Last modified on Jan 19, 2026 13:53

Contributors: Martin Wooster (EI), Weidong Xu (EI), Carsten Brockmann (BROCKMANN CONSULT GMBH), Grit Kirches (BROCKMANN CONSULT GMBH), Martin Boettcher (BROCKMANN CONSULT GMBH)

Issued by: Earth Insight (EI)/ M.Wooster & W. Xu

Date: 10/12/2025

Ref: C3S2_313e_BC_WP1-DDP-FIRE-AF&FRP-S3-SLSTR-v1.2-2024_202506_PQAR

Official reference number service contract: 2024/C3S2_313e_BC/SC1

History of modifications

List of datasets covered by this document

Acronyms

General definitions

Active Fire (AF): A landscape fire that was actively burning when the satellite observations was made. Satellite ‘Active Fire’ Products are those that report information on these types of fires using thermal remote sensing techniques. AF pixels are pixels classified as containing one or more actively burning fires when the observation was made.

Brightness Temperature (BT) : The temperature of a hypothetical blackbody emitting an identical amount of radiation as being measured in the waveband.

Fire Radiative Power (FRP): The rate of radiant heat output from a landscape fire, typically expressed in Watts ´ 10⁶ (MW). FRP is typically very well related to a fire´s combustion rate (how much material is being burned per unit time) and rate of smoke emission, and hence remotely-sensed FRP measures are commonly used to estimate these terms. At the pixel scale, a satellite product typically is reporting the total FRP from all fires burning within that pixel at the time the observation was made. 

Fire Radiative Energy (FRE): The temporal integral of fire radiative power calculated over the fire’s lifetime, equating to the total amount of energy radiated by the fire. FRE is typically used to estimate how much material was burned in a fire and how much smoke was released.

 Error of Omission: A type of error where data is erroneously excluded from membership of a class when it should have been included. In satellite AF products this typically means a pixel being incorrectly left out of being classified as an AF pixel, when other data suggest it should have been.

 Error of Commission: A type of error where data is erroneously included in the membership of a class when it should have been excluded. In satellite active fire products this typically means a pixel being incorrectly classified as an AF pixel when other data suggest it should not have been. 

Gas Flare: A controlled open flame used to burn off excess natural or other combustible gas, typically at oil and gas facilities, chemical plants and natural gas processing plants, or landfills. Sometimes known as a flare stack, flare boom, ground flare, or flare pit.

Satellite Data Processing Levels

• Level 0 (L0) data are reconstructed, unprocessed instrument and payload data at full resolution, with any and all communications artefacts (e.g., synchronization frames, communications headers, duplicate data) removed.
• Level 1A (L1A) data are reconstructed, unprocessed instrument data at full resolution, time-referenced, and annotated with ancillary information, including radiometric and geometric calibration coefficients and georeferencing parameters (e.g., platform ephemeris) computed and appended but not applied to L0 data.
• Level 1B (L1B) data are L1A data that have been processed into sensor units (not all instruments have L1B source data).
• Level 1C (L1C) data are L1B data that include new variables to describe the spectra. These variables allow the user to identify which L1C channels have been copied directly from the L1B and which have been synthesized from L1B and why.
• Level 2 (L2) data are derived geophysical variables at the same resolution and location as L1 source data.
• Level 2A (L2A) data contains information derived from the geolocated sensor data, such as ground elevation, highest and lowest surface return elevations, energy quantile heights (“relative height” metrics), and other waveform-derived metrics describing the intercepted surface.
• Level 2B (L2B) data are L2A data that have been processed to sensor units (not all instruments will have a L2B equivalent).
• Level 3 (L3) are variables mapped on uniform space-time grid scales, usually with some completeness and consistency.
• Level 3A (L3A) data are generally periodic summaries (weekly, ten-day, monthly) of L2 products.
• Level 4 data are model output or results from analyses of lower-level data (e.g., variables derived from multiple measurements).

The description of data processing levels ranging from Level 0 to Level 4 have been cited from the following National Aeronautics and Space Administration (NASA) Earth Observation Data website: https://www.earthdata.nasa.gov/engage/open-data-services-and-software/data-information-policy/data-levels 

Target Requirements (TRs): Requirements on the datasets, production system and documentation for this service to meet the needs of the users. The C3S targets a variety of users, ranging from policy makers to scientists, industry sectors and businesses, although policy makers (e.g., DG CLIMA and Climate-ADAPT) and sectoral users (e.g., water, energy, urban, agriculture) are the main focus. The TRs takes the Global Climate Observing System (GCOS) requirements as a starting point and refine them to meet the specific needs of the C3S services ranging from climate monitoring, constraint of models, data assimilation, initialization of simulations, and attribution. For example, C3S might have requirements on latency for constraining climate prediction. The TRs will continuously evolve in time to capture new needs from users.

Breakthrough (B): An intermediate level between threshold and goal which, if achieved, would result in a significant improvement for the targeted application. The breakthrough value may also indicate the level at which specified uses within climate monitoring become possible. It may be appropriate to have different breakthrough values for different uses.

Goal (G): An ideal requirement above which further improvements are not necessary.

Threshold (T): The minimum requirement to be met to ensure that data are useful. 

Validation: The validation is essential for providing a high-quality product that is accepted and applied by the user community. The different steps of validation that jointly lead to the achievement of the validation objectives are anticipated:

• Internal validation
• Inter-comparison with other validated products
• Independent product validation and comparison
• User assessment and feedback

Verification: The process of confirming that a data product or processing system has been implemented correctly and produces outputs consistent with its design specifications and reference implementations. Verification focuses on internal consistency and technical correctness rather than accuracy against independent ground truth. The key steps of verification include:

• Input consistency checks - Confirming that input data files are correctly ingested and processed identically across different system implementations
• Intermediate product comparison - Systematic comparison of processing outputs at each stage (e.g., Level-2 products) between baseline and reference processors to identify algorithmic discrepancies
• End-product traceability - Comprehensive analysis of final products (e.g., Level-3 Gridded products) across different spatial and temporal resolutions to ensure processing chain integrity
• Discrepancy source identification - Characterization and attribution of any differences to specific processing stages, enabling targeted debugging and quality assurance

Executive summary

Landscape fires consume substantial quantities of vegetation and organic soil fuel across the globe, releasing combustion products as smoke containing numerous chemical compounds in both gaseous and particulate form. The Copernicus Climate Change Service (C3S) provides Climate Data Records (CDRs) for the Landscape Fire Essential Climate Variable (ECV), encompassing active fire (AF) pixel detections and associated fire radiative power (FRP) estimates, the latter of which has been demonstrated to correlate well with rates of fuel consumption and smoke emission.

The C3S Active Fire & FRP products deliver global hotspot detection and FRP information across temporal scales ranging from daily to monthly intervals, comprising daytime and night-time Level-2 Summary products in text format together with three Level-3 NetCDF gridded synthesis products at different spatio-temporal resolutions.  Each C3S product file consolidates the key information contained within hundreds to thousands of individual Sentinel-3 Sea and Land Surface Temperature Radiometer (SLSTR) Level-2 Active Fire & FRP product granules into a single file providing global coverage over periods of one day, 27 days, or one calendar month depending on the specific product type. Most hotspots identified within the C3S Active Fire & FRP products are associated with actively burning landscape fires. Accompanying the C3S Active Fire & FRP products is a complementary set of C3S Gas Flare products that focus exclusively on land and ocean-based gas flaring, identified through spectral radiance signatures and temporal persistence criteria, with observations limited to night-time acquisitions only. Together, these C3S products are designed to facilitate ease of use and are well-suited to support global modelling applications, fire emissions estimation, long-term trend analysis, and model evaluation studies. The C3S products have been designed to maintain alignment with the long-running MODIS Terra fire records, which are anticipated to conclude around 2026/27 when the Terra satellites cease operation. The similar equatorial crossing time of Sentinel-3 relative to Terra is expected to enable continuation of the consistent long-term global active fire and FRP record initiated by MODIS in 2000, which is particularly important given the strong diurnal variability characteristic of landscape fire activity.

Detailed product specifications and user guidance are provided in the accompanying C3S Product User Guide and Specification (PUGS) document, whilst this Product Quality Assessment Report (PQAR) provides a detailed description of the product validation methodology, consist of a systematic confidence building procedure and in an independent statistical validation, and a summary of the validation results. Prior to 2025, the assessment of the C3S FRP product is conducted via comparison to data from the MODIS sensor that operates onboard Terra since this instrument has a similar equatorial overpass time to SLSTR.

For the C3S FRP night-time products, the S3A and S3B datasets exhibit spatial patterns closely aligned with those observed by MODIS, indicating a substantial degree of agreement despite most MODIS night-time observations not being collected near-simultaneously with SLSTR. Analysis of active fire (AF) pixel counts reveals that the SLSTR product detects significantly more AF pixels than MODIS; however, the grid-cell FRP totals remain comparable between the two records, as the additional AF pixels identified by SLSTR—attributable to its enhanced sensitivity to night-time active fires—are predominantly of low FRP. Similarly, for the C3S FRP daytime products, the S3A and S3B datasets display spatial patterns that closely match those of MODIS, again demonstrating broad agreement despite the lack of near-simultaneous observations. The comparison of AF pixel counts shows that Sentinel-3 generally detects more AF pixels than MODIS, although the difference is less pronounced than at night. Again the grid-cell FRP totals are similar, as the extra AF pixels detected by SLSTR are mostly low-FRP pixels located at the periphery of fire clusters. These are partially offset by single-pixel fires detected by MODIS but missed by SLSTR during daytime. Overall, the additional pixels detected by SLSTR are of low FRP magnitude.

The assessment of the C3S Gas Flare product is performed by comparison with the NightFire product, which is derived from VIIRS sensor observations onboard Suomi-NPP and NOAA polar-orbiting environmental satellites. Although VIIRS has a different overpass time than SLSTR, this is not considered a significant limitation, as gas flares are more temporally stable and persistent than landscape fires. SLSTR achieved a 90.0% detection confirmation rate with VIIRS, while VIIRS achieved a 90.1% confirmation rate with C3S. The nearly identical co-location rates of approximately 90% and complementary omission rates of 10% between the two products underscore their high consistency, particularly given the differing overpass times of SLSTR and VIIRS.

After 2025, with the Terra satellite approaching the end of its operational life and scheduled for retirement in late 2025 or early 2026, MODIS-based assessment is no longer feasible. Consequently, the evaluation of C3S daytime and night-time FRP products is conducted through continuous verification between the C3S baseline dataset and the scientific prototype. Verification results from 2024–2025 indicate approximately 98% agreement between the two data versions. Due to the persistent nature of gas flares, validation of the corresponding products against data from previous year, with a comparison between the 2024 and 2023 gas flare datasets demonstrating approximately 90% agreement. These consistently high levels of agreement confirm that all twelve C3S products perform in accordance with expectations.

The PQAR is organized as follows: Section 1 outlines the product validation and verification methodology. Section 2 presents the results; detailed validation analyses for specific multi‑year periods are provided on dedicated sub‑pages (Sections 2.3–2.6), while the most recent year’s outcomes are reported as verification results (Section 2.7), reflecting the transition from MODIS‑based validation to continuous baseline–prototype verification as Terra approaches end‑of‑life. Section 3 introduces a progressive climate change assessment focused on extreme fire events; Section 4 summarises application‑specific assessments; and Section 5 details compliance with user data‑quality requirements and GCOS‑aligned targets.

1. Product validation & verification methodology

1.1. Validated and Verified Product Types and Descriptions

The C3S products comprise both 'Active Fire & FRP' products, covering global land during daytime and night-time, and 'Gas Flare’ products, available for global land and water but limited to night-time observations. Twelve product types are generated in total, all derived from observations made by the (SLSTR) operating onboard the Sentinel-3 satellites. The non-time critical (NTC) Level-2 Active Fire Detection and FRP products, generated using the algorithm designed by Wooster et al. (2012) and subsequently modified by Xu et al. (2020) and Xu & Wooster (2023), provide the source data for all C3S products. The eight C3S Active Fire & FRP product types comprise two Level-2 Summary products (daytime and night-time) providing text-based summaries of Level-2 hotspot information detected primarily using thermal infrared channels at 1 km resolution, and six Level-3 gridded synthesis products representing statistical summaries at daily, 27-day, and monthly intervals for both daytime and night-time observations. These products focus on land-detected hotspots, the vast majority of which relate to actively burning landscape fires. The Level-3 synthesis products include fractional cloud cover information to enable adjustment of detected hotspot counts, as utilised by the Global Fire Assimilation System (GFAS) within the Copernicus Atmosphere Monitoring Service (Kaiser et al., 2012). Fire radiative power derived from the MIR radiance method is considered most appropriate for landscape fire pixels (Fisher and Wooster, 2019).

The four C3S Gas Flare product types focus exclusively on gas flaring hotspots identified through spectral and temporal filtering of night-time SLSTR shortwave infrared (SWIR) observations at 500 m spatial resolution over both land and water. Gas flare FRP is optimally derived from SWIR spectral radiance signatures (Fisher and Wooster, 2019), which is less effective during daytime due to reflected solar radiation. Consequently, consistent with other gas flare remote sensing products such as VIIRS NightFire (Zhizhin  et al., 2021), C3S Gas Flare products are limited to night-time acquisitions. This temporal restriction is considered less problematic than for landscape fires given the relatively persistent nature of gas flaring activity. The products comprise a Level-2 Monthly Global Gas Flare Summary Product providing text-based summaries of hotspots classified as gas flares, and three Level-3 gridded synthesis products at daily, 27-day, and monthly intervals. Since the spectral and temporal filtering approach differs from the fixed geographic coordinate classification used in Level-2 products, gas flares identified in the C3S Gas Flare products may differ from those classified as 'gas flare' in the C3S Active Fire & FRP Summary products.

The formats and specifications of each of the C3S FRP product types are detailed in Table 1‑1.

Table 1‑1: Specifications of the twelve C3S product types discussed herein

1.2. Description of Validating Datasets

The NTC Sentinel-3 Level 2 Active Fire (AF) Detection and FRP Products are the data source used to generate all the C3S products detailed in Table 1‑1. These Level 2 products are subject to continuing quality checks and evaluation, including dedicated comparison to airborne data and to AF detection and FRP data coming from other satellite-based sensors, including from the MODIS sensor onboard the Terra satellite which has a similar overpass time to Sentinel-3 and which provides near simultaneous data at certain locations, dates and times. Since their first iteration in 2000, these MODIS AF products have been used to help address a very broad range of scientific questions concerning fire characterisation and the role of biomass burning within the Earth system (e.g. Wooster and Zhang, 2004; Ichoku and Kaufman, 2005; Giglio et al., 2006; Ichoku et al., 2008; Freeborn et al., 2011; Kaiser et al., 2012; Archibald et al., 2013; Hantson et al., 2013; Peterson and Wang, 2014; Sembhi et al., 2020; Zhang et al., 2018), and these are the primary data used to compare to both the NTC Sentinel-3 Level 2 Active Fire (AF) Detection and FRP Products and the C3S products derived from them. These MODIS data are best compared to the C3S Active Fire and FRP products. The C3S Gas Flare products on the other hand are best compared to the VIIRS NightFire product, which is the first global, long-term Earth Observation-based dataset that is commonly used for gas flare analysis (Zhizhin et al.,2021). These data are generated from observations made by the Visible Infrared Imaging Radiometer Suite (VIIRS) sensor carried on the Suomi-NPP satellite, and the subsequent NOAA polar orbiting satellites as well. They include information on hotspots detected on Earth at night by VIIRS, along with information that can be used to separate gas flares from other heat sources.

1.2.1. MODIS AF & FRP Product

MODIS provides radiometrically calibrated and geo-coded remote sensing observations of the Earth in 34 spectral bands over a 2330 km swath, including at times similar to those of the SLSTR sensor onboard Sentinel-3. Terra MODIS’ data are used to generate the pixel-level MOD14 MODIS Active Fire and Thermal Anomaly products (Giglio et al., 2016), which have similar characteristics and timing as the Sentinel-3 Level 2 Active Fire (AF) Detection and FRP Products and the C3S data derived from them. The latest Collection 6 MOD14 products are used as the reference data for the C3S FRP product evaluation. From the granule-level MOD14 Level 2 MODIS AF products, a series of summary products are generated, including the MODIS Climate Modelling Grid (CMG) AF products (Giglio et al., 2006), primarily intended for use in regional and global modelling. MODIS has a 16-day repeat cycle, and these MODIS CMG products are generated on a 0.25° spatial resolution grid – either every calendar month (MOD14CMQ) or every eight days (MOD14C8Q). An example of the corrected AF pixel count layer (CorrFirePix) from the MOD14CMQ product for January 2001 is shown in Figure 1‑1. Since Terra MODIS lowered its orbit its overpass time started drifting significantly in 2022 as it prepared for its end of life in later 2025/ early 2026, so we focus here on Terra MODIS data from before that period.

Figure 1‑1: Global ‘corrected active fire’ (AF) pixel count data layer (‘CorrFirePix’) extracted from the January 2001 Terra MODIS MOD14CMQ monthly Climate Modelling Grid (CMG) AF product.

1.2.2. VIIRS NightFire Product

The VIIRS NightFire product is the first global, long-term Earth Observation-based gas flaring dataset and leverages VIIRS' multi-spectral observations, the continued operation of some of the solar reflective channels at night, and the sensors low-light imaging capability. The NightFire product detects and characterizes hotspots globally at night, and uses a temperature retrieval algorithm to identify which ones are gas flares. This provides insights into global flaring activity and supports efforts to reduce flaring emissions (Zhizhin et al., 2021). The NightFire algorithm operates on night-time multispectral VIIRS observations, including in the shortwave infrared (SWIR) bands. The algorithm uses a Planck fitting approach using multi-band spectral radiance measures to estimate the effective flaring temperature (in Kelvin) and area (in square meters) of sub-pixel hot targets. Gas flares are identified by thresholding this temperature—generally those detections exceeding 1,400 K are assumed to be flares — and via their temporal persistence. These dual criteria help ensure that the algorithm reliably differentiates gas flares from transient or non-flaring heat sources, allowing for their accurate discrimination. From these sub-pixel temperature and area parameters, the radiative power output of the flares, also referred to as Fire Radiative Power (FRP) in the context of landscape fires, can be calculated using Stefan's Law. Since 2012, the VIIRS dataset has been used to provide annual global surveys of gas flare sites - derived from the NightFire products thresholded at 1,400 K. Nightly data are made accessible in CSV and Keyhole Markup Zipped (KMZ) formats. These datasets support both longitudinal studies and real-time assessments, enabling scientists to analyse flaring trends and their implications on a global scale. Validation studies have corroborated the robustness of the VIIRS NightFire methodology, showing a strong correlation with reported flaring data, with some regions achieving a coefficient of variation value as high as 0.92 (Zhizhin et al.,2021). Figure 2-2 below shows one year of the global gas flare data mapped from the VIIRS NightFire product of 2012.

Figure 1‑2: Global gas flares (red areas) detected in 2012 as present in the VIIRS NightFire products of that year. 

1.3. Description of Product Verification Methodology

The C3S products have undergone validation via comparison to the MODIS AF products and the VIIRS Night-fire products detailed above. They have also undergone and continue to undergo a verification process - comparing the output of the C3S processor to that from the scientific prototype code used to develop, test and improve the algorithms prior to C3S implementation. This scientific prototype incorporates and tests all advances designed to improve the accuracy, reliability, and consistency of the C3S products, and implements these to output a set of prototype products that are then tested prior to inclusion of the updates in the C3S baseline processor. The verification framework provides a robust means to confirm that once any algorithm enhancement is included in the latter, the performance of the C3S processor continues to correctly match that of the scientific prototype and thus the enhancements have been incorporated correctly in the operational system. Due to the persistent nature of gas flares, verification of the gas flare products is conducted by comparing C3S data from two neighbouring years.

1.3.1. Landscape Active Fire and FRP Product Verification

Our analysis employed a hierarchical verification framework to rigorously evaluate the correctness of the C3S processor performance and the resulting products throughout the entire processing chain. The framework commences with fundamental input file comparison, advances through Level-2 Monthly Summary product comparisons, and culminates in a comprehensive analysis of the three types of Level-3 Gridded products which each have different spatial and temporal resolutions. This systematic methodology enables the characterization of any discrepancies present at each stage, thereby facilitating the identification of specific sources of divergence between the C3S processor baseline and scientific prototype products.

The C3S processor baseline refers to the operational processing system developed by the C3S for the production of AF&FRP products. This baseline processor is designed to ensure consistency, reliability, and compliance with specified product requirements for routine data generation and dissemination. In contrast, the scientific prototype is developed by Earth Insight, the C3S science partner, to advance research and conduct scientific experiments. The prototype incorporates the latest algorithmic innovations and experimental methodologies, serving as a testbed for evaluating improvements prior to their integration into the operational C3S processor. The comparative analysis between the baseline and the scientific prototype thus allows for the rigorous validation of algorithmic enhancements and ensures that scientific advances are accurately and effectively translated into operational products. The core of the evaluation centered on direct statistical assessments of daily outputs from both the C3S baseline and prototype algorithms. Separate analyses were conducted for daytime and night-time processing, recognizing that algorithm performance may vary under different illumination and thermal conditions. Key metrics included fire pixel counts and FRP magnitudes, with particular attention given to differences between day and night processing.

Conventional temporal product verification often introduces artefacts due to repeated format conversions and aggregation steps, potentially masking algorithmic differences. To mitigate these issues, we employed the scientific prototype to grid the Level-2 Active Fire and FRP product source data directly to the Level-3 temporal aggregation periods, thereby bypassing the intermediate NetCDF processing steps—such as sequential gridding and aggregation to the 27-day and Monthly product intervals —the latter being steps that can potentially introduce other errors or obscure discrepancies. By processing the Level-2 Active Fire and FRP products in this manner, we generated temporal aggregations from the prototype processor that serve as a reference for the baseline C3S processor and product verification. This approach eliminates uncertainties associated with intermediate steps and enables confident evaluation of the Level-3 products. Overall, our methodology enhances the reliability and transparency of comparisons between operational baseline and experimental prototype algorithms.

1.3.2. Gas Flare Comparison

Given the highly persistent, slowly changing nature of gas flares, verification of the C3S Gas Flare product production is performed by comparing C3S data from the current year to that of the previous year, to identify any spurious large changes. Both datasets are initially aggregated to a global 0.25-degree grid over a one-year period. Within each grid cell, the presence of a gas flare is encoded as a binary indicator: cells containing at least one gas flare pixel detection are assigned a value of 1, irrespective of the total number of detections.

To assess the spatial agreement between the two years, a spatial window technique is employed. Specifically, for each SLSTR gas flare pixel detection, a 3×3 cell window centred on the detection is examined. If at least one gas flare pixel detection from the other year exists within this window, the detection is considered an agreement between the two years. The adoption of a 3×3 spatial window serves to accommodate minor geolocation inaccuracies and inherent uncertainties in gas flare positioning, thereby enhancing the robustness of interannual spatial agreement assessments.

Detection errors are identified through reciprocal analysis of both datasets. For SLSTR gas flare pixel detections in the current year, any detection lacking a corresponding detection within its 3×3 window in the previous year is classified as a commission error. Conversely, gas flare pixel detections from the previous year without a corresponding current year detection in their respective 3×3 windows are designated as omission errors. This methodology enables a quantitative assessment of spatial agreement between the two annual gas flare datasets and provides a robust framework for evaluating the continued reliability of the C3S Gas Flare product datasets.

1.4. Description of Product Validation Methodology

Beyond the verification approach referred to above, the validation methodology applied to perform quality assessment of the various C3S FRP and AF products referred to in this document was based on the method described in the PQAD [E.U. Copernicus Climate Change Service - PQAD-NT, 2024  & E.U. Copernicus Climate Change Service-PQAD-DT, 2024]. Certain of these validation approaches are performed in the specific regions used by the Global Fire Emissions Database (GFED), described in detail below.

1.4.1.  Level 2 Active Fire Detection and FRP Performance Assessment with Terra MODIS

1.4.1.1. Internal Validation

The NTC Level 2 Sentinel-3 AF detection and FRP Products are summarized and stored in CSV format in the C3S Level 2 Monthly Global Fire Location and FRP Summary Products (Daytime and Night-Time). Thus, the first validation step (internal validation) in the evaluation of the C3S FRP products is comparison of both the C3S Level 2 Monthly Summary Products and the Level 3a Daily Gridded FRP Products to the information contained within the set of NTC Level 2 Sentinel-3 AF detection and FRP Products from which they are derived. This ‘verification’ of the data contained in the C3S Level 2 Monthly Summary Product and the C3S Level 3a Daily Gridded FRP Product underlies the evaluation of all the C3S FRP products, since all C3S FRP products are derived from the same NTC Level 2 FRP Product datasets (and specifically the data stored within the FRP_in.nc file of those products).

1.4.1.2. Independent Validation

The second (independent) validation step is comparison of the C3S products to the (similar) MOD14 products generated from Terra MODIS. As detailed in the dataset descriptions above, at certain areas and times both sensors collect near-simultaneous data at a similar pixel resolution (~ 1 km at the near nadir point). With MOD14 as the comparison dataset, the C3S Level 2 Monthly Global Fire Location and FRP Summary Product AF detection errors of omission and commission can be calculated with respect to MODIS, as well as the degree of FRP agreement under two conditions - (i) when both sensors view the same individual fire cluster at almost the same time (e.g., within ± 6 minutes, following Xu et al., 2017; 2020; 2021), and (ii) when both sensors view the same larger land surface region within the same time interval. In these comparisons, in addition to requiring near simultaneous-views, MODIS data can be restricted to those with a scan angle maximum of ±30° to avoid geometric issues associated with the MODIS ‘bow-tie’ effect (Freeborn et al., 2011; 2014a; Xu et al., 2020; Xu & Wooster, 2023). This restriction limits the MODIS pixel area to a maximum of 1.7 km². To match this the SLSTR data can also be restricted to those with an S7 pixel area maximum of 1.7 km² (the matching SLSTR F1 pixel area maximum at this scan angle is 1.2 km²). To facilitate the inter-comparison, MODIS AF pixels are re-projected to the SLSTR Level 1b projection data grid, and Sentinel-3 AF errors of omission with respect to MODIS evaluated by considering whether an SLSTR AF detection was present within a 7 × 7 pixel window centred on each MODIS AF pixel location (following satellite AF product intercomparison methodologies adopted by Freeborn et al., 2014a; Xu et al., 2017; 2020; 2021).

Matching night-time SLSTR and MODIS example data are shown in Figure 1‑3. In this case SLSTR identifies all AF pixels detected in the MOD14 product (a total of 283), plus an additional 67 AF pixels that remained undetected by MODIS. The additional AF pixels detected by SLSTR are either isolated single pixel detections, or pixels lying at the edges of clusters of detected AF pixels. Figure1-3c also shows that at night many fires clearly show up in the 500 m spatial resolution SLSTR SWIR signal recorded in the sensor´s ‘S6' band, confirming that the AF detections based on the MIR channel data are the result of real fires and not false alarms (since there should be very few false alarms in SWIR-detected AF pixel data at night; Xu et al., 2020). A per-fire based FRP analysis of the data contained within the C3S Level 2 Monthly Summary FRP Product is conducted to intercompare its FRP values to those provided by MODIS of the same view viewed near-simultaneously. In this case a ‘fire’ is taken as comprising of a set of a spatially contiguous (or near-contiguous) AF pixels, sometimes termed a fire pixel ‘cluster’, since the two sensors may not detect an identical set of AF pixels even when both view the same fire at the same time. This 'cluster-based' intercomparison provides an estimate of the level of agreement in FRP when both sensors identify the same fire at almost the same time. The regional analysis extends this type of per-fire comparison, now intercomparing the total FRP identified across an area observed near simultaneously by SLSTR and by MODIS. The regional analysis indicates the effect of any AF detection errors of omission or commission on the regional-scale FRP total. These types of methodologies are common in satellite AF product intercomparisons (e.g., Freeborn et al., 2014a; Roberts et al., 2015; Xu et al., 2017; 2020; 2021).

Figure 1‑3: Comparison between near-simultaneous night-time observations of the same fire made by Sentinel-3A SLSTR and Terra MODIS. The location is that of the Fort McMurray wildfire in Alberta, Canada. The SLSTR image subset covers 200 km by 150 km and was collected at 04:56 UTC on 6th May 2016, with the matching MODIS data collected at 05:00 UTC. The intercomparison is based on the methodology outlined in Section 1.4.1. (a) SLSTR MIR-LWIR Channel (S7 – S8) Brightness Temperature (BT) difference image, where higher BT differences are depicted as brighter pixels. (b) Same as (a) but with the SLSTR AF detections overlain with a one pixel offset for clarity. (c) SLSTR S6 (SWIR) channel data matching (a), and (d) the same as (c) but with SLSTR AF detections superimposed as red ‘…’ and near-simultaneous MODIS AF detections as green ‘+’ (both with one pixel offsets for clarity). In (d), the blue rectangle highlights the AF pixels detected by SLSTR but missed by MODIS. Example taken from Xu et al. (2020).

Matching SLSTR and MODIS example data are shown in Figure 1‑4. In this case S3 SLSTR detected a total of 47 AF pixels, with 44 (94%) of these present in the MODIS AF product, and 3 (6%) not (area boxed in green in Figure 1‑4b).  MODIS detected a total of 48 AF pixels, most detected by SLSTR and of those missed ~ 80% were small fires located near water or cloud edges (see an example in the yellow boxed area of Figure 1‑4c). A per-fire based FRP analysis of the data contained within the C3S Level 2 Monthly Summary FRP Product is conducted to intercompare its FRP values to those provided by MODIS of the same view viewed near-simultaneously. In this case a ‘fire’ is taken as comprising a set of a spatially contiguous (or near-contiguous) AF pixels, since the two sensors may not detect exactly the same AF pixels. This intercomparison provides an estimate of the level of agreement in FRP when both sensors identify the same fire at almost the same time. The regional analysis extends this type of per-fire comparison, now intercomparing the total FRP identified across an area observed near simultaneously by SLSTR and by MODIS. The regional analysis indicates the effect of any AF detection errors of omission or commission on the regional-scale FRP total. These types of methodologies are common in satellite AF product intercomparisons (e.g. Freeborn et al., 2014a; Roberts et al., 2015; Xu et al., 2017; 2020; 2021).

Figure 1‑4: Comparison between near-simultaneous S3 SLSTR and Terra MODIS active fire (AF) data near Lake Rukwa in Tanzania, Southern Africa, based on the methodology detailed in section 1.4.1. The SLSTR image subset covers 200 km by 200 km and was collected at 07:51 UTC on 6th Aug 2020, and the matching MODIS data at 08:00 UTC. (a) SLSTR MIR Brightness Temperature (BT) difference between BT4 and S8 channel and where higher BT difference are depicted as brighter pixels. (b) Same as (a) but with the SLSTR AF detections overlain. (c) Same as (a) but with near-simultaneous MODIS AF detections (yellow boxed area - example of a fire only identified in the MODIS data; green boxed area in (b) - example of a fire only identified in the SLSTR data). 

1.4.2. Fire Pattern & FRP Magnitude Analysis

The performance of the C3S Level 2 Monthly Global Fire Location and FRP Summary Product can be best assessed against near-simultaneous MOD14 product files generated from Terra MODIS observations. In contrast, the three C3S Level-3 gridded FRP Products can be best evaluated via analysis of their spatial-temporal patterns. Comparison to those in the wider set of MOD14 data are expected to show similar patterns. Similar intercomparisons made between the Sentinel-3 SLSTR and Terra MODIS AF data records are also expected to form the basis of transfer functions used to blend these data together to develop a long-term AF data record spanning from the early 2000’s from MODIS and into the Sentinel-3 measurement period of the 2020's and beyond.

Figure 1‑5 shows a visual comparison between the spatial pattern of AF and FRP data retrieved from SLSTR night-time observations made by Sentinel-3A and -3B from March 2020 to Feb 2021 compared to those from Terra MODIS. Very similar spatial patterns are seen, indicating broad agreement despite the MODIS data including all night-time observations – including some not collected near-simultaneously with S3 SLSTR. A similar comparison of AF pixel counts shows that the SLSTR product includes more AF detections than does MOD14 (Xu and Wooster, 2020), but the grid-cell FRP totals shown in Figure 1‑5 are similar between the two records because the additional AF pixels that SLSTR detects in many of the grid cells are dominated by low FRP values.

Figure 1‑5: Total active fire (AF) pixel count and total FRP of fires detected within 0.25° grid cells from March 2020 to February 2021 using all Sentinel-3A, -3B SLSTR and Terra MODIS data. Note the Terra MODIS active fire pixel count has a scale which is 5 times smaller than that of S3A and S3B due to the lower number of AF pixels the former sensor typically detects.

1.4.3. Global Fire Emission Database (GFED) Regions

The Global Fire Emissions Database (GFED) ¹ divides the world into 14 basis regions ² for fire analyses:

• BONA: Boreal North America
• TENA: Temperate North America
• CEAM: Central America
• NHSA: Northern Hemisphere South America
• SHSA: Southern Hemisphere South America
• EURO: Europe
• MIDE: Middle East
• NHAF: Northern Hemisphere Africa
• SHAF: Southern Hemisphere Africa
• BOAS: Boreal Asia
• CEAS: Central Asia
• SEAS: Southeast Asia
• EQAS: Equatorial Asia
• AUST: Australia and New Zealand

These regions are used to categorise and analyse fire activity and emissions data in the GFED dataset, and are used for certain of the validation and intercomparison analyses performed herein

Figure 1‑6: Global Fire Emission Database (GFED) regions used herein. The abbreviations are as follows: Boreal North America (BONA); Temperate North America (TENA); Central America (CEAM); NH South America (NHSA); SH South America (SHSA); Europe (EURO); Middle East (MIDE); NH Africa (NHAF); SH Africa (SHAF); Boreal Asia (BOAS); Central Asia (CEAS); SE Asia (SEAS); Equatorial Asia (EQAS); Australia & New Zealand (AUST).

1. https://daac.ornl.gov/VEGETATION/guides/fire_emissions_v4.html ↩
2. https://www.globalfiredata.org/data.html ↩

1.4.4. Fire Season Metrics

The C3S Level 3 Gridded AF & FRP Night-time & Daytime Products are intended primarily for large scale analysis of fire patterns, seasonality, anomalies and trends. Such characteristics help define regional ‘fire regimes’, which help describe the role of landscape fires in an area, and under this broad definition, their physical attributes such as fire frequency, size, intensity, type and timing. Fire regimes may alter with changing climate and with human activity associated with e.g., landuse management and landuse change; (Moritz 2009; Flannigan et al., 2009; Archibald et al., 2013; Hantson et al, 2015). Characterizing past and current fire regimes has historically been performed by analysing field data such as charcoal records, fire-scar networks and fire occurrence databases. However, the regular and continuous information on landscape fires that can be provided by EO satellites is increasingly being used to determine certain fire regime characteristics (e.g., Chuvieco et al., 2008; Freeborn et al., 2014b). Fire seasonality is a key characteristic of regional fire regimes, and the C3S Level 3 gridded FRP Products are well suited to its determination, as are the existing MODIS MOD14 products and their MOD14CMQ and MOD14C8Q Climate Modelling Grid (CMG) summaries. To evaluate the C3S Level 3 Gridded FRP Products, we will derive fire regime seasonality characteristics from them and compare these to the same metrics derived from the MOD14 data used to create the MOD14CMQ and MOD14C8Q products. We will also directly intercompare the AF detection and FRP patterns present in the matching monthly temporal resolution C3S Level 3 Monthly Summary FRP Product and MOD14CMQ Products.

In terms of fire season, the start and end of the fire season for a grid cell or region can be defined as the time when the total FRP in the region exceeds a certain percentage of the total FRP of the whole year respectively, as illustrated for two grid cells in the Central African Republic in Figure 1‑7 (Freeborn et al., 2014b). The exact percentage thresholds can be altered as desired (in Figure 1‑7 the values of 10% and 90% are used). A key advantage of this cumulative approach is that it does not rely on a single threshold of active fire ‘amount’ being exceeded at any particular time-step (Freeborn et al., 2014b). This means it can be meaningfully and successfully applied in both (a) areas showing both short, intense periods of fire activity characterized by a clear peak above a threshold value (e.g., grid cell WGC in Figure 1‑7), and (b) areas with far longer but less intense fire seasons (e.g., grid cell EGC) where the absolute amount of fire activity can fluctuate and may sometimes drop below any pre-defined threshold. Once the start and end of the fire season are derived, the fire season duration can then be defined as the difference between them, whilst the fire season peak can be defined as that time when maximum fire activity is reached (Freeborn et al., 2014b). We will derive these fire season metrics from the C3S Level 3a Daily Gridded AF and FRP Night-time products and compare them to those same metrics derived from the night-time MODIS data of the same period.

Figure 1‑7: Fire season data for the Central African Republic (CAR) -  Map showing the locations of the western and eastern grid cells (labelled WGC and EGC) whose data is analysed in (Figure 1‑8), superimposed on percent tree cover characterized according to the 500 m Global Land Cover Facility (GLCF) Version 3 of the Collection 4 Vegetation Continuous Field (VCF) product (Bangui, the capital city of the Central African Republic, is indicated by a yellow star, and major road networks are also superimposed).

Figure 1‑8: Fire season data for the Central African Republic (CAR)  - Normalized seasonal profiles of (a) MODIS active fire (AF) pixel counts and (b) cumulative distributions of MODIS AF pixel counts for two 0.05° grid cells at 16-day temporal resolution. Seasonal profiles are generated from 10 years of aggregated observations, and the locations of the example grid cells, referred to as the western and eastern grid cells (WGC and EGC), are shown in (Figure 1‑7). The peak of the fire season is represented by the maxima shown in (a). The 10th and 90th percentiles of the cumulative AF pixel counts are shown in (b) to demonstrate the start and end of the fire season, whilst the duration is the difference between these dates.

Fire season metrics derived from the C3S Level 3a Daily Gridded AF & FRP Night-time Product files will be compared to those derived from MOD14 data gridded to the same 0.1° spatial resolution grid. The comparisons can be made per-grid cell, per biome or per region (e.g., see regions in Figure 1‑9a), and also both in terms of AF detections (as in Figure 1‑7 & Figure 1‑8), but also with FRP. As Sentinel-3 satellites have been reported to have the capability of detecting from three to five times more AF pixels that does near simultaneous MODIS data - while the total FRP of the fire cluster and regions is very close for both sensors (Xu et al., 2020), the fire season metrics derived from FRP data will be more similar than the ones calculated using numbers of AF pixels. Therefore, we focus on deriving fire season information from FRP data for this quality assessment. The most appropriate level of geographic comparison will be determined in part from the number of AF pixels present within the data (too few in a grid cell and the statistical analysis will be less meaningful and a larger concatenation of the data from multiple grid cells or over a region or biome will be required). As an example, at the global scale, Figure 1‑9b shows monthly global total night-time FRP as derived from Sentinel-3A, -3B and Terra MODIS gridded global products. All three products show a very similar temporal development at this global scale.

The C3S Level 3a 27-Day Gridded AF & FRP Daytime Product is simply the accumulation of twenty-seven C3S Level 3a Daily Gridded FRP Products, so its evaluation will simply focus on verifying the correctness of the lower temporal resolution statistical summary derived from the former data (i.e., internal validation/verification).

The C3S Level 3 Monthly Summary FRP Product will be compared to the MODIS MOD14CMQ product. Since the monthly temporal resolution is potentially too low for comparing the precise fire season start and end, the comparison will focus also on the degree of spatio-temporal FRP pattern agreement, defined by such metrics as the statistical summaries (mean, standard deviation, etc.), coefficient of determination (r²) and the slope of the linear best fit.

Figure 1‑9: Fire seasonality metrics. (a) Global regions used in the Global Fire Emissions Database (Van der Werf, 2017), and (b) night-time global monthly total FRP as derived from all fires across all regions in (a) using Sentinel-3A, -3B and Terra MODIS (The abbreviation of the names is as follows: Boreal North America (BONA); Temperate North America (TENA); Central America (CEAM); NH South America (NHSA); SH South America (SHSA); Europe (EURO); Middle East (MIDE); NH Africa (NHAF); SH Africa (SHAF); Boreal Asia (BOAS); Central Asia (CEAS); SE Asia (SEAS); Equatorial Asia (EQAS); Australia and New Zealand (AUST).

1.4.5. C3S Gas Flare Product Detection Performance

1.4.5.1. Internal Validation

Similar to the NTC Level 2 Monthly AF detection and FRP Summary Products, the C3S Level 2 Monthly Global Gas Fire Summary Products store their information in CSV format. Thus, the first validation step (internal validation) in the evaluation of these C3S Gas Flare products is comparison of both the C3S Level 2 Monthly Gas Flare Summary Products and the Level 3a Daily Gridded Gas Flare Products with information contained within the set of NTC Level 2 Sentinel-3 AF Detection and FRP Products from which they are derived. This ‘verification’ of the data contained in the C3S Level 2 Monthly Gas Flare Summary Product and the C3S Level 3a Daily Gridded FRP Product underlies the evaluation of all the C3S Gas Flare products, since all C3S FRP products are derived from the same NTC Level 2 FRP Product datasets (and specifically the data stored within the FRP_an.nc file of those products).

1.4.5.2. Independent Validation

The independent validation of the S3 SLSTR Gas Flare data is conducted using the VIIRS NightFire products described in Section 1.2.2. The comparison of data from the S3 and VIIRS gas flare products involves a systematic spatial analysis using a grid-based approach. Initially, both datasets are aggregated to a 0.25-degree global grid over a one-month to one-year period. For each grid cell, the presence of a gas flare is binary-coded, where cells containing at least a single gas flare pixel detection are assigned a value of 1, and this unity value remains regardless of the actual pixel detection count.

The assessment of the degree of agreement between the two products employs a spatial window technique. For each SLSTR gas flare pixel detection, a 3x3 cell window centred on the detection is examined and if at least one VIIRS gas flare pixel detection exists within this window, the detection is classified as an agreement between the two products. This approach accounts for potential minor geolocation differences and the different spatial resolution characteristics of the two sensors.

Commission errors are identified through a reciprocal analysis of both datasets. For SLSTR gas flare pixel detections, any detection without a corresponding VIIRS detection in its 3´3 window is classified as a commission error. Similarly, VIIRS gas flare pixel detections without corresponding SLSTR detections in their respective 3´3 windows are marked as omission errors. This approach enables a quantitative assessment of the spatial agreement between the two gas flare datasets whilst also accounting for the differing technical and operational characteristics of both sensor systems. As such the method provides a robust framework for evaluating the relative performance and complementarity of SLSTR and VIIRS gas flare detection capabilities, remembering that the original VIIRS NightFire product contains all hot-spot detections made and that these are reduced to only those of the gas flares using thresholding of the effective hotspot temperature and the persistence of detection, as detailed in Section 1.2.2

Figure 1‑10 shows an example of the spatial co-location analysis between VIIRS and SLSTR gas flare pixel detections in July 2023. The intercomparison reveals strong agreement between the data coming from the two instruments. Of the 1,850 total VIIRS gas flare pixel detections, 83.8% (1,551) were co-located with an SLSTR detection, while only 16.2% (299) were unique to VIIRS, indicating an SLSTR omission error of 16.2%. Similarly, for SLSTR's 1,571 total gas flare pixel detections, 92.0% (1,446) were co-located with VIIRS detections, with just 8.0% (125) being SLSTR commission errors with respect to VIIRS. This high degree of correspondence between the two instruments (>80% co-location rate for both) suggests robust detection capabilities and suggests a good reliability of gas flare identification from both satellite systems, considering for example that the VIIRS and SLSTR instruments observe the same location at a very different time of the night when cloud cover (through which gas flares cannot be detected) may also be quite different.

Figure 1‑10: Comparison of global gas flare pixel detections made by (a) Sentinel-3A SLSTR and (b) SNPP VIIRS in July 2023, based on data held within the C3S Level 2 Monthly Gas Flare Summary Products and the matching set of VIIRS NightFire products respectively.

2. Validation and Verification Results

All four C3S AF & FRP daytime products undergo annual quality checks and evaluation. Prior to 2025, assessments were based on comparisons with MODIS Terra (MOD14) products. Results for March 2023–February 2024 are presented in Section 2.6, with the prior period (March 2022–February 2023) in Section 2.5. An overall summary for both periods and both sensors is provided in Section 2.1 for the daytime products. Night-time products have been generated over a longer period: results for March 2023–February 2024 are presented in Section 2.6; prior results for March 2022–February 2023, March 2021–February 2022, and March 2020–February 2021 are presented in Sections 2.5, 2.4, and 2.3, respectively. For the 2023 C3S Gas Flare products, assessments are based on comparisons with the 2023 SNPP VIIRS NightFire products (see Sections 2.6.10–2.6.14).

After 2025, with the Terra satellite approaching end-of-life (expected late 2025 or early 2026), Terra MODIS-based assessment of the C3S FRP products is no longer feasible. The Aqua satellite carrying the second MODIS instrument overpasses at a completely different time of day to Sentinel-3, and given the very strong fire diurnal cycle this is also not feasible to use. Consequently, subsequent evaluation of the C3S daytime and night-time FRP products is conducted instead via continuous verification between the C3S baseline dataset and the scientific prototype, reflecting the transition from MODIS‑based validation to baseline–prototype verification. The most recent verification results from 2024–2025 demonstrate approximately 98% agreement between the two data versions as detailed in Section 2.7. For the Gas Flare products, where we expect slow changes over time in comparison to active fires, verification assessment is conducted through inter-comparison with the prior years' C3S dataset. A comparison of the annual gas flare datasets from 2024 and 2023 reveals approximately 90% agreement. These consistently high levels of concordance confirm that all twelve C3S products perform in accordance with established expectations.

2.1.  Validation Summary of C3S AF and FRP Daytime Products Validation

Table 2‑1 summarizes the validation metrics for the daytime C3S Level 3 Monthly Summary FRP products over the evaluated multi-year periods (2022-2024). The reported metrics include the coefficient of determination (r²) and the slope of the ordinary least squares linear best fit (SLOPE).

The results demonstrate exceptionally high agreement between the C3S products and the MODIS Terra reference data for both S3A and S3B SLSTR. For both sensors, the global r² values are consistently 1.0 across all evaluated years, indicating near-perfect correlation at the global scale. Within the GFED regions, r² values range from 0.88 to 1.00 for S3A and from 0.91 to 1.00 for S3B, reflecting strong regional consistency.

The SLOPE values, which assess the proportionality between the C3S and MODIS FRP estimates, are also close to unity. For S3A, global SLOPE values are 1.0 for 2023–2024 and 0.98 for 2022–2023, with regional values ranging from 0.74 to 1.02. For S3B, global SLOPE values are 1.0 for 2023–2024 and 0.98 for 2022–2023, with regional values between 0.71 and 1.09. These results confirm that the C3S Level 3 Monthly Summary FRP products provide highly reliable and consistent estimates of fire radiative power, both globally and across diverse fire-prone regions.

Table 2‑1. Summary of annual comparison results for the Daytime C3S Level 3 Monthly Summary FRP products.

2.2. Validation Summary of C3S AF and FRP Night-Time Products 

Table 2‑2 provides a summary of the metrics for the validation of the C3S Level 3 Monthly Summary FRP products for the four year long periods evaluated (2020-2024). The metrics include coefficient of determination (r²) and the slope of the ordinary least squares linear best fit (SLOPE).

For both S3A and S3B SLSTR, the global coefficient of determination (r²) is consistently 1.00 across all years, indicating an exceptionally high degree of correlation with the MODIS reference. Regional r² values within the GFED regions range from 0.74 to 1.00, reflecting strong agreement but with some variability across different regions and years.

The SLOPE values, representing the proportional relationship between the C3S and MODIS FRP estimates, are also close to unity at the global scale, with values ranging from 1.02 to 1.08 for S3A and from 1.00 to 1.08 for S3B. Regional SLOPE values span from 0.50 to 1.15, suggesting generally robust agreement with occasional regional deviations. These results collectively demonstrate that the Night-time C3S Level 3 Monthly Summary FRP products provide highly reliable and consistent estimates of fire radiative power, both globally and across diverse fire-prone regions, over multiple years.

Table 2‑2. Summary of annual comparison results for the Night-time C3S Level 3 Monthly Summary FRP products.

3. coefficient of determination  ↩
4. slope of the ordinary least squares linear best fit  ↩
5. coefficient of determination   ↩
6. slope of the ordinary least squares linear best fit   ↩

2.3. Product Evaluation Results: March 2020 to Feb 2021 AF & FRP Night-time

Subpage link - 2.3. Product Evaluation Results: March 2020 to Feb 2021 AF & FRP Night-time

2.4. Product Evaluation Results: March 2021 to Feb 2022 AF & FRP Night-time

Subpage link - 2.4. Product Evaluation Results: March 2021 to Feb 2022 AF & FRP Night-time

2.5. Product Evaluation Results: March 2022 to Feb 2023 AF & FRP Daytime & Night-time

Subpage link - 2.5. Product Evaluation Results: March 2022 to Feb 2023 AF & FRP Daytime & Night-time

2.6. Product Evaluation Results: March 2023 to Feb 2024 AF & FRP Daytime & Night-time and 2023 Gas Flare

Subpage link - 2.6. Product Evaluation Results: March 2023 to Feb 2024 AF & FRP Daytime & Night-time and 2023 Gas Flare

2.7. Product Verification Results: March 2024 to Feb 2025 C3S AF and FRP Daytime & Night-time Products and C3S Gas Flare Products

Subpage link - 2.7. Product Verification Results: March 2024 to Feb 2025 C3S AF and FRP Daytime & Night-time Products and C3S Gas Flare Products

3. Climate Change Assessment (to be implemented progressively)

Active Fire and FRP 'anomalies' refer to significant deviations in fire activity and magnitude (sometimes referred to as 'intensity'; though this is not the same as the standard 'fireline intensity' metric used in fire behaviour analysis). These 'anomalies' are derived typically from long-term historical baselines, as constructed from  satellite observation time-series. These anomalies are important because they provide quantitative identification of, and measures for, of unusual wildfire events and shifts, which can then potentially be linked to broader environmental or human activity changes, including those possibly related to climate change.

In recent years, analyses of AF detections and FRP measures have revealed that some regions maybe experiencing wildfire seasons with a length that has significantly exceeded historical norms, are seeing an increased frequency of high magnitude fire seasons, or somehow experiencing other potentially anomalous fire patterns. For example, satellite observations have shown that for certain years some areas of the planet appear to be exhibiting fires with FRP magnitude statistics several times greater than the long-term average, indicating periods of extreme fire activity. Such anomalies are often associated with prolonged heatwaves, droughts, and other climate-related factors that increase chances of extreme fire behaviour.  Abram et al. (2021) describe one such fire event in Australia for example. 

Because AF and FRP anomalies capture these exceptional fire events, they can serve as valuable indices for monitoring and assessing impacts of climate change on wildfire regimes. Persistent or widespread anomalies in AF and FRP metrics may suggest a shift in baseline fire behaviour, providing empirical evidence that climate change is influencing the frequency, duration, and/or intensity of wildfires across a region for example.

3.1. Global and Regional Extreme Fire Events 

Regional extreme fire events, such as the 2023 Canadian wildfires and the 2023–2024 Greek fires, underscore the possibility of increasing frequency of extreme wildfire events at different locations worldwide. MODIS Terra collected active fire and FRP data at a very similar time of day to Sentinel-3 SLSTR. Therefore, to illustrate how such changes in fire activity can be examined, 20 years of night-time MODIS Terra AF data (2001–2020) were analysed and compared to the 2023 Terra MODIS night-time observations, and to the night-time data contained in the C3S Sentinel-3 Active Fire and FRP products (Sentinel-3A and 3B). The scale of the 2023 Canadian wildfires, notable for their unprecedented size and impact, have already been linked to climate change in recent studies (i.e Byrne et al., 2024).

Figure 3-1 shows the cumulative night-time Fire Radiative Power (FRP) across Canada by month, comparing the 20-year monthly mean and maximum FRP derived from the MODIS Terra record to the 2023 MODIS FRP data and the 2023 Sentinel-3A/B SLSTR-derived FRP data extracted from the C3S products. These metrics demonstrate the seasonal progression of wildfire activity - starting in the spring - and highlights the anomalous 2023 fire season. In particular, in 2023 the FRP metrics increased substantially from May to September compared to the historical norm. The 2023 total annual FRP from MODIS (e.g. the cumulative value reached in Dec) was nearly an order of magnitude higher than the 20-year mean, and twice the 20-year maximum. The FRP data extracted from the C3S Active Fire and FRP Products (both S3A and S3B) also reported comparable anomalies, with total annual FRP values from both satellites also approximately an order of magnitude higher than the 20-year MODIS mean and twice the 20-year MODIS maximum. These findings underscore the comparable fire activity characterization capabilities of the Sentinel-3A/B sensors and thus the C3S Active Fire and FRP products in relation to MODIS Terra, which provides the long-term record but which is ending its life in the coming year or two. Extending this type of analysis to other regions, such as Greece and Portugal in 2024, may yield further insights into the degree of anomalous behaviour in these fires, and ultimately maybe used to study how such behaviours may be linked to climate change.

Figure 3‑1: Cumulative monthly night-time Fire Radiative Power (FRFP) assessed across Canada, comparing the 20-year monthly FRP mean and maximum as derived from the long-term MODIS Terra record with night-time observations of the 2023 'extreme fire year'. The 2023 observations also come from MODIS Terra, but also from the C3S Sentinel-3 Active Fire and FRP products (Sentinel-3A and 3B).

4. Application(s) specific assessments 

This section will be updated as further information becomes available.

5. Compliance with user requirements concerning data quality

The target requirements (TR) for the AF & FRP product are specified in E.U. Copernicus Climate Change Service-TRGAD, 2024. The TR were defined and updated according to the GCOS (see Table 5‑1 - Table 5‑3). GCOS requirements for the Fire ECV are defined in the 2022 GCOS ECVs Requirements (GCOS-245, 2022). Since 2016, both FRP and active fire (AF) detection have been classified as full ECVs (GCOS-200, 2016), and their 2016 specifications (Table 5‑1) were most recently updated in the 2022 plan (Table 5‑2; Table 5‑3). Whilst the 2016 ECV specifications for FRP and AF detection were largely achievable with current sensors, the 2022 specifications represent to some extent specifications beyond those capable of being provided with current observing systems. For details of the Goal (G), Breakthrough (B) and Threshold (T) classes, please see the General Definitions.

Table 5‑1. GCOS requirements (GCOS-200, 2016) for Active Fire detection and FRP

Table 5‑2: GCOS requirements (GCOS-245, 2022) for Active Fire detection

Table 5‑3: GCOS requirements (GCOS-245, 2022) for Fire Radiative Power

Table 5-4 and Table 5-5 respectively indicate the compliance of the C3S Active Fire & FRP and the companion C3S Gas Flare products with the GCOS-245 requirements listed in Table 5-2 for Active Fire Detection, and Table 5-3 for FRP. The compliance is related to the characteristics of the Level 2 data used to create the C3S Level 2 Summary products and the Level 3 Gridded Products, with the Level 2 Summary products including data at the full spatio-temporal resolution and the Level 3 Gridded products providing statistical averages.

Table 5‑4: C3S product compliance with GCOS-245 Requirements for Active Fire Detection (see Table 5.2). Values meeting GCOS-245 target requirements are indicated by cell highlighting in yellow.

Table 5‑5: C3S product compliance with GCOS-245 Requirements for FRP (see Table 5.3). Values meeting GCOS-245 target requirements are indicated by cell highlighting in yellow.

7. 2022 specification terms: G (Goal), B (Breakthrough), T(Threshold). Definitions for these concepts can be found in the General definitions. ↩
8. ESA CCI ECV Fire Disturbance: D.1.1 User requirement document, version 5.2.  - Heil, 2017 ↩
9. with respect to active fires burning with FRP equal to 5 MW km-2 in the detector ground footprint. ↩
10. Ibid ↩
11. with respect to active fires burning with FRP equal to 10 MW km-2 in the detector ground footprint ↩
12. with respect to active fires burning with FRP equal to 20 MW km-2 in the detector ground footprin ↩
13. 2022 specification terms: G (Goal), B (Breakthrough), T (Threshold). Definitions for these concepts can be found in the General definitions ↩

References

Abram, N.J., Henley, B.J., Sen Gupta, A., Lippmann, T.J., Clarke, H., Dowdy, A.J., Sharples, J.J., Nolan, R.H., Zhang, T., Wooster, M.J. and Wurtzel, J.B., 2021. Connections of climate change and variability to large and extreme forest fires in southeast Australia. Communications Earth & Environment, 2(1), pp.1-17.

Archibald, S., Lehmann, C.E., Gómez-Dans, J.L. and Bradstock, R.A., 2013. Defining pyromes and global syndromes of fire regimes. Proceedings of the National Academy of Sciences, 110(16), pp.6442-6447.

Byrne, B., Liu, J., Bowman, K. W., Pascolini-Campbell, M., et al. (2024). Carbon emissions from the 2023 Canadian wildfires. Nature. https://doi.org/10.1038/s41586-024-07878-z 

Chuvieco, E., Giglio, L. and Justice, C., 2008. Global characterization of fire activity: toward defining fire regimes from Earth observation data. Global change biology, 14(7), pp.1488-1502.

 E.U. Copernicus Climate Change Service - PQAD-NT, 2024. "Product Quality Assurance Document - ICDR Sentinel-3 Active Fire & Fire Radiative Power Night-time Products", E.U. Copernicus Climate Change Service, Document ref: WP2-ICDR-FRP-NIGHTTIME-2020-2023-SENTINEL3-v1.2_PQAD_v3.1, Available from: https://dast.copernicus-climate.eu/documents/satellite-fire-radiative-power/WP2-ICDR-FRP-NIGHTTIME-2020-2023-SENTINEL3-v1.2_PQAD_v3.1_final.pdf

E.U. Copernicus Climate Change Service-PQAD-DT, 2024. Product Quality Assurance Document (PQAD) CDR and ICDR Sentinel-3 Active Fire & Fire Radiative Power Daytime Products, Document ref: WP2-FDDP-FRP-DAYTIME-2023-SENTINEL3-v1.2_PQAD_v1.6, Available from: https://dast.copernicus-climate.eu/documents/satellite-fire-radiative-power/WP2-FDDP-FRP-DAYTIME-2023-SENTINEL3-v1.2_PQAD_v1.6_final.pdf

E.U. Copernicus Climate Change Service-TRGAD, 2024 . Target Requirements and Gap Analysis Document, Document ref: WP3-TR-GAD-2024_BA_v1.3, Available from: http://dast.data.compute.cci2.ecmwf.int/documents/satellite-fire-radiative-power/WP3-TR-GAD-2024_BA_v1.3.pdf

Donlon, C., Berruti, B., Mecklenberg, S., Nieke, J., Rebhan, H., Klein, U., Buongiorno, A., Mavrocordatos, C., Frerick, J., Seitz, B. and Goryl, P., 2012, July. The Sentinel-3 Mission: Overview and status. In 2012 IEEE International Geoscience and Remote Sensing Symposium (pp. 1711-1714). IEEE.

Elvidge, C. D., Zhizhin, M., Hsu, F. C., & Baugh, K. E. (2013). VIIRS Nightfire: Satellite Pyrometry at Night. Remote Sensing, 5(9), 4423-4449. https://www.mdpi.com/2072-4292/5/9/4423 

Fisher, D., & Wooster, M. J. (2018). Shortwave IR adaption of the mid-infrared radiance method of fire radiative power (FRP) retrieval for assessing industrial gas flaring output. Remote Sensing, 10(2), 305. 

Fisher, D., & Wooster, M. J. (2019). Multi-decade global gas flaring change inventoried using the ATSR-1, ATSR-2, AATSR and SLSTR data records. Remote Sensing of Environment, 232, 111298. 

Flannigan, M.D., Krawchuk, M.A., de Groot, W.J., Wotton, B.M. and Gowman, L.M., 2009. Implications of changing climate for global wildland fire. International journal of wildland fire, 18(5), pp.483-507. 

Freeborn, P.H., Wooster, M.J., Hao, W.M., Ryan, C.A., Nordgren, B.L., Baker, S.P. and Ichoku, C., 2008. Relationships between energy release, fuel mass loss, and trace gas and aerosol emissions during laboratory biomass fires. Journal of Geophysical Research: Atmospheres, 113(D1). 

Freeborn, P.H., Wooster, M.J. and Roberts, G., 2011. Addressing the spatiotemporal sampling design of MODIS to provide estimates of the fire radiative energy emitted from Africa. Remote Sensing of Environment, 115(2), pp.475-489. 

Freeborn, P.H., Wooster, M.J., Roberts, G. and Xu, W., 2014a. Evaluating the SEVIRI fire thermal anomaly detection algorithm across the Central African Republic using the MODIS active fire product. Remote Sensing, 6(3), pp.1890-1917. 

Freeborn, P.H., Cochrane, M.A. and Wooster, M.J., 2014b. A decade long, multi-scale map comparison of fire regime parameters derived from three publicly available satellite-based fire products: a case study in the Central African Republic. Remote Sensing, 6(5), pp.4061-4089. 

Giglio, L., Csiszar, I. and Justice, C.O., 2006. Global distribution and seasonality of active fires as observed with the Terra and Aqua Moderate Resolution Imaging Spectroradiometer (MODIS) sensors. Journal of geophysical research: Biogeosciences, 111(G2). 

Giglio, L., Schroeder, W. and Justice, C.O., 2016. The collection 6 MODIS active fire detection algorithm and fire products. Remote Sensing of Environment, 178, pp.31-41. 

GCOS (2016) The Global Observing System for Climate (GCOS): Implementation Needs. (GCOS-154). Geneva, Switzerland: World Meteorological Organization. 

GCOS (2022). The Global Observing System for Climate (GCOS): Implementation Needs. (GCOS-245). Geneva, Switzerland: World Meteorological Organization. 

Hantson, S., Padilla, M., Corti, D. and Chuvieco, E., 2013. Strengths and weaknesses of MODIS hotspots to characterize global fire occurrence. Remote Sensing of Environment, 131, pp.152-159. 

Hantson, S., Pueyo, S. and Chuvieco, E., 2015. Global fire size distribution is driven by human impact and climate. Global Ecology and Biogeography, 24(1), pp.77-86. 

Heil A., Yue C., Mouillot F. and Kaiser J.W. (2017): ESA CCI ECV Fire Disturbance: D.1.1 User requirement document, version 5.2. Available from: https://climate.esa.int/en/projects/fire/key-documents/ 

Ichoku, C. and Kaufman, Y.J., 2005. A method to derive smoke emission rates from MODIS fire radiative energy measurements. IEEE transactions on Geoscience and Remote Sensing, 43(11), pp.2636-2649. 

Ichoku, C., Giglio, L., Wooster, M.J. and Remer, L.A., 2008. Global characterization of biomass-burning patterns using satellite measurements of fire radiative energy. Remote Sensing of Environment, 112(6), pp.2950-2962. 

Kaiser, J.W., Heil, A., Andreae, M.O., Benedetti, A., Chubarova, N., Jones, L., Morcrette, J.J., Razinger, M., Schultz, M.G., Suttie, M. and Van Der Werf, G.R., 2012. Biomass burning emissions estimated with a global fire assimilation system based on observed fire radiative power. Biogeosciences, 9(1), pp.527-554. 

Morisette, J.T., Giglio, L., Csiszar, I., Setzer, A., Schroeder, W., Morton, D. and Justice, C.O., 2005. Validation of MODIS active fire detection products derived from two algorithms. Earth Interactions, 9(9), pp.1-25. 

Moritz, M.A., Parisien, M.A., Batllori, E., Krawchuk, M.A., Van Dorn, J., Ganz, D.J. and Hayhoe, K., 2012. Climate change and disruptions to global fire activity. Ecosphere, 3(6), pp.1-22. 

Roberts, G., Wooster, M.J., Perry, G.L., Drake, N., Rebelo, L.M. and Dipotso, F., 2005. Retrieval of biomass combustion rates and totals from fire radiative power observations: Application to southern Africa using geostationary SEVIRI imagery. Journal of Geophysical Research: Atmospheres, 110(D21). 

Roberts, G., Wooster, M.J., Xu, W., Freeborn, P.H., Morcrette, J.J., Jones, L., Benedetti, A., Jiangping, H., Fisher, D. and Kaiser, J.W., 2015. LSA SAF Meteosat FRP products–Part 2: Evaluation and demonstration for use in the Copernicus Atmosphere Monitoring Service (CAMS). Atmospheric Chemistry and Physics, 15(22), pp.13241-13267. 

Schroeder, W., Prins, E., Giglio, L., Csiszar, I., Schmidt, C., Morisette, J. and Morton, D., 2008. Validation of GOES and MODIS active fire detection products using ASTER and ETM+ data. Remote Sensing of Environment, 112(5), pp.2711-2726. 

Sembhi, H., Wooster, M., Zhang, T., Sharma, S., Singh, N., Agarwal, S., Boesch, H., Gupta, S., Misra, A., Tripathi, S.N. and Mor, S., 2020. Post-monsoon air quality degradation across Northern India: assessing the impact of policy-related shifts in timing and amount of crop residue burnt. Environmental Research Letters, 15(10), p.104067.

Van der Werf, G.R., Randerson, J.T., Giglio, L., Leeuwen, T.T.V., Chen, Y., Rogers, B.M., Mu, M., Van Marle, M.J., Morton, D.C., Collatz, G.J. and Yokelson, R.J., 2017. Global fire emissions estimates during 1997–2016. Earth System Science Data, 9(2), pp.697-720. 

Wooster, M.J., Zhukov, B. and Oertel, D., 2003. Fire radiative energy for quantitative study of biomass burning: derivation from the BIRD experimental satellite and comparison to MODIS fire products. Remote Sensing of Environment, 86(1), pp.83-107. 

Wooster, M.J. and Zhang, Y.H., 2004. Boreal forest fires burn less intensely in Russia than in North America. Geophysical Research Letters, 31(20). 

Wooster, M.J., Roberts, G., Perry, G.L.W. and Kaufman, Y.J., 2005. Retrieval of biomass combustion rates and totals from fire radiative power observations: FRP derivation and calibration relationships between biomass consumption and fire radiative energy release. Journal of Geophysical Research: Atmospheres, 110(D24). 

Wooster, M.J., Xu, W. and Nightingale, T., 2012. Sentinel-3 SLSTR active fire detection and FRP product: Pre-launch algorithm development and performance evaluation using MODIS and ASTER datasets. Remote Sensing of Environment, 120, pp.236-254. 

Xu, W., Wooster, M.J., Kaneko, T., He, J., Zhang, T. and Fisher, D., 2017. Major advances in geostationary fire radiative power (FRP) retrieval over Asia and Australia stemming from use of Himarawi-8 AHI. Remote Sensing of Environment, 193, pp.138-149. 

Xu, W., Wooster, M.J., He, J. and Zhang, T., 2020. First study of Sentinel-3 SLSTR active fire detection and FRP retrieval: Night-time algorithm enhancements and global intercomparison to MODIS and VIIRS AF products. Remote Sensing of Environment, 248, p.111947. 

Xu, W., Wooster, M.J., He, J. and Zhang, T., 2021. Improvements in High-Temporal Resolution Active Fire Detection and FRP Retrieval over the Americas using GOES-16 ABI with the Geostationary Fire Thermal Anomaly (FTA) Algorithm. Science of Remote Sensing, p.100016. 

Xu, W. and Wooster, M.J. (2023). Sentinel-3 SLSTR active fire (AF) detection and FRP daytime product-Algorithm description and global intercomparison to MODIS, VIIRS and Landsat AF data. Science of Remote Sensing, p.100087.

Zhizhin, M., Matveev, A., Ghosh, T., Hsu, F. C., Howells, M., & Elvidge, C. (2021). Measuring gas flaring in Russia with multispectral VIIRS nightfire. Remote Sensing, 13(16), 3078.

Related articles

• Page:
  Active Fire & Fire Radiative Power (AF & FRP) & Gas Flare Products V1.2: Product Quality Assessment Report (PQAR) (Copernicus Knowledge Base)

  • c3s
  • ecv
  • af
  • frp
  • pqar
• Page:
  Active Fire & Fire Radiative Power (AF & FRP) & Gas Flare Products V1.2: Product User Guide and Specifications (PUGS) (Copernicus Knowledge Base)

  • c3s
  • ecv
  • af
  • frp
  • pugs
• Page:
  Active Fire & Fire Radiative Power (AF & FRP) & Gas Flare Products V1.2: Algorithm Theoretical Basis Document (ATBD) (Copernicus Knowledge Base)

  • c3s
  • ecv
  • af
  • frp
  • atbd