Wildfire Mapping Research Articles

Automatic Methodology for Forest Fire Mapping with SuperDove Imagery.

The global increase in wildfires due to climate change highlights the need for accurate wildfire mapping. This study performs a proof of concept on the usefulness of SuperDove imagery for wildfire mapping. To address this topic, we present an automatic methodology that combines the use of various vegetation indices with clustering algorithms (bisecting k-means and k-means) to analyze images before and after fires, with the aim of improving the precision of the burned area and severity assessments. The results demonstrate the potential of using this PlanetScope sensor, showing that the methodology effectively delineates burned areas and classifies them by severity level, in comparison with data from the Copernicus Emergency Management Service (CEMS). Thus, the potential of the SuperDove satellite sensor constellation for fire monitoring is highlighted, despite its limitations regarding radiometric distortion and the absence of Short-Wave Infrared (SWIR) bands, suggesting that the methodology could contribute to better fire management strategies.

Impact of fire severity on forest structure and biomass stocks using NASA GEDI data. Insights from the 2020 and 2021 wildfire season in Spain and Portugal

Wildfires have been progressively shrinking the C sink capacity of forest accelerating climate change effects on forest biodiversity, especially where megafires are recurrent and of increased frequency such as in the Mediterranean. Data from The Global Ecosystem Dynamics Investigation (GEDI) mission can inform on changes on forest structure to inform on fire impacts on vegetation. In this study, we assessed the performance of GEDI at measuring biomass and structural change from wildfires using the 2020/21 summer fire seasons in Spain and Portugal. The GEDI hybrid-inference method was used to calculate mean and total biomass in pre- and post-fire stages, while GEDI footprint data was further used to explain the fire severity classes derived from optical data. Our results showed the increasing impact of wildfires on biomass stocks and GEDI ecological metrics by increasing fire severity. More than over biomass stocks, severe fires substantially altered trends in structural metrics such as plant area volume density. The integration of GEDI metrics to explain fire severity had an accuracy of 52% considering five severity classes and an accuracy of 69% when considering the three main classes: unburned, moderate and high. Structural metrics from GEDI can be used to improve optical-based fire severity estimates used globally and to evaluate potential fire impacts based on time-series of GEDI tracks as showcased in the study, but also to measure forest recovery between fire seasons. The extension of GEDI is a major support for wildfire mapping efforts, integrated approaches to capture the increasing impact of fire on forest biodiversity and the monitoring of changes in carbon stocks.

Assessing Sentinel-2, Sentinel-1, and ALOS-2 PALSAR-2 Data for Large-Scale Wildfire-Burned Area Mapping: Insights from the 2017–2019 Canada Wildfires

Wildfires play a crucial role in the transformation of forest ecosystems and exert a significant influence on the global climate over geological timescales. Recent shifts in climate patterns and intensified human–forest interactions have led to an increase in the incidence of wildfires. These fires are characterized by their extensive coverage, higher frequency, and prolonged duration, rendering them increasingly destructive. To mitigate the impact of wildfires on climate change, ecosystems, and biodiversity, it is imperative to conduct systematic monitoring of wildfire progression and evaluate their environmental repercussions on a global scale. Satellite remote sensing is a powerful tool, offering precise and timely data on terrestrial changes, and has been extensively utilized for wildfire identification, tracking, and impact assessment at both local and regional levels. The Canada Centre for Mapping and Earth Observation, in collaboration with the Canadian Forest Service, has developed a comprehensive National Burned Area Composite (NBAC). This composite serves as a benchmark for curating a bi-temporal multi-source satellite image dataset for change detection, compiled from the archives of Sentinel-2, Sentinel-1, and ALOS-2 PALSAR-2. To our knowledge, this dataset is the inaugural large-scale, multi-source, and multi-frequency satellite image dataset with 20 m spatial resolution for wildfire mapping, monitoring, and evaluation. It harbors significant potential for enhancing wildfire management strategies, building upon the profound advancements in deep learning that have contributed to the field of remote sensing. Based on our curated dataset, which encompasses major wildfire events in Canada, we conducted a systematic evaluation of the capability of multi-source satellite earth observation data in identifying wildfire-burned areas using statistical analysis and deep learning. Our analysis compares the difference between burned and unburned areas using post-event observation solely or bi-temporal (pre- and post-event) observations across diverse land cover types. We demonstrate that optical satellite data yield higher separability than C-Band and L-Band Synthetic Aperture Radar (SAR), which exhibit considerable overlap in burned and unburned sample distribution, as evidenced by SAR-based boxplots. With U-Net, we further explore how different input channels influence the detection accuracy. Our findings reveal that deep neural networks enhance SAR’s performance in mapping burned areas. Notably, C-Band SAR shows a higher dependency on pre-event data than L-Band SAR for effective detection. A comparative analysis of U-Net and its variants indicates that U-Net works best with single-sensor data, while the late fusion architecture marginally surpasses others in the fusion of optical and SAR data. Accuracy across sensors is highest in closed forests, with sequentially lower performance in open forests, shrubs, and grasslands. Future work will extend the data from both spatial and temporal dimensions to encompass varied vegetation types and climate zones, furthering our understanding of multi-source and multi-frequency satellite remote sensing capabilities in wildfire detection and monitoring.

A Lightweight Remote Sensing Payload for Wildfire Detection and Fire Radiative Power Measurements

Small uncrewed aerial systems (sUASs) have the potential to serve as ideal platforms for high spatial and temporal resolution wildfire measurements to complement aircraft and satellite observations, but typically have very limited payload capacity. Recognizing the need for improved data from wildfire management and smoke forecasting communities and the potential advantages of sUAS platforms, the Nighttime Fire Observations eXperiment (NightFOX) project was funded by the US National Oceanic and Atmospheric Administration (NOAA) to develop a suite of miniaturized, relatively low-cost scientific instruments for wildfire-related measurements that would satisfy the size, weight and power constraints of a sUAS payload. Here we report on a remote sensing system developed under the NightFOX project that consists of three optical instruments with five individual sensors for wildfire mapping and fire radiative power measurement and a GPS-aided inertial navigation system module for aircraft position and attitude determination. The first instrument consists of two scanning telescopes with infrared (IR) channels using narrow wavelength bands near 1.6 and 4 µm to make fire radiative power measurements with a blackbody equivalent temperature range of 320-1500 °C. The second instrument is a broadband shortwave (0.95-1.7 µm) IR imager for high spatial resolution fire mapping. Both instruments are custom built. The third instrument is a commercial off-the-shelf visible/thermal IR dual camera. The entire system weighs about 1500 g and consumes approximately 15 W of power. The system has been successfully operated for fire observations using a Black Swift Technologies S2 small, fixed-wing UAS for flights over a prescribed grassland burn in Colorado and onboard an NOAA Twin Otter crewed aircraft over several western US wildfires during the 2019 Fire Influence on Regional to Global Environments and Air Quality (FIREX-AQ) field mission.

Geospatial Wildfire Risk Assessment from Social, Infrastructural and Environmental Perspectives: A Case Study in Queensland Australia

Although it is hard to predict wildfires, risky areas can be systematically assessed and managed. Some of the factors for decision-making are hazard, vulnerability, and risk maps, which are the end product of wildfire mapping. This study deals with wildfire risk analysis in Queensland, Australia. A review of the previous studies focusing on each aspect has been done and used with wildfire records from 2011 to 2019 in Queensland, Australia, to compile the required input models to detect risky wildfire regions. Machine learning (ML) methods of Decision Tree (DT) and Support Vector Machine (SVM) were used to perform hazard assessment. The reason was to select the most accurate outcomes for the rest of the analysis. Among accuracy assessment techniques, the Area Under Curvature (AUC) method was used to evaluate the hazard maps. Prediction rates of 89.21% and 83.78% were obtained for DT and SVM, respectively. The DT prediction value showed that the DT-hazard map was more accurate than the SVM-hazard map. Vulnerability analysis was implemented by assigning weights to each factor according to the literature. Lastly, in order to create the wildfire risk map, the hazard and vulnerability indices were combined. The risk map showed that particularly dense urbanization regions are under future wildfire risk. To perform preliminary land use planning, this output can be used by local governmental authorities.

A Machine Learning Approach to Waterbody Segmentation in Thermal Infrared Imagery in Support of Tactical Wildfire Mapping

Wildfire research is working toward near real-time tactical wildfire mapping through the application of computer vision techniques to airborne thermal infrared (IR) imagery. One issue hindering automation is the potential for waterbodies to be marked as areas of combustion due to their relative warmth in nighttime thermal imagery. Segmentation and masking of waterbodies could help resolve this issue, but the reliance on data captured exclusively in the thermal IR and the presence of real areas of combustion in some of the images introduces unique challenges. This study explores the use of the random forest (RF) classifier for the segmentation of waterbodies in thermal IR images containing a heterogenous wildfire. Features for classification are generated through the application of contextual and textural filters, as well as normalization techniques. The classifier’s outputs are compared against static GIS-based data on waterbody extent as well as the outputs of two unsupervised segmentation techniques, based on entropy and variance, respectively. Our results show that the RF classifier achieves very high balanced accuracy (>98.6%) for thermal imagery with and without wildfire pixels, with an overall F1 score of 0.98. The RF method surpassed the accuracy of all others tested, even with heterogenous training sets as small as 20 images. In addition to assisting automation of wildfire mapping, the efficiency and accuracy of this approach to segmentation can facilitate the creation of larger training data sets, which are necessary for invoking more complex deep learning approaches.

A Quadratic Morphological Deep Neural Network Fusing Radar and Optical Data for the Mapping of Burned Areas

Wildfires are considered as one of the most disturbing factors in forest areas and high-density vegetation regions. The mapping of wildfires is particularly important for fire prediction and burned biomass estimation. Therefore, accurate and timely mapping of burned areas is of great importance and has a key role in disaster management. Estimation of burned areas from multi-spectral datasets is challenging, because of the complexity of the background and the different reflections of wildfires on the targets. To this end, this study presents a novel burned-area mapping framework based on fusion of multi-temporal Sentinel-1 coherence imagery and post-event Sentinel-2 imagery. The proposed framework uses hybrid quadratic morphological (QM) operations and convolution layers for deep feature extraction. The proposed architecture is known as the QMDNN-Net, where the overall framework of QMDNN-Net is based on a deep Siamese network. QMDNN-Net has double streams for extracting deep features from multi-temporal coherence data and from Sentinel-2 imagery. The streams are similar to each other, and have the same number of group-dilated convolution blocks and QM layers. QMDNN-Net is defined based on quadratic dilation and erosion, and then it takes the average of these as output. The results of wildfire mapping with QMDNN-Net are evaluated here with two real-world wildfire datasets based on Sentinel-1 and Sentinel-2 imagery. The results show that QMDNN-Net achieves an overall accuracy and Kappa coefficient of 95.5% and 0.9, respectively, and outperforms other state-of-the-art methods.

A machine learning framework for multi-hazards modeling and mapping in a mountainous area

This study sought to produce an accurate multi-hazard risk map for a mountainous region of Iran. The study area is in southwestern Iran. The region has experienced numerous extreme natural events in recent decades. This study models the probabilities of snow avalanches, landslides, wildfires, land subsidence, and floods using machine learning models that include support vector machine (SVM), boosted regression tree (BRT), and generalized linear model (GLM). Climatic, topographic, geological, social, and morphological factors were the main input variables used. The data were obtained from several sources. The accuracies of GLM, SVM, and functional discriminant analysis (FDA) models indicate that SVM is the most accurate for predicting landslides, land subsidence, and flood hazards in the study area. GLM is the best algorithm for wildfire mapping, and FDA is the most accurate model for predicting snow avalanche risk. The values of AUC (area under curve) for all five hazards using the best models are greater than 0.8, demonstrating that the model’s predictive abilities are acceptable. A machine learning approach can prove to be very useful tool for hazard management and disaster mitigation, particularly for multi-hazard modeling. The predictive maps produce valuable baselines for risk management in the study area, providing evidence to manage future human interaction with hazards.

Spatiotemporal Variations and Factors of Air Quality in Urban Central China during 2013–2015

Spatiotemporal behaviors of particulate matter (PM2.5 and PM10) and trace gases (SO2, NO2, CO, and O3) in Hefei during the period from December 2013 to November 2015 are investigated. The mean annual PM2.5 (PM10) concentrations are 89.1 ± 59.4 µg/m3 (118.9 ± 66.8 µg/m3) and 61.6 ± 32.2 µg/m3 (91.3 ± 40.9 µg/m3) during 2014 and 2015, respectively, remarkably exceeding the Chinese Ambient Air Quality Standards (CAAQS) grade II. All trace gases basically meet the requirements though NO2 and O3 have a certain upward trend. Old districts have the highest pollution levels, followed by urban periphery sites and new districts. Severe haze pollution occurs in Hefei, with frequent exceedances in particulate matter with 178 (91) days in 2014 (2015). The abnormal PM2.5 concentrations in June 2014 attributed to agricultural biomass burning from moderate resolution imaging spectroradiometry (MODIS) wildfire maps and aerosol optical depth (AOD) analysis. PM2.5 is recognized as the major pollutant, and a longer interspecies relationship is found between PM2.5 and other criteria pollutants for episode days as compared to non-episode days. The air pollution in Hefei tends to be influenced by local primary emissions, secondary formation, and regional transport from adjacent cities and remote regions. Most areas of Anhui, southern Jiangsu, northern Zhejiang, and western Shandong are identified as the common high-potential source regions of PM2.5. Approximately 9.44 and 8.53 thousand premature mortalities are attributed to PM2.5 exposure in 2014 and 2015. The mortality benefits will be 32% (24%), 47% (41%), 70% (67%), and 85% (83%) of the total premature mortalities in 2014 (2015) when PM2.5 concentrations meet the CAAQS grade II, the World Health Organization (WHO) IT-2, IT-3, and Air Quality Guideline, respectively. Hence, joint pollution prevention and control measures need to be strengthened due to pollutant regional diffusion, and much higher health benefits could be achieved as the Hefei government adopts more stringent WHO guidelines for PM2.5.

Exploring the Potential of Sentinels-1 & 2 of the Copernicus Mission in Support of Rapid and Cost-effective Wildfire Assessment

The present study explores the use of the recently launched Sentinel-1 and -2 data of the Copernicus mission in wildfire mapping with a particular focus on retrieving information on burnt area, burn severity as well as in quantifying soil erosion changes. As study area, the Sierra del Gata wildfire occurred in Spain during the summer of 2015 was selected. First, diverse image processing algorithms for burnt area extraction from Sentinel-2 data were evaluated. In the next step, burn severity maps were derived from Sentinel-2 data alone, and the synergy between Sentinel-2 & Sentinel-1 for this purpose was evaluated. Finally, the impact of the wildfire to soil erodibility estimates derived from the Revised Universal Soil Loss Equation (RUSLE) model implemented to the acquired Sentinel images was explored. In overall, the Support Vector Machines (SVMs) classifier obtained the most accurate burned area mapping, with a derived accuracy of 99.38%. An object-based SVMs classification using as input both optical and radar data was the most effective approach of delineating burn severity, achieving an overall accuracy of 92.97%. Soil erosion mapping predictions allowed quantifying the impact of wildfire to soil erosion at the studied site, suggesting the method could be potentially of a wider use. Our results contribute to the understanding of wildland fire dynamics in the context of the Mediterranean ecosystem, demonstrating the usefulness of Sentinels and of their derived products in wildfire mapping and assessment.

Is a picture worth a thousand words? Evaluating the effectiveness of maps for delivering wildfire warning information

Maps are a sensible approach for communicating wildfire early warnings to the public as such warnings often contain a multitude of spatial information. However, a reluctance of agencies was found in using accurate and timely wildfire maps for public warnings, a sentiment potentially fuelled by beliefs that the public are not fluent map-readers and may be overwhelmed by the large amount of information. To test the validity of these beliefs, this study empirically compared the effectiveness of maps versus traditional text-based approaches for communicating spatial-related wildfire warning information. Through an online survey, 261 residents from wildfire prone areas in Western Australia were asked to view multidimensional spatial information regarding a simulated wildfire scenario presented as either text messages or maps, and were subsequently queried for their comprehension, their risk perceptions, and the attractiveness of the presentation format. Additionally, the survey captured the time required to interpret the varied information representations. The results showed that appropriately designed maps prevailed over text messages for the communication of most wildfire warning information by improving comprehension, elevating risk perceptions, and increasing appeal to the public. However, an optimal communication approach would be to couple map designs with several imperative textual descriptors. Especially, the textual description of safe shelters in the community (i.e. location names and addresses) yielded indispensible meaning when the locations were well-known landmarks, and hence should not be replaced by map-based depiction. Furthermore, several heuristics were identified to facilitate the design of effective warning maps across hazards in general.

Increased PM Concentrations during a Combined Wildfire and Saharan Dust Event Observed at High-Altitude Sonnblick Observatory, Austria

A period of increased particulate matter concentrations was observed at the high-altitude Sonnblick Observatory in August 2013. Trajectory analysis, wildfire maps and the evaluation of aerosol measurements revealed a combined and sometimes alternating influence of long-range transport of Saharan dust and emissions of wildfires. The occurrence of Saharan dust was confirmed by an increase of coarse particle number concentration and a negative exponent of the single scattering albedo wavelength dependence, determined by Nephelometer and Aethalometer measurements. During time periods less influenced by Saharan dust, number concentration of accumulation mode particles increased and a marked correlation of aerosol mass concentrations and CO mixing ratios was observed. By analyzing the wavelength dependence of the absorption coefficients determined with a seven wavelength Aethalometer, the influence of the two aerosol sources was decoupled. Therefore, absorption exponents of 3 and 1.3 were assumed for Saharan dust and wildfires, respectively. Mass concentrations of particulate matter caused by Saharan dust and wildfire emissions were estimated, with the contribution of Saharan dust to overall particulate matter mass ranging from 5% to 80%.

Black carbon aerosols in urban central China

The first ever (to our knowledge), year-round measurements of Black Carbon (BC) aerosols in Hefei, an urban site of central China, from June 2012 to May 2013 are performed. The aim of this paper is to evaluate the black carbon in Hefei in terms of seasonal, monthly and diurnal variations, including their source identification. The annual mean BC mass concentration MBC is found to be 3.5±2.5μgm−3 in Hefei, while the aerosol optical depth shows a yearly average value of ~0.6. The seasonality of MBC depicts minimum values in the summer, moderate levels in the spring and fall, and maximum in the winter. The monthly average values of MBC vary threefold, ranging from the lowest average value of 2.0±1.0μgm−3 in July to the highest 6.0±2.6μgm−3 during January. Diurnal variations exhibit two BC peaks, corresponding to the morning and evening rush hours. Higher median BC concentrations are observed during haze episodes compared with non-haze periods, although low MBC is sometimes observed for high visibility, which is probably indicative of the aerosol scattering dominating diminished visibility. Based on trajectory analyses, the haze BC pollutions are mostly classified into three types from local areas, long-range transport from the Yangtze Delta, and transport from the North China Plain. The median MBC values for haze groups attributed to biomass burning from MODIS wildfire maps are higher than related groups that are not, which is indicative of the significant enhancement of BC aerosols due to agricultural biomass burning. The study suggests that aerosol absorption contributes more to the observed haze episodes in fall compared to other seasons.

LOW COST SURVEYING USING AN UNMANNED AERIAL VEHICLE

Abstract. Traditional manned airborne surveys are usually expensive and the resolution of the acquired images is often limited. The main advantage of Unmanned Aerial Vehicle (UAV) system acting as a photogrammetric sensor platform over more traditional manned airborne system is the high flexibility that allows image acquisition from unconventional viewpoints, the low cost in comparison with classical aerial photogrammetry and the high resolution images obtained. Nowadays there is a necessity for surveying small areas and in these cases, it is not economical the use of normal large format aerial or metric cameras to acquire aerial photos, therefore, the use of UAV platforms can be very suitable. Also the large availability of digital cameras has strongly enhanced the capabilities of UAVs. The use of digital non metric cameras together with the UAV could be used for multiple applications such as aerial surveys, GIS, wildfire mapping, stability of landslides, crop monitoring, etc. The aim of this work was to develop a low cost and accurate methodology in the production of orthophotos and Digital Elevation Models (DEM). The study was conducted in the province of Almeria, south of Spain. The photogrammetric flight had an altitude of 50 m over ground, covering an area of 5.000 m2 approximately. The UAV used in this work was the md4-200, which is an electronic battery powered quadrocopter UAV developed by Microdrones GmbH, Germany. It had on-board a Pextax Optio A40 digital non metric camera with 12 Megapixels. It features a 3x optical zoom lens with a focal range covering angles of view equivalent to those of 37–111 mm lens in 35 mm format. The quadrocopter can be programmed to follow a route defined by several waypoints and actions and it has the ability for vertical take off and landing. Proper flight geometry during image acquisition is essential in order to minimize the number of photographs, avoid areas without a good coverage and make the overlaps homogeneous. The flight planning was done using the MdCockpit software, with the module waypoint editor. Flight route file was downloaded into the quadrocopter autonomous chip via cable. A total of twelve vertical images with a longitudinal and transversal overlapping of 60% and 50% respectively were taken. The digital camera was previously geometrically calibrated. Field control points covering the whole studied area were defined over the area of interest and their coordinates were measured by a GPS. Natural targets were used as field control points. The close range photogrammetric software Photomodeler Scanner v.7 was used in this work to calibrate the camera and to carry out the photogrammetric process. The software Golden Surfer was used to produce the DEM. The planimetric and the altimetric root mean square error (RMSE) were calculated in order to check the accuracy of the products. The RMSEx was 6 cm, the RMSEy was 4 cm and the RMSEy was 7 cm. Our preliminary results demonstrate the feasibility and accuracy of orthophotos and DEMs obtained from images captured from a quadrocopter using low cost photogrammetric software. A future work can be the comparison of the products obtained following the route used in this study where the images are taken vertically with the products obtained with an orbital route where the number of images will be diminished and the photos will be taken oblique.

Haze types in Beijing and the influence of agricultural biomass burning

Abstract. Emissions from agricultural biomass burning (ABB) in northern China have a significant impact on the regional and global climate. The monthly average aerosol optical depth (AOD) at 550 nm in northern China in 2007 had a maximum of 0.7 in June. The AOD measurements are consistent with regional brown hazes that occurred at that time, which was a period of severe aerosol pollution. Aerosol particles were collected in urban Beijing from 12 to 30 June 2007, during a period of high haze, and studied using transmission electron microscopy with energy-dispersive X-ray spectrometry. The dominant particle types collected in the fine fraction (diameter <1 μm) were ammonium sulfate, soot, K2SO4, KNO3, and organic matter, except that the K salts were minor between 21 and 30 June. K-rich particles as tracers of biomass burning, together with wildfire maps, show that intense regional ABB in northern China contributed significantly to the regional haze between 12 and 20 June. We therefore grouped the episodes into type-1 and -2 haze, with the former occurring between 12 and 20 June and the latter between 21 and 30 June. After long-range transport, ABB particles in the type-1 haze exhibited marked changes in morphology, composition, and mixing state. KCl particles were absent, presumably having been converted by heterogeneous reactions to K2SO4 and KNO3. Soot particles were mixed with the other particle types. Abundant organic matter and soluble salts emitted by ABB increased their sizes during transport and resulted in more hygroscopic aerosol particles in downwind areas, becoming additional cloud condensation nuclei. The high AOD (average value 2.2) in Beijing during 12 to 20 June is partly explained by the hygroscopic growth of fine aerosol particles and by the strong absorption of internally mixed soot particles, both coming from regional ABB emissions. Therefore, it is important to consider the origins of the haze, which in turn leads to the different particle types.

Assessing Accuracy of Manually-mapped Wildfire Perimeters in Topographically Dissected Areas

Accurate mapping of wildfires is critical to fire management. Technological advances in remote sensing and Geographic Information Systems (GIS) over the last decade have been widely incorporated into wildfire mapping and management, but neither have been assessed for accuracy nor compared to established manual methods. Since Landsat-based mapping of wildfires will soon replace manual mapping methods, this type of comparison is critical to understanding the strengths of each method. Landsat ETM+ imagery was classified to create fire perimeter maps for 53 fires in Nevada, USA. These maps were then assessed for agreement with published, manually mapped fire perimeters. Published perimeters were found to correlate poorly to remotely sensed fire perimeters, and significantly overestimated area burned (p ≤ 0.05) by an average of 18 percent. Mapping disagreement was then correlated to a measure of topographic roughness at four spatial scales to determine whether increasing terrain complexity was a factor in increased disagreement. Mapping disagreement showed a significant positive correlation (r = 0.57) to topographic roughness. For fire research spanning multiple decades, these results indicate that it may be difficult to utilize fire perimeter data sets comprising both satellite-derived and manually mapped perimeters because the two data sets are significantly different.

Regionally adaptable dNBR-based algorithm for burned area mapping from MODIS data

Recent advances in instrument design have led to considerable improvements in wildfire mapping at regional and global scales. Global and regional active fire and burned area products are currently available from various satellite sensors. While only global products can provide consistent assessments of fire activity at the global, hemispherical or continental scales, the efficiency of their performance differs in various ecosystems. The available regional products are hard-coded to the specifics of a given ecosystem (e.g. boreal forest) and their mapping accuracy drops dramatically outside the intended area. We present a regionally adaptable semi-automated approach to mapping burned area using Moderate Resolution Imaging Spectroradiometer (MODIS) data. This is a flexible remote sensing/GIS-based algorithm which allows for easy modification of algorithm parameterization to adapt it to the regional specifics of fire occurrence in the biome or region of interest. The algorithm is based on Normalized Burned Ratio differencing (dNBR) and therefore retains the variability of spectral response of the area affected by fire and has the potential to be used beyond binary burned/unburned mapping for the first-order characterization of fire impacts from remotely sensed data. The algorithm inputs the MODIS Surface Reflectance 8-Day Composite product (MOD09A1) and the MODIS Active Fire product (MOD14) and outputs yearly maps of burned area with dNBR values and beginning and ending dates of mapping as the attributive information. Comparison of this product with high resolution burn scar information from Landsat ETM+ imagery and fire perimeter data shows high levels of accuracy in reporting burned area across different ecosystems. We evaluated algorithm performance in boreal forests of Central Siberia, Mediterranean-type ecosystems of California, and sagebrush steppe of the Great Basin region of the US. In each ecosystem the MODIS burned area estimates were within 15% of the estimates produced by the high resolution base with the R 2 between 0.87 and 0.99. In addition, the spatial accuracy of large burn scars in the boreal forests of Central Siberia was also high with Kappa values ranging between 0.76 and 0.79.

Evaluation of Wildfire Mapping with NOAA/AVHRR Data by Land Cover Types and Eco-Regions in California

In order to determine the efficacy of archived remotely sensed data to create historic fire maps, this study compares a remote sensing based wildland fire map with an historical fire database for California, USA. Fires occurring in two years (1996 and 1999) were mapped using data obtained from the NOAA Advanced Very High Resolution Radiometer (AVHRR) sensor. A vector database of fire perimeters, compiled and maintained by the State of California was obtained as a source of comparison for the AVHRR based fire map. The two datasets were overlaid and spatially compared in seven land cover types and ten ecoregions. The sources of disagreement and overlap between the datasets were quantified in order to elucidate trends in fire detection algorithm performance over the land cover types and ecoregions. Various methods of vector based comparison were examined in order to more effectively describe the spatial relationships between the two fire maps. The results show that algorithm performance varies over both ecosystem type and geographic region of California. The remote sensing method was able to map between 62% and 74% (1999 and 1996, respectively) of fire area mapped in the State maintained database. There was between 40% and 45% (1999 and 1996, respectively) of geographic overlap in the datasets. The results illustrate the need to calibrate remotely sensed algorithms by ecosystem type and geographic location in order to more effectively produce historic fire map products for research and other purposes.

A dynamic algorithm for wildfire mapping with NOAA/AVHRR data

A wildfire-mapping algorithm is proposed based on fire dynamics, called the dynamic algorithm. It is applied to daily NOAA/AVHRR/HRPT data for wildland areas (scrub, chaparral, grassland, marsh, riparian forest, woodland, rangeland and forests) in California for September and October 1999. Daily AVHRR images acquired from two successive days are compared for active fire detection and burn scar mapping. The algorithm consists of four stages: data preparation; hotspot detection; burn scar mapping; and final confirmation of potential burn scar pixels. Preliminary comparisons between the result mapped by the dynamic algorithm and the fire polygons collected by the California Department of Forestry and Fire Protection through ground survey indicate that the algorithm can track burn scars at different developmental stages at a daily level. The comparisons between wildfire mapping results produced by a modified version of an existing algorithm and the dynamic algorithm also indicate this point. This is the major contribution of this algorithm to wildfire detection methods. The dynamic algorithm requires highly precise registration between consecutive images.

Mapping Fire-Induced Vegetation Mortality Using Landsat Thematic Mapper Data: A Comparison of Linear Transformation Techniques

Forests in the U.S. southwest experience large, intense wildfires. Fire severity maps can assist management of such fire-scarred landscapes. Remote sensing appears suitable for wildfire mapping, provided data have sufficient spatial, radiometric, and spectral resolutions. Using a 1995 Thematic Mapper (TM) post-fire scene of the 8900 ha Rattlesnake Fire in southeastern Arizona as a case study, two linear transformation techniques, the Kauth–Thomas (KT) and principal components analysis (PC) transforms were invoked to enhance Thematic Mapper data prior to supervised classification. The KT and PC transformations were selected to enhance fire-related brightness, greenness, and wetness variations in the image, detecting the extent of different fire severities. The KT transform produced 17% higher overall classification accuracies than the PC transform. The higher accuracy recorded by the KT transform results from brightness, greenness, and wetness variations which, in this case, are associated with fire severity.