Monitoring active fires in Borneo from Sentinel-2 MSI images

A preliminary evaluation of GOES-16 active fire product using Landsat-8 and VIIRS active fire data, and ground-based prescribed fire records

Validation of MCD64A1 and FireCCI51 cropland burned area mapping in Ukraine

Abstract. Characterising the dynamics of landscape-scale wildfires at very high temporal resolutions is best achieved using observations from Earth Observation (EO) sensors mounted onboard geostationary satellites. As a result, a number of operational active fire products have been developed from the data of such sensors. An example of which are the Fire Radiative Power (FRP) products, the FRP-PIXEL and FRP-GRID products, generated by the Land Surface Analysis Satellite Applications Facility (LSA SAF) from imagery collected by the Spinning Enhanced Visible and Infrared Imager (SEVIRI) onboard the Meteosat Second Generation (MSG) series of geostationary EO satellites. The processing chain developed to deliver these FRP products detects SEVIRI pixels containing actively burning fires and characterises their FRP output across four geographic regions covering Europe, part of South America and Northern and Southern Africa. The FRP-PIXEL product contains the highest spatial and temporal resolution FRP data set, whilst the FRP-GRID product contains a spatio-temporal summary that includes bias adjustments for cloud cover and the non-detection of low FRP fire pixels. Here we evaluate these two products against active fire data collected by the Moderate Resolution Imaging Spectroradiometer (MODIS) and compare the results to those for three alternative active fire products derived from SEVIRI imagery. The FRP-PIXEL product is shown to detect a substantially greater number of active fire pixels than do alternative SEVIRI-based products, and comparison to MODIS on a per-fire basis indicates a strong agreement and low bias in terms of FRP values. However, low FRP fire pixels remain undetected by SEVIRI, with errors of active fire pixel detection commission and omission compared to MODIS ranging between 9–13 % and 65–77 % respectively in Africa. Higher errors of omission result in greater underestimation of regional FRP totals relative to those derived from simultaneously collected MODIS data, ranging from 35 % over the Northern Africa region to 89 % over the European region. High errors of active fire omission and FRP underestimation are found over Europe and South America and result from SEVIRI's larger pixel area over these regions. An advantage of using FRP for characterising wildfire emissions is the ability to do so very frequently and in near-real time (NRT). To illustrate the potential of this approach, wildfire fuel consumption rates derived from the SEVIRI FRP-PIXEL product are used to characterise smoke emissions of the 2007 "mega-fire" event focused on Peloponnese (Greece) and used within the European Centre for Medium-Range Weather Forecasting (ECMWF) Integrated Forecasting System (IFS) as a demonstration of what can be achieved when using geostationary active fire data within the Copernicus Atmosphere Monitoring Service (CAMS). Qualitative comparison of the modelled smoke plumes with MODIS optical imagery illustrates that the model captures the temporal and spatial dynamics of the plume very well, and that high temporal resolution emissions estimates such as those available from a geostationary orbit are important for capturing the sub-daily variability in smoke plume parameters such as aerosol optical depth (AOD), which are increasingly less well resolved using daily or coarser temporal resolution emissions data sets. Quantitative comparison of modelled AOD with coincident MODIS and AERONET (Aerosol Robotic Network) AOD indicates that the former is overestimated by ~ 20–30 %, but captures the observed AOD dynamics with a high degree of fidelity. The case study highlights the potential of using geostationary FRP data to drive fire emissions estimates for use within atmospheric transport models such as those implemented in the Monitoring Atmospheric Composition and Climate (MACC) series of projects for the CAMS.

The Sea and Land Surface Temperature Radiometer (SLSTR) senses the Earth from onboard two concurrently operating European Copernicus Sentinel-3 (S3) satellites. As the Terra platform carrying the Moderate Resolution Imaging Spectroradiometer (MODIS) is reaching its end of life, the S3 Active Fire Detection and FRP products generated from data captured by S3 SLSTR are expected to soon become the main global active fire (AF) product for the mid-morning and evening low Earth orbit timeslots. The S3 night-time AF product issued by the European Space Agency (ESA) has been operational since March 2020, and here we report on the significant adjustments made to enable the generation of a complimentary daytime product. Similar to MODIS, SLSTR possesses two middle infrared channels, both a ‘standard’ (normal gain; S7) channel and a ‘fire’ (low-gain; F1) channel - but in contrast to MODIS by day even the ambient background land surface is often saturated in the SLSTR standard gain MIR (S7) channel. This saturation necessitates far greater use of the F1 channel data by day for active fire detection than by night, even though F1 has characteristics which make its data more challenging to combine with that from the other SLSTR thermal infrared channels. Here we report on the approaches used to combine S7 and F1 data for optimized daytime AF detection, and also detail the other algorithm adjustments found necessary to include in the daytime AF product algorithm. We compare the resulting daytime SLSTR AF product data to that generated from near-simultaneous views provided by MODIS onboard Terra. When both sensors detect the same active fire cluster at similar time, there is minimal bias shown between the two FRP retrievals (the ordinary least squares linear best fit between matched SLSTR and MODIS per-fire FRP matchups has a slope of 0.97). At the regional scale, the S3 product detects 70% of the AF pixels that the matching MODIS product reports, but also provides a further (16%) set of unique AF pixel detections. Regional FRP totals derived from SLSTR appear slightly lower than those from MODIS, and the OLS linear best fit between these regional FRP matchup datasets has a slope of 0.91. This is largely due to SLSTR performing less well in detecting the lowest FRP fires by day, whereas by night the S3 product performs a little better than MODIS due to the increased night-time use of S7 in the earlier AF pixel detection stages. Global fire mapping at a 0.25° grid cell resolution shows very similar daytime fire patterns and FRP totals from S3 and Terra MODIS, with SLSTR detecting around twice the number of AF pixels due to the algorithm being more effective at identifying low FRP pixels at the edges of fire clusters. Regional time series case studies also show very similar temporal patterns between S3 and Terra MODIS. Longer-term intercomparisons such as these will provide the knowledge necessary to use MODIS and SLSTR AF products together to analyse long-term AF trends. Comparing near simultaneous observations of fires by SLSTR and from the 30 m spatial resolution Landsat Operational Land Image (OLI) data, we find that once there are around 150 OLI active fire pixels detected within the area of an SLSTR pixel, the chances of that SLSTR pixel being classed as an active fire by the daytime algorithm rises to almost 100%. The daytime SLSTR AF Detection and FRP product based on the algorithm described herein has been fully operational since March 2022 and is available from the Sentinel-3 Science Hub (https://scihub.copernicus.eu/).

An increasing number of end-users looking for ground data about fire activity in regions where accurate official datasets are not available adopt a free-of-charge global burned area (BA) and active fire (AF) products for applications at the local scale. One of the pressing requirements from the user community is an improved ability to detect small fires (less than 50 ha), whose impact on terrestrial environments is empirically known but poorly quantified, and is often excluded from global earth system models. The newest generation of BA algorithms combines the capabilities of both the BA and AF detection approaches, resulting in a general improvement of detection compared to their predecessors. Accuracy assessments of these products have been done in several ecosystems; but more complex ones, such as regions that are characterized by frequent small fires and steep terrain has never been assessed. This study contributes to the understanding of the performance of global BA and AF products with a first assessment of four selected datasets: MODIS-based MCD45A1; MCD64A1; MCD14ML; and, ESA’s Fire_CCI in a mountainous region of northwest Yunnan; P.R. China. Due to the medium to coarse resolution of the tested products and the reduced sizes of fires (often smaller than 50 ha) we used a polygon intersection assessment method where the number and locations of fire events extracted from each dataset were compared against a reference dataset that was compiled using Landsat scenes. The results for the two sample years (2006 and 2009) show that the older, non-hybrid products MCD45A1 and, MCD14ML were the best performers with Sørensen index (F1 score) reaching 0.42 and 0.26 in 2006, and 0.24 and 0.24 in 2009, respectively, while producer’s accuracies (PA) were 30% and 43% in 2006, and 16% and 47% in 2009, respectively. All of the four tested products obtained higher probabilities of detection when smaller fires were excluded from the assessment, with PAs for fires bigger than 50 ha being equal to 53% and 61% in 2006, 41% and 66% in 2009 for MCD45A1 and MCD14ML, respectively. Due to the technical limitations of the satellites’ sensors, a relatively low performance of the four products was expected. Surprisingly, the new hybrid algorithms produced worse results than the former two. Fires smaller than 50 ha were poorly detected by the products except for the only AF product. These findings are significant for the future design of improved algorithms aiming for increased detection of small fires in a greater diversity of ecosystems.

Sentinel-3 active fire detection and FRP product performance - Impact of scan angle and SLSTR middle infrared channel selection

In this paper, the optimal threshold for fire detection is derived based on the probability density functions of the fire and background pixels. The threshold values and the commission and omission errors are computed according to several detection criteria: maximum likelihood, null-hypothesis testing, and minimum cost. This statistical approach enables the determination of optimal threshold value according to the specific requirements of the detection tasks I. INTRODUCTION Active fire detection using satellite thermal sensors usually involves thresholding the detected brightness temperature in several bands. The frequently used features for fire detection are the brightness temperature in the 4-µm wavelength band (T4) and the brightness temperature difference between the 4- µm and 11- µm bands (∆T=T4-T11) (1), (2), (3). Due to the statistical nature of the fire and background temperature distribution, no matter how the thresholds are chosen, there are bound to be commission errors (i.e. false alarms) and omission errors (false negatives) in the detection outcomes. In the commonly used fire detection algorithms, the thresholds are often arbitrarily defined, and hence may not be optimal for a given fire detection task. The optimal threshold values are expected to be site dependent. They are different in different regions and for different fire conditions (such as smoldering and flaming fires). Hence, a global set of threshold values may not work optimally for a specific site. The task of active fire hot spot detection can be modeled as a stochastic target detection procedure (4). In this model, each pixel of the image occupies a point (T4, ∆T) in the two- dimensional feature space, and each pixel belongs to either the fire class or the non-fire background class. This approach considers the probability density functions of the fire and background pixels and optimal thresholds are derived depending on the specific objectives of the detection tasks. In a previous work (4), we implemented a method for deriving the optimal detection threshold by minimizing a cost function which is a weighted sum of the omission and commission errors. Alternatively, the threshold can also be derived based on other criteria. In this paper, we derive the optimal thresholds for detecting active fires in MODIS scenes using various detection criteria: maximum likelihood, constant false alarm rate, and null-hypothesis testing. The test data set consists of several MODIS scenes over Sumatra and Borneo islands and coincidental high resolution SPOT scenes (5). The SPOT data serves as ground truth for validation of MODIS hotspots detected by the new algorithms based on stochastic models. The probability density functions (pdf's) of the fire and background pixels are derived from this dataset and these pdf's are used to derive the optimal thresholds for active fire detection.

Sentinel-3 SLSTR active fire detection and FRP product: Pre-launch algorithm development and performance evaluation using MODIS and ASTER datasets

Assessment of VIIRS 375 m active fire detection product for direct burned area mapping

Validation of GOES-16 ABI and MSG SEVIRI active fire products

Building a small fire database for Sub-Saharan Africa from Sentinel-2 high-resolution images

Since 2001, the Moderate Resolution Imaging Spectroradiometer (MODIS) sensor on board the Aqua and Terra platforms has made great strides in providing information on global burned areas (BA). However, the MODIS mission is nearing its end. The Visible Infrared Imaging Radiometer Suite (VIIRS) sensors, presented as the MODIS Aqua heritage, could be an excellent alternative to ensure the temporal continuity of this information at a moderate resolution. This paper describes and evaluates the effectiveness of our developed hybrid algorithm, which utilizes VIIRS reflectance and active fire products on the Google Earth Engine platform, in producing efficient information about BA. The study investigates the algorithm's performance in sub-Saharan Africa as the region of interest in 2019, using biweekly outputs and a spatial resolution of 250 m. The algorithm encompasses several steps, including pre-processing individual scenes, creating cloud-free composites, generating binary reference data for burned and non-burned areas, conducting a supervised classification using random forest, and performing region shaping. The VIIRS-BA final product, which includes three confidence levels (low, moderate, and high) known as the uncertainty layer, is compared to four other burned area products. The validation is conducted against 27 reference sampling units from the Sentinel-2 Burned Area Reference Database dataset, allowing for a comprehensive uncertainty assessment across five various biomes. The VIIRS-BA product identified 5.1 million km2 of BA, which was significantly larger than other global coarse resolution BA products such as FireCCI51, FireCCIS310, and MCD64A1 and close to the fine resolution FireCCISFD20 with a difference of 7.3%. The differences were less significant in biomes such as “Tropical Savannas” and “Temperate Grasslands” which are characterized by persistent biomass burning. Based on a stratified random sampling, the validation results demonstrate varying levels of accuracy for the VIIRS-BA product across different confidence levels. The commission error (CE) ranges from 7.8% to 23.4%, while the omission error (OE) falls between 29.4% and 58.8%. Notably, there is a significant reduction in OE (ranging from 40.7% to 50.5%) compared to global BA products like FireCCI51, FireCCIS310, and MCD64A1. When compared to VIIRS-BA, the FireCCISFD20 regional product has a 37% better OE performance. While VIIRS-BA shows great potential in detecting fires that global products miss, the VIIRS-BA with low confidence level tends to overestimate BA in regions with high fire activity. To address this, future versions of the algorithm will integrate the updated VIIRS reflectance data alongside VIIRS active fire from the National Oceanic and Atmospheric Administration to reduce CE and improve understanding spatial patterns.

In the absence of historical fire records, end-users intend to adopt free satellite-derived fire products, including global burn area (BA) and active fire (AF) products, to understand the historical fire dynamics for better forest management. Previous literature evaluated the accuracy of these fire products in regions with different environments, but no study evaluated the performance of these fire products in fire-prone, cloudy and mountainous areas. This study contributed to filling this gap, through the first evaluation of four broadly used fire products: MODIS-based MCD64A1, MCD14ML, VIIRS-based VNP14DLIMGTDL_NRT (hereafter simplified as VNP14DL), and ESA Fire_CCI51 over subtropical China. Two methods were applied to this end, the spatio-temporal clustering algorithm based on official historical fire records and the density-based random sampling and estimation method from the Landsat 8 fire scenes. The results show that both the AF and BA products show poor fire detection ability in this area (with all the F1-Score < 0.5). Among them, the VNP14DL performed best, followed by MCD14ML, Fire_CCI51, and MCD64A1. The MCD14ML had the best detection capability for small fires (< 50 hectares). AF products have an overall higher fire detection ability than BA products. These findings provide insights for the improvement of fire detection algorithms of these fire products in the fire-prone, cloudy and mountainous area over subtropical China.

ABSTRACTA global operational land imager (GOLI) Landsat-8 daytime active fire detection algorithm is presented. It utilizes established contextual active fire detection approaches but takes advantage of the significant increase in fire reflectance in Landsat-8 band 7 (2.20 μm) relative to band 4 (0.66 μm). The detection thresholds are fixed and based on a statistical examination of 39 million non-burning Landsat-8 pixels. Multi-temporal tests based on band 7 reflectance and relative changes in normalized difference vegetation index in the previous six months are used to reduce commissions errors. The probabilities of active fire detection for the GOLI and two recent Landsat-8 active fire detection algorithms are simulated to provide insights into their performance with respect to the fire size and temperature. The algorithms are applied to 11 Landsat-8 images that encompass a range of burning conditions and environments. Commission and omission errors are assessed by visual interpretation of detected active fire locations and by examination of the Landsat-8 images and higher spatial resolution Google Earth imagery. The GOLI algorithm has lower omission and comparable commission errors than the recent Landsat-8 active fire detection algorithms. The GOLI algorithm has demonstrable potential for global application and is suitable for implementation with other Landsat-like reflective wavelength sensors.

Madagascar's burned area from Sentinel-2 imagery (2016–2022): Four times higher than from lower resolution sensors