Burned Area Products Research Articles

Dynamic prediction of global monthly burned area with hybrid deep neural networks.

Wildfires not only severely damage the natural environment and global ecological balance but also cause substantial losses to global forest resources and human lives and property. Unprecedented fire events such as Australia's bushfires have alerted us to the fact that wildfire prediction is a critical scientific problem for fire management. Therefore, robust, long-lead models and dynamic predictions of wildfire are valuable for global fire prevention. However, despite decades of effort, the dynamic, effective, and accurate prediction of wildfire remains problematic. There is great uncertainty in predicting the future based on historical and existing spatiotemporal sequence data, but with advances in deep learning algorithms, solutions to prediction problems are being developed. Here, we present a dynamic prediction model of global burned area of wildfire employing a deep neural network (DNN) approach that produces effective wildfire forecasts based on historical time series predictors and satellite-based burned area products. A hybrid DNN that combines long short-term memory and a two-dimensional convolutional neural network (CNN2D-LSTM) was proposed, and CNN2D-LSTM model candidates with four different architectures were designed and compared to construct the optimal architecture for fire prediction. The proposed model was also shown to outperform convolutional neural networks (CNNs) and the fully connected long short-term memory (FcLSTM) approach using the refined index of agreement and evaluation metrics. We produced monthly global burned area spatiotemporal prediction maps and adequately reflected the seasonal peak in fire activity and highly fire-prone areas. Our combined CNN2D-LSTM approach can effectively predict the global burned area of wildfires 1 month in advance and can be generalized to provide seasonal estimates of global fire risk.

Conjunctive Use of Landsat-8 OLI and MODIS Data for Delineation of Burned Areas

For regional-level monitoring of burned areas, Moderate Resolution Imaging Spectroradiometer (MODIS ) MCD64A1 and MCD45A1 products have been operationally used. However, because of their coarser spatial resolution, such products do not allow for detection of small patches (<50 ha) of burned areas, which are very important for modeling gas emissions. In order to bridge this gap, we undertook a study to evaluate the synergy of MCD64A1 and Landsat-8 Operational Land Imager (OLI ) data for delineating burned areas in part of the mountainous terrain of the Himalayas, northern India. The approach involves generating a differenced normalized burn ratio image from Landsat-8 OLI data before and after fire events, refining the burned areas thus delineated using the MCD64A1 burned-area product and field data, and estimating accuracy. An overall mapping accuracy of 90.0% was achieved using Landsat-8 OLI data. Inclusion of the MODIS MCD64A1 burned-area product resulted in marginal improvement in overall accuracy (to 94.0%).

How Much of a Pixel Needs to Burn to Be Detected by Satellites? A Spectral Modeling Experiment Based on Ecosystem Data from Yellowstone National Park, USA

We present a simple modeling technique based on linear spectral mixture analysis to assess satellite detectability of sub-pixel burned area. Pixel observations are modeled using a linear combination of pure land covers, called endmembers. We executed an experiment using spectral data from Yellowstone National Park, USA. Using endmember samples from spectral libraries, pixel samples were assessed on burn detectability using the widely used differenced Normalized Burn Ratio (dNBR). While individual samples yielded differing results for Landsat 8, Sentinel-2, and the Moderate Resolution Imaging Spectroradiometer (MODIS), the average park-wide detectability of burned area was consistent across satellites. For the commonly used dNBR threshold of 0.15, the results indicated that detectability is reached when around a quarter of a pixel’s area is burned. However, a significant percentage of the modeled burned pixels remained undetectable, especially those with low pre-fire vegetation cover. This has consequences for burned area estimates, as smaller fires in sparsely vegetated terrain may remain undetected in moderate resolution burned area products.

Reply to Giglio et al. Comment on “Otón et al. Analysis of Trends in the FireCCI Global Long Term Burned Area Product (1982–2018). Fire 2021, 4, 74”

First of all, we would like to thank the authors of the comment [...]

Evaluation of low-resolution remotely sensed datasets for burned area assessment within the wildland-urban interface

The impacts of wildfires are often intensified within the wildland-urban interface (WUI) where built structures intermingle with wildland vegetation. Because fire impacts in the WUI are high, detailed fire occurrence information is especially valuable to fire scientists and risk managers. While it is known that burned area datasets with coarse spatial resolution frequently underestimate burned area due to the omission of small fires, the extent to which they do so has not been quantified in states such as Oklahoma in the south-central United States. Here, we explore how much burned area low-resolution datasets miss, where they miss burned area, and how different datasets detect burned area within the wildland-urban interface in Oklahoma, USA. We compare the MODIS MCD64A1 burned area product, the Monitoring Trends in Burn Severity (MTBS) product and a higher-resolution dataset developed using Sentinel-2 imagery and find that the low-resolution datasets underestimate burned area by approximately 46% and were unable to detect major hotspots of fire occurrence. Overall, our study provides an estimate of the extent to which large-scale, low-resolution burned area datasets underestimate the number and distribution of small fires in Oklahoma.

Detecting and quantifying residue burning in smallholder systems: An integrated approach using Sentinel-2 data

• Smallholder agricultural emissions in Central India are largely unaccounted. • A method is devised to detect smallholder’s burned area using Sentinel-2 data. • Agricultural emissions are estimated using burned area derived at 10 m resolution. • A novel visual image interpretation is applied to collect ground truth data. • Our approach showed significant smallholder (<10 ha) emissions than global products. Accurate and rapid mapping of agriculture burned areas (ABA) is essential to track agricultural emissions for sustainable agronomic development, better air quality management, and mitigation of climate change. Despite its climate and environmental impacts, India is largely unexplored considering the magnitude and frequency of agricultural residue burning. This study focuses on Central India and explores the potential of Sentinel-2 (10 m) multispectral data to accurately estimate ABA, which can be used to calculate associated emissions. We used a locally adapted multi-temporal burned area detection methodology to quantify burned areas after the winter crop harvest in Central India from 2019 to 2021. A novel satellite data assisted virtual sampling method was used to collect burned and unburned training samples. Monthly ABAs were extracted at 10 m resolution and were compared with existing global burned area products. On average, the derived ABAs were larger (∼9179 km 2 per month) than those reported by global burned area products that are based on MODIS at 500 m and 250 m resolutions. We show that our approach maps ABA from smallholder farms more accurately than other global products, particularly for fields that are smaller than 10 ha. The derived ABAs were further used to estimate emissions due to agricultural residue burning, which can be used as an initial estimate of potential emissions due to ABA in Central India. Overall, the study demonstrates the ability of Sentinel-2 data to detect and quantify ABA in smallholder systems, which can be used to more accurately estimate emissions compared to existing global products.

Wildland urban interface of the City of Cape Town 1990–2019

AbstractWildfire affecting urban areas at the wildland urban interface (WUI) is a growing global concern, where management is important for urban residents as well as wildland vegetation. We used a socio‐ecological system perspective to investigate the interactions of urban land with a fire‐dependent wildland in South Africa’s City of Cape Town (CoCT). We examined changes in population growth, land cover change and related WUI footprint, occurrence of large fires, and related policies over time. We used Landsat data to track changes over the period 1990–2019 in the formal and informal urban and wildland footprint, census data to track changes in population, and difference normalised burn ratio and MODIS burned area product to track large fires. The urban footprint has expanded greatly and through consolidation has led to the reduction of the WUI. Furthermore, evidence points to an increase in fire suppression, even though national policies ask for wildfires to run their course where possible and appropriate. As a result of pressure from urban residents, local managers prefer short term fire suppression to long term risk reduction for urban areas and ecological management of wildland. Framing the problem as a socio‐ecological system enabled us to highlight how WUI management is a product of interaction between urban development, wildland type and policies. Our findings emphasise the point that wildland management is driven by urban residents and local municipalities, with national fire and disaster management policies not fully implemented.

Comparing the Ability of Burned Area Products to Detect Crop Residue Burning in China

Burning crop residues is a common way to remove them during the final stages of crop ripening in China. To conduct research effectively, it is critical to reliably and quantitatively estimate the extent and location of a burned area. Here, we investigated three publicly available burned area products—MCD64A1, FireCCI 5.1, and the Copernicus Burnt Area—and evaluated their relative performance at estimating total burned areas for cropland regions in China between 2015 and 2019. We compared these burned area products at a fine spatial and temporal scale using a grid system comprised of three-dimensional cells spanning both space and time. In general, the Copernicus Burnt Area product detected the largest annual average burned area (37,095.1 km2), followed by MCD64A1 (21,631.4 km2) and FireCCI 5.1 (12,547.99 km2). The Copernicus Burnt Area product showed a consistent pattern of monthly burned areas during the study period, whereas MCD64A1 and FireCCI 5.1 showed frequent changes in monthly burned area peaks. The greatest spatial differences between all three products occurred in Northeast and North China, where cultivated land is concentrated. The burned area detected by Copernicus in Xinjiang Province was larger than that detected by the other two products. In conclusion, we found that all three products underestimated the amount of crop residues present in a burned area. This limits the ability of end users to understand fire-related impacts and burned area characteristics, and hinders them in making an informed choice of which product is most appropriate for their application.

Assessing the Accuracy of MODIS MCD64A1 C6 and FireCCI51 Burned Area Products in Mediterranean Ecosystems

The catastrophic impact of wildfires on the economy and ecosystems of Mediterranean countries in recent years, along with insufficient policies that favor disproportionally high funding for fire suppression, demand a more comprehensive understanding of fire regimes. Satellite remote sensing products support the generation of relevant burned-area (BA) information, since they provide the means for the systematic monitoring of large areas worldwide at low cost. This research study assesses the accuracy of the two publicly available MODIS BA products, MCD64A1 C6 and FireCCI51, at a national scale in a Mediterranean country. The research period covered four fire seasons, and a comparison was conducted against a higher-resolution Sentinel-2 dataset. The specific objectives were to assess their performance in detecting fire events occurring primarily in forest and semi-natural lands and to investigate their spatial and temporal uncertainties. Monthly fire observations were processed and analyzed to derive a comprehensive set of accuracy metrics. We found that fire size has an impact on their detection accuracy, with higher detection occurring in fires larger than 100 ha. Detection of smaller (&lt;100 ha) fires was favored by the 250 m FireCCI51 product, but not from MCD64A1 C6, which exhibited less than 50% detection probability in the same range. Their spatial estimates of burned area exhibited a fairly satisfactory agreement with the reference data, reaching an average of 78% in detection rate. MCD64A1 C6 exhibited a more consistent spatial performance overall and better temporal accuracy, whereas FireCCI51 did not substantially outperform the former despite its finer resolution. Additional research is required for a more rigorous assessment of the variability of these burned area products, yet this research provides further insight and has implications for their use in fire-related applications at the local to the national scale.

Contemporary (1984–2020) fire history metrics for the conterminous United States and ecoregional differences by land ownership

Background Remotely sensed burned area products are critical to support fire modelling, policy, and management but often require further processing before use. Aim We calculated fire history metrics from the Landsat Burned Area Product (1984–2020) across the conterminous U.S. (CONUS) including (1) fire frequency, (2) time since last burn (TSLB), (3) year of last burn, (4) longest fire-free interval, (5) average fire interval length, and (6) contemporary fire return interval (cFRI). Methods Metrics were summarised by ecoregion and land ownership, and related to historical and cheatgrass datasets to demonstrate further applications of the products. Key results The proportion burned ranged from 0.7% in the Northeast Mixed Woods to 74.1% in the Kansas Flint Hills. The Flint Hills and Temperate Prairies showed the highest burn frequency, while the Flint Hills and the Sierra Nevada and Klamath Mountains showed the shortest TSLB. Compared to private, public land had greater burned area (19 of 31 ecoregions) and shorter cFRI (25 of 31 ecoregions). Conclusions Contemporary fire history metrics can help characterise recent fire regimes across CONUS. Implications In regions with frequent fire, comparison of contemporary with target fire regimes or invasive species datasets enables the efficient incorporation of burned area data into decision-making.

Agricultural burned area detection using an integrated approach utilizing multi spectral instrument based fire and vegetation indices from Sentinel-2 satellite

This study presents a methodology that focuses on detecting agricultural burned areas using Sentinel-2 multispectral data at 10 m. We developed a simple, locally adapted, straightforward approach of multi-index threshold to extract post-winter agricultural burned areas at high resolution for 2019-21. Further, we design a new method for virtual sample collection using already validated fire location data and visual interpretation conditioned using strict selection criteria to improve sample accuracy. Sampling accuracy showed near-perfect agreement with an average Cohen's Kappa value of 0.98. We retrieved monthly ABAs at a resolution of 10 m, and these products were validated against reference burned sample plots identified using visual interpretation of Planet (3m) satellite data. Overall, we found that our method performed well, with an F1 score of 83.63% and low commission (20%) and omission (7%) errors. When compared to global burnt area products, validation accuracy demonstrated an exceptional subpixel scale detecting capability. The study also addresses the complexity of residue burnings and burn signatures’ volatile nature by performing multilevel masking and temporal corrections.•A novel remotely sensed data aided virtual sampling approach to acquire burned and unburned samples.•An integrated method to extract smallholder agricultural burned area using Sentinel-2 multispectral data at a high resolution of 10 m

Mapping Cropland Burned Area in Northeastern China by Integrating Landsat Time Series and Multi-Harmonic Model

Accurate cropland burned area estimation is crucial for air quality modeling and cropland management. However, current global burned area products have been primarily derived from coarse spatial resolution images which cannot fulfill the spatial requirement for fire monitoring at local levels. In addition, there is an overall lack of accurate cropland straw burning identification approaches at high temporal and spatial resolution. In this study, we propose a novel algorithm to capture burned area in croplands using dense Landsat time series image stacks. Cropland burning shows a short-term seasonal variation and a long-term dynamic trend, so a multi-harmonic model is applied to characterize fire dynamics in cropland areas. By assessing a time series of the Burned Area Index (BAI), our algorithm detects all potential burned areas in croplands. A land cover mask is used on the primary burned area map to remove false detections, and the spatial information with a moving window based on a majority vote is employed to further reduce salt-and-pepper noise and improve the mapping accuracy. Compared with the accuracy of 67.3% of MODIS products and that of 68.5% of Global Annual Burned Area Map (GABAM) products, a superior overall accuracy of 92.9% was obtained by our algorithm using Landsat time series and multi-harmonic model. Our approach represents a flexible and robust way of detecting straw burning in complex agriculture landscapes. In future studies, the effectiveness of combining different spectral indices and satellite images can be further investigated.

Global intercomparison of functional pyrodiversity from two satellite sensors

ABSTRACT Fires create substantial ecosystem disturbances, which affect biodiversity and climate processes. Previously, satellite-derived burned areas were assessed, and fire patches and morphological traits related to fire spread functions identified. The cut-off value in days is a critical parameter for the algorithms used to identify fire patches, as it identifies when two contiguous burned pixels should be grouped together as one patch. Our understanding of how fire patches and their morphologies change with the cut-off values between satellite sensors is currently insufficient for model development and pyrogeographic studies. We have addressed this using the FRY v1.0 global database of fire patch functional traits, with cut-off values of 3, 5, 9, and 14 days, derived from two burned area products, the MERIS Fire_cci version 4.1 and MODIS MCD64A1 collection 6, from two satellite sensors with different spatial and temporal resolutions. We intercompared the FRY products with functional diversity trait methods that assess their morphological pyrodiversity structures and dissimilarities. We have assessed the functional pyrodiversity of the FRY products with four cut-off values, for fire patches ≥ 300 ha, on 1° resolution tiles for the world and Global Fire Emissions Database (GFED) region scales, from 2005–2011. The functional structures and dissimilarities between the FRY products varied with the regions, while cut-off value sensitivity varied with the regions and between the satellite sensor products. These variations in functional structures and dissimilarities mirrored the regional fire spread processes and satellite sensor resolutions. The functional pyrodiversity that varied with the regions and similar cut-off values did not lead to the best agreement between the satellite sensor products. These variations reinforce the need to combine high temporal resolution optical sensors with radar in tropical regions, to consistently meet the monitoring requirements and for international organizations, to enable improved assessments of climate and biodiversity.

Accuracy estimation of two global burned area products at national scale

In this work we perform an initial assessment of the accuracy of two publicly available MODIS burned area products, MCD64A1 C6 and MODIS FireCCI51, at national scale in a Mediterranean region. The research focused on two fire seasons for the years 2016 and 2017 and comparison was performed against a higher resolution Sentinel-2 dataset. The specific objectives were to assess their capabilities in detection of fire events occurring primarily in forest and semi-natural lands and also to investigate their spatial uncertainties. The analysis combined monthly fire observations and accuracy metrics derived from error matrices. Satisfactory performance was achieved by the two products in detection of larger fires (> 100 ha) whereas their spatial performance exhibited good agreement with the reference data. MCD64A1 C6 exhibited a more consistent performance overall and the 250 m FireCCI51 product exhibited relatively higher sensitivity in detection of smaller (<100 ha) fires. Although additional work is required for a more rigorous assessment of the variability of these burned area products, our research has implications for their usability in fire-related applications at finer scales.

Investigating El Nino Southern Oscillation as the main driver of forest fire in Kalimantan

Strong El Nino events have been identified as major factors contributing to the forest fire in Indonesia. This is due to El Nino Southern Oscillation (ENSO) variability has a distinct connection with tropical precipitation, mainly in Indonesia. El Nino years are typically drier, while La Nina events generally increase precipitation in Indonesia. Although very strong El Nino events have been connected with massive forest fires, fire continue to occur during the other phases of ENSO: La Nina and normal. Here, the research reports a time series of monthly counted fires in Kalimantan between the period 2001-2020 from MODIS fire hotspot and MODIS Burned Area products. The region is divided into three categories, Primary Intact Forest, Primary Degraded Forest and Outside Forest/ Deforested area. This categorization validates the location of the fire. Our results show that in general wildfires in Kalimantan follow a similar temporal pattern with Oceanic Nino Index (ONI), with several anomalies. If ONI is high, wildfires are more intense and vice versa. The wildfire appears almost every month and increases drastically in June-October of El Nino years. However, the proportion of wildfires in Primary Intact Forest are tiny and insignificant. The primary intact forest fire only appears in July-October and have a different pattern with wildfires in general. In conclusion, wildfires are highly correlated with El Nino but limited in primary intact forests. The fires dominantly appear in the deforested area, about 80%. The rest 20% are in degraded forest and only less than 1% in primary intact forest. Keywords: Deforested, degraded forest, ENSO, forest fire, ONI, primary intact forest

Using long temporal reference units to assess the spatial accuracy of global satellite-derived burned area products

In recent years, the growing availability of global satellite-derived burned area (BA) products has led to the development of methods and protocols to rigorously estimate their accuracy metrics. These protocols are based on design-based inference and provide unbiased estimators of various dimensions of accuracy. Current procedures consider the spatial and temporal dimension when obtaining the independent reference data used to assess accuracy, commonly based on the Landsat imagery archive as the basic source. The protocol in which the temporal dimension is addressed in the reference data impacts the accuracy metrics. For example, the 8–16-day sampling units usually recommended in Stage 3 BA validation protocols may result in confounding of spatial and temporal classification errors. However, both errors have different implications from a user's perspective, depending on whether the spatial detection or the temporal dating are relevant. While maintaining the fundamentals of current validation protocols, this study presents a new approach based on long temporal reference units (> 48 days) to diminish the influence of temporal reporting (i.e., dating errors) on the spatial accuracy estimates. This methodology is applied to estimate the accuracy of several global BA products for the period 2017–2019, including two European BA products, the FireCCI51 and C3SBA10, and NASA's standard BA product, the MCD64A1 collection 6 (MCD64C6). Global estimates showed similar performance for the three products; BA commission errors ranged from 17.2% ± 1.1% for C3SBA10 to 19.4% ± 1.1% for FireCCI51, and BA omission errors ranged from 43.1% ± 1.9% for FireCCI51 to 49.3% ± 2.2% for MCD64C6 (± values are one standard error). The total burned area was consistently underestimated in all products. These errors are much lower relative to those obtained in recent Stage 3 validation exercises based on short temporal reference units, which estimated global commission and omission errors greater than 40% and 70%, respectively. Thus, this study demonstrates that using long reference units provides a method to address the impact of BA product dating errors on estimates of spatial accuracy metrics, particularly for those products with lower temporal resolution or for areas with greater cloud coverage. Validation methods developed in this study may contribute to improving future protocols adopted by the Committee on Earth Observation Satellites (CEOS) Land Product Validation (LPV) subgroup.

A space view of agricultural and industrial changes during the Syrian civil war

The agricultural sector in Syria was heavily affected by the civil war that started in 2011. We investigate the war’s impact on the country’s atmospheric ammonia (NH3) from 2008 to 2019, using measurements from the infrared atmospheric sounding interferometer instrument on board the Metop satellites. We examine the changes in NH3 close to a fertilizer industry, whose activities were suspended due to conflict-related events. We also explore the effect of war-induced land use/land cover changes on agriculture-emitted ammonia in north-east Syria that has witnessed battles between different groups. The interpretation of the changes in NH3 is supported by different datasets: visible satellite imagery to assess the effect on industrial activity, reanalysis data from the European center for medium-range weather forecasts to look at the effect of meteorology (temperature, wind speed, and precipitation), and land cover and burned area products from the moderate resolution imaging spectroradiometer (MODIS) to examine land use/land cover changes and fire events during the study period. We show that the NH3 columns are directly affected by the war. Periods of intense conflict are reflected in lower values over the industry reaching –17%, –47%, and –32% in 2013, 2014, and 2016, respectively, compared to the [2008–2012] average, and a decrease reaching –14% and –15% in the croplands’ area in northeast Syria during 2017 and 2018 (compared to 2011), respectively. Toward the end of the control of Islamic State in Iraq and Syria, an increase in atmospheric NH3 was accompanied by an increase in croplands’ area that reached up to +35% in 2019 as compared to prewar (2011). This study shows the relevance of remote-sensing data of atmospheric composition in studying societal changes at a local and regional scale.

Implementation of Fire Policies in Brazil: An Assessment of Fire Dynamics in Brazilian Savanna

In 2012, the Brazilian government implemented the Federal Brigades Program (FBP), a fire policy strategy to hire and train firefighters to combat wildfires. This study analyzed the impact of this program on fire behavior before (2008–2012) and after (2013–2017) its implementation in the Parque do Araguaia Indigenous Land, the largest indigenous territory with the highest occurrence of fires in the Brazilian tropical savanna. We analyzed the annual pattern of fire incidence in the dry season, the fire impact per vegetation type, the recurrence, and the relationship between fire and precipitation. The datasets were based on active fire products derived from the Moderate Resolution Imaging Spectroradiometer (MODIS), the Landsat and Resourcesat-based burned area products, and the records of the fire combat operations. Our results showed that FBP contributed to the reduction of the number of areas affected by fires and to the formation of a more heterogeneous environment composed of fire-resistant and fire-sensitive native vegetation fragments. On the other hand, after the implementation of the FBP, there was an increase in the recurrence of 3–4 years of fires. We concluded that the FBP is an important public policy capable of providing improvements in fire management activities.

Analysis of Trends in the FireCCI Global Long Term Burned Area Product (1982–2018)

We present an analysis of the spatio-temporal trends derived from long-term burned area (BA) data series. Two global BA products were included in our analysis, the FireCCI51 (2001–2019) and the FireCCILT11 (1982–2018) datasets. The former was generated from Moderate Resolution Imaging Spectroradiometer (MODIS) 250 m reflectance data, guided by 1 km active fires. The FireCCILT11 dataset was generated from Land Long-Term Data Record data (0.05°), which provides a consistent time series for Advanced Very High Resolution Radiometer images, acquired from the NOAA satellite series. FireCCILT11 is the longest time series of a BA product currently available, making it possible to carry out temporal analysis of long-term trends. Both products were developed under the FireCCI project of the European Space Agency. The two datasets were pre-processed to correct for temporal autocorrelation. Unburnable areas were removed and the lack of the FireCCILT11 data in 1994 was examined to evaluate the impact of this gap on the BA trends. An analysis and comparison between the two BA products was performed using a contextual approach. Results of the contextual Mann-Kendall analysis identified significant trends in both datasets, with very different regional values. The long-term series presented larger clusters than the short-term ones. Africa displayed significant decreasing trends in the short-term, and increasing trends in the long-term data series, except in the east. In the long-term series, Eastern Africa, boreal regions, Central Asia and South Australia showed large BA decrease clusters, and Western and Central Africa, South America, USA and North Australia presented BA increase clusters.

Missing Burns in the High Northern Latitudes: The Case for Regionally Focused Burned Area Products

Global estimates of burned areas, enabled by the wide-open access to the standard data products from the Moderate Resolution Imaging Spectroradiometer (MODIS), are heavily relied on by scientists and managers studying issues related to wildfire occurrence and its worldwide consequences. While these datasets, particularly the MODIS MCD64A1 product, have fundamentally improved our understanding of wildfire regimes at the global scale, their performance may be less reliable in certain regions due to a series of region- or ecosystem-specific challenges. Previous studies have indicated that global burned area products tend to underestimate the extent of the burned area within some parts of the boreal domain. Despite this, global products are still being regularly used by research activities and management efforts in the northern regions, likely due to a lack of understanding of the spatial scale of their Arctic-specific limitations, as well as an absence of more reliable alternative products. In this study, we evaluated the performance of two widely used global burned area products, MCD64A1 and FireCCI51, in the circumpolar boreal forests and tundra between 2001 and 2015. Our two-step evaluation shows that MCD64A1 has high commission and omission errors in mapping burned areas in the boreal forests and tundra regions in North America. The omission error overshadows the commission error, leading to MCD64A1 considerably underestimating burned areas in these high northern latitude domains. Based on our estimation, MCD64A1 missed nearly half the total burned areas in the Alaskan and Canadian boreal forests and the tundra during the 15-year period, amounting to an area (74,768 km2) that is equivalent to the land area of the United States state of South Carolina. While the FireCCI51 product performs much better than MCD64A1 in terms of commission error, we found that it also missed about 40% of burned areas in North America north of 60° N between 2001 and 2015. Our intercomparison of MCD64A1 and FireCCI51 with a regionally adapted MODIS-based Arctic Boreal Burned Area (ABBA) shows that the latter outperforms both MCD64A1 and FireCCI51 by a large margin, particularly in terms of omission error, and thus delivers a considerably more accurate and consistent estimate of fire activity in the high northern latitudes. Considering the fact that boreal forests and tundra represent the largest carbon pool on Earth and that wildfire is the dominant disturbance agent in these ecosystems, our study presents a strong case for regional burned area products like ABBA to be included in future Earth system models as the critical input for understanding wildfires’ impacts on global carbon cycling and energy budget.