc.: Cary, NC, USA, 2008. 35. Key, C.H. Remote Sensing Sensitivity to Fire Severity and Fire Recovery. In Proceedings of the 5th International Workshop on Remote Sensing and GIS Appl ications to Forest Fire Management: Fire Effects Assessment Zaragoza, Spain, 16–18 June 2005. 36. Key, C.H.; Benson, N.C. Landscape Assessment: Ground Measur e of Severity, the Composite Burn Index; and Remote Sensing of Severity, the Normalized Burn Ratio ; Technical Report of the USDA Forest Service; Rocky Mountain Re search Station: Ogden, UT, USA, 2006. 37. Lavoie, M.; Starr, G.; Mack, M. C.; Martin, T.A.; Gholz, H.L. Ef fects of a prescribed fire on understory vegetation, carbon pools, a nd soil nutrients in longleaf pine-slash pine forest in Florida. Natl. Areas J. 2010 30 82-94. 38. Lemon, P.C. Successional respons es of herbs in the longleafsl ash pine forest after fire. Ecology 1949 30 135-145. 39. Holden, Z.A.; Morgan, P.; Hudak, A.T. Burn seve rity of areas reburned by wildfires in the Gila National Forest, New Mexico, USA. Fire Ecol. 2010 6 77-85.

PAGE 24

Remote Sens. 2011 3 2028 40. Mitchell, J.K.; Hiers, J.K.; O’Brien, J.J.; Jack, S. B.; Engstrom, R.T. Silvic ulture that sustains: The nexus between silviculture, frequent prescrib ed fire, and conservation of biodiversity in longleaf pine forests of the southeastern United States. Can. J. For. Resour. 2006 36 2724-2736. 41. Abrahamson, W.G. Species response to fire on the Florida Lake Wales Ridge. Am. J. Bot. 1984 71 35-43. 42. Abrahamson, W.G.; Abrahamson, C.R. Effect s of fire on long unburned Florida uplands. J. Veg. Sci. 1996 7 565-574. 43. Maliakal, S.K.; Menges, E.S. Community com position and regeneration of Lake Wales Ridge wiregrass flatwoods in relati on to time since last fire. J. Torrey Bot. Soc. 2000 127 125-138. 44 Moore, W.H.; Swindel, B.F.; Terry, S.W. Vegetativ e response to prescribed fire in a north Florida flatwoods forest. J. Range Manag. 1982 35 386-389. 45. DeBano, L.F.; Neary, D.G.; Ffolliott, P.F. Fire’s Effects on Ecosystems ; Wiley: New York, NY, USA, 1998; pp. 333. 46. Kinghorn, S. USFS Osceola National Forest. Ol ustee, Florida, Personal Communication, 2010. 47. Van Wagtendonk, J.W.; Lutz, J.A. Fire regime a ttributes wildland fires in Yosemite National Park, USA. Fire Ecol. 2007 3 34-52 2011 by the authors; licensee MDPI, Basel, Switz erland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution license (http://creativecommons.org/licenses/by/3.0/)