Between 1750 and 2016, atmospheric concentrations of carbon dioxide (CO2) have increased from ~277 ppm to ~403 ppm, an increase attributed to anthropogenic release through fossil fuel and cement emissions, deforestation, and other land use changes. By replacing ecosystems often dominated by woody plants with agriculture dominated by annual herbaceous vegetation, land use accelerates the biomass turnover and depletes the biomass and soil carbon stocks. Hence, land use and land cover change is the second largest anthropogenic contributor to current climate change, causing ~10% of the atmospheric increase in CO2 concentrations. Even though it plays such a major role within the global carbon cycle, it is still its most uncertain component. Hence, a comprehensive, spatially explicit, global quantification of the effect of land cover change is of major importance for improved understanding of spatial and inter-annual variability in the global net terrestrial carbon sink.

The dynamic global land cover product 1992-2015 from the European Space Agency (ESA) climate Change Initiative (CCI) based on earth observation data provides a timely opportunity to quantify the specific role of land use and land cover change on the net terrestrial carbon sink. We aim at studying the impact of land use and land cover change on the contribution of the terrestrial biomes on the global net terrestrial carbon sink 1992-2015. We therefore forced the dynamic global vegetation model LPJ- GUESS with the ESA-CCI dynamic land cover data set 1992- 2015. We thereafter partitioned the LPJ-GUESS derived terrestrial carbon sink to the different land cover classes and analysed their contribution to the mean, trend and inter-annual variability of the global net terrestrial carbon sink 1992-2015. Preliminary results indicate that tropical and boreal forest were the main contributors to the mean global net terrestrial carbon sink We found that drylands were the biome that contributed the most to the inter-annual variability of the global net terrestrial carbon sink. Finally, the biome with the strongest contribution the trend in the global net terrestrial carbon sink was boreal forests.

Area A - Board 30

The surface albedo is an Essential Climate Variable (ECV) that needs to be generated on a regular basis in order to ensure continuous estimates as a contribution of the radiation budget to water and carbon balance. First of all, the removal of atmospheric effects must be properly handled. Thus, the consistency of the MAJA method for Sentinel-2 and Landsat-8 is clearly an asset. The present developments are intended to generate a time evolving surface albedo product at the enhanced resolution of 10m to foster an advanced research in agricultural area. Further, surface albedo product will be considered to estimate the radiation forcing due to the surface since for instance maintaining permanently vegetation, instead of having episodes of bare soil, would contribute to cool the atmosphere notably, thereby reducing carbon emission.
Hence, the method – applied to become operational since inherited from Copernicus Global Land Service - makes use of the well-established approach based on a semi-empirical BRDF kernel-driven model. Such model is applied to Level 2a data sets of Sentinel-2 and Landsat-8. BRDF coefficients can serve to perform a normalization of the data and also to compute the spectral albedo in weighing the angular integrated kernels. Narrow to broadband albedo conversion rely on PROSAIL model ditto. Surface albedo products are generated with a quality control and an uncertainty assessment.
A composite period of several weeks must be considered to gather sufficient observations in order to build a BRDF product from which the surface albedo is derived. The potential impact due to the frequency of revisit offered by Sentinel2-A & -B plus Landsat-8 is assessed through a time evolution analysis. For time being, the surface albedo product is refreshed during synthesis periods about 10 days or less based on an update of its intensity. Demonstrative results are shown for selected time periods of 2018. The validation is carried on for two anchor ICOS-like (Integrated Carbon Observation System) stations operated by CESBIO and located near Toulouse, which are covered by crops (maize, wheat, and sunflower, merely).

Area A - Board 31

Effects from rising global temperatures pose great risks to agricultural ventures and wildlife. The main cause for global warming are anthropogenic greenhouse gas emissions, such as CO₂. Better understanding of the global carbon cycle, photosynthesis, and global terrestrial gross primary production (GPP) as the largest uptake mechanism of atmospheric CO₂ is an essential part of how we respond to the climate change that we face today.

Recently, Sun-Induced Fluorescence (SIF) emerged as a possibility to measure photosynthetic activity in large scales via satellites. SIF correlates with GPP and is used to study the carbon cycle. Photosynthesis responses to environmental changes can be studied via SIF measurements.

As SIF is long-wavelength electromagnetic radiation and not fully absorbed by the atmosphere, large areas can be probed for SIF intensity remotely by sensors shouldered on satellites. However, to date there are only sensors with coarse resolutions and small swaths available. The need for more highly resolved SIF measurements is evident.

Facing the problem of too coarse spatial and temporal resolutions in SIF measurements, we propose to use highly resolved images from the Sentinel-2 satellites and apply a machine learning solution to calculate a SIF map with improvements on space and time resolution compared to the existing SIF measurements.

The study employs neural networks to learn a mapping from 13 spectral bands of Sentinel-2 reflectances to the SIF signal of the same location. This is achieved by training the algorithm to predict SIF given by the OCO-2 mission from the input of Sentinel-2 bands. Thus, we can produce highly resolved SIF, which is usually computed from a highly resolved spectrum, using only a coarse spectrum of 13 bands. Furthermore, we contribute spatially highly resolved SIF data to the research in GPP, CO₂ uptake, and global warming - laying the foundations for more accurate estimates and predictions.

Area A - Board 32

Gross Primary Productivity (GPP), which is a critical indicator for the plant photosynthesis, is sensitive to the climate change according to plant trait and growing condition. On canopy scale and regional scale, Light Use Efficiency (LUE) model was a typical tool to estimate GPP, but had many structures for different regions and vegetation types. This study aimed to determine a general LUE model structure, and propose a method to predict the corresponding model parameters, which determines the sensitivity of GPP to climate forcing factors, using vegetation and climate traits. Firstly, based on a kind of LUE model which define GPP as the product of absorbed photosynthetically active radiation (APAR), maximum LUE, and its sensitivity functions of air temperature, vapor pressure deficit, water availability index, radiation and cloudiness index, totally 1920 model structures were compared for 13 plant functional types of vegetation in 177 sites globally. The general model was determined by the overall GPP estimation accuracy, which was assessed by Nash-Sutcliffe Model Efficiency (NSE) and Akaike Information Criteria (AIC). Then the model parameters were predicted using the vegetation and climate features, such as mean seasonal cycle, annual and interannual variability of vegetation index, air temperature and precipitation. The cross-validated result show that the general model was including APAR, maximum LUE and the sensitivity functions of air temperature, vapor pressure deficit, water availability index and cloudiness index, and could simulate GPP of 177 sites with NSE of 0.731 and AIC of 481714. The model parameters could be predicted by: 1) the median of the parameters of sites with identical plant functional type and climate type; 2) the parameters of site with smallest vegetation and climate distance; 3) the parameters of site with smallest sensitivity function distance. However, the optimal method for parameters prediction still need to be further explored. The result of this study could also be applied to upscale LUE models, and enhance the understanding of the plant photosynthesis reaction to the climate change on canopy and global scale.

Area A - Board 33

Northern terrestrial ecosystems store high amounts of carbon, and play an important role in the global carbon cycle. However, carbon balance mechanisms in these areas are not completely understood. This project aims at better understanding the ecosystem scale carbon cycle which will be used to upscale carbon fluxes to landscape level using optical remote sensing data.

In this study the main objective is to understand the influence of spatial heterogeneity of vegetation on ecosystem carbon fluxes, namely gross primary production (GPP), ecosystem respiration (ER), and net ecosystem exchange (NEE). Vegetation variation near a flux tower has a strong influence on observed carbon fluxes and therefore an accurate description of surface characteristics within a flux tower footprint area is important for understanding the ecosystem carbon exchange.

The project is based on novel methodology and combines carbon flux measurements from eddy covariance towers with remote sensing data. Research infrastructures such as SITES (Swedish Infrastructure for Ecosystem Science) and ICOS (Integrated Carbon Observation System) provide ecosystem level measurements of carbon exchange in major Nordic ecosystems: forests, wetlands and agriculture sites. Field stations in Sweden, Denmark and Finland also represent different climatic zones from temperate to boreal and sub-arctic areas. New European Sentinel-2 satellite provides high-resolution (10 meters) data for mapping the land surface properties such as land use classification and spectral vegetation indices. UAV (unmanned aerial vehicle) data will be used for mapping fine scale vegetation heterogeneity, for example in the transition zones where vegetation types change. Footprint models will be used to match carbon fluxes with the area that contributes the flux tower measurements. Spatial and temporal modeling will then be used to upscale carbon fluxes from flux towers to the landscape level.

The combination of footprint modeling and high-resolution remote sensing data can be used to describe the land surface heterogeneity over the footprint area and improve the accuracy of footprint models.

Area A - Board 34

More than one third of carbon dioxide (CO2) in the atmosphere is exchanged annually with the terrestrial ecosystems through processes of photosynthesis, plant respiration and heterotrophic respiration. The uptake of CO2 by terrestrial ecosystems through gross primary productivity (GPP) reduces the rate of atmospheric CO2 build-up and hence help mitigate global warming. Therefore, accurate estimation CO2 exchange between the atmosphere and terrestrial ecosystems is critical in understanding the global carbon cycle and the ecosystem response to climate change. At local scale, the prominent method of estimating terrestrial ecosystems carbon exchange is through the use of Eddy Covariance (EC) measurements. While EC techniques are important at these scales, they are often limited to the extent of the designated flux tower footprint. Furthermore, the spatial distribution of flux tower sites globally is limited making their use in quantifying carbon flux at continental to global scales impractical. The development of satellite sensors and GPP algorithms that use these data has provided an opportunity to address these limitations. Many models with varying assumptions on how ecosystems respond to environmental factors have been developed for estimating GPP at regional to global scale from remote sensing data. However, at present, only one satellite data based GPP model (i.e. the MOD17 model) is configured to produce GPP products operationally. Furthermore, the continuity of the MODIS satellite mission and its products in their current form, is not guaranteed beyond its current mission lifespan. The launch of European Space Agency (ESA) Sentinel missions through the Copernicus programme, specifically Sentinel-3 with its moderate spatial and high temporal resolution, provide a unique opportunity to explore new ways of operationally producing GPP at regional to global scales. A new model called Southampton Carbon Flux Model (SCARF), which uses the fraction of of photosynthetically active radiation (PAR) absorbed only by the photosynthetic elements of the canopy (i.e. green FAPAR rather than total canopy FAPAR) and exploits the intrinsic quantum yield of the two main photosynthetic pathways of plants (i.e. C3 and C4 photosynthesis) thereby minimising the need for a detailed vegetation functional type classification, was recently developed and showed promise in deriving GPP at regional to global scales. In this study, we evaluated the potential of using the recently available Sentinel-3 Ocean and Land Colour Instrument Terrestrial Chlorophyll Index- OTCI data in the SCARF model to derive GPP across a various biomes in Europe. The results from the modelling exercise were validated using data from 15 flux tower sites representing various vegetation types. Overall, the results clearly demonstrate the potential of Sentinel-3 OTCI data to provide routine estimates of terrestrial ecosystem GPP at regional to global scale.

Area A - Board 35

The African tropics and subtropics is the most dynamic region of the world with respect to terrestrial carbon flux variability and population growth, which imposes direct carbon emissions and perturbations to the natural ecosystems. The underlying mechanisms such as photosynthesis, respiration, natural and man-made biomass burning, as well as fossil fuel related emissions vary on sub-daily timescales and, they are subject to external forcing by human, meteorological, and climatic factors. Thus, gaining insights into the functioning of African carbon cycling and into its sensitivity to environmental variability and perturbations needs observations from the sub-daily to the seasonal to the year-to-year time-scale. However, the African continent is poorly sampled by current and planned atmospheric observation systems. Establishing a dense and robust ground-based network is logistically challenging. Satellite observations from low-Earth-orbit (LEO) are limited to a single local overpass time per day and, they are frequently cloudy.

The Middle East and Europe are major global players in fossil fuel extraction and usage, respectively. Leakage of carbon gases has been reported throughout the extracting-processing-transportation-consumption chain. Satellite observations from LEO will become progressively available as part of envisioned surveillance concepts. These LEO sensors, however, do not capture characteristic diurnal variations of fluxes, potentially resulting in biased emission estimates and lacking the ability to discriminate between man-made and biospheric flux signals.

The AbsoRption spectRometric patHfindEr for carboN regIonal flUx dynamicS (ARRHENIUS) will overcome the sampling gaps in the tropics and subtropics and on the sub-daily time scale by adopting a process-focused sampling strategy from geostationary orbit (GEO). For selectable focus regions, ARRHENIUS will deliver quasi-contiguous maps of atmospheric carbon species concentrations (carbon dioxide (CO₂), methane (CH₄), and carbon monoxide (CO)) and a photosynthesis process marker (solar induced plant fluorescence (SIF)) with sub-daily, seasonal, and year-to-year coverage. These observations will be used by carbon cycle models to inform on regional carbon cycle processes. Being process focused instead of surveillance driven, ARRHENIUS will pioneer a flexible and intelligent sampling approach with short lead times for pointing adjustments. Sampling will be flexible with regard to focus region selection, region extent (typically 1300x2400 km² with 2x2 km² footprint at sub-satellite), dwell times (typically 1h) and the number of revisits per day (up to 5 times per day), per season and per year. Sampling will be intelligent by actively avoiding regions which are expected cloudy based on observations of meteorological sounders (such as Meteosat Third Generation) from adjacent orbits.

Area A - Board 36

Increases in atmospheric carbon dioxide (CO2) concentration due to anthropogenic carbon emissions are partly buffered by carbon uptake into the ocean and by the land surface [1]. Dynamic changes in the strength of the land sink are directly mirrored in fluctuations of atmospheric CO2 concentrations [e.g. 2] with major implications for our climate. However, CO2 fluxes between the land surface and the atmosphere are not sufficiently constrained and different quantification approaches yield strongly diverging estimates both regionally and globally [3,4]. One major source of uncertainty in the CO2 flux estimates is an incomplete understanding of different feedback mechanisms between environmental conditions and vegetation functioning/photosynthetic carbon uptake and respirational release [e.g. 5].
We present advances in data-driven modeling of the biogenic fluxes (net ecosystem exchange NEE, gross primary productivity GPP, and respiration R) based on upscaling eddy-covariance measurements to global scale. In the FLUXCOM project (fluxcom.org, and its predecessors), machine-learning methods use satellite observations describing the state of the land surface and meteorological information inferred from space or reanalysis to upscale relationships between CO2 fluxes and environmental conditions learned at site level to the globe [6-9]. Simulations have been done on progressively finer temporal resolutions from monthly to daily [6-9]. A multitude of carbon cycle related studies analysed these machine learning estimates as a complementary means to process based model results and for comparisons to ‘top-down’ approaches of atmospheric inversions. Recent work developed a method to produce the first global subdaily estimates of carbon fluxes for every half hour, despite having available predictors only at daily resolution [10]. The ability to resolve the diurnal cycle is a crucial step required for the correct interpretation of snapshot-like satellite images of atmospheric CO2 or land surface variables usually taken at a certain time of the day. The subdaily flux estimates can also serve as priors for atmospheric inversions or to check their consistency. We further have extended the same method by using additional hourly meteorological information from the newly released ERA5 reanalysis data as explanatory variables. Simulated diurnal carbon fluxes show higher subdaily variability, even partly resolving changes in the shape of the diurnal cycle. We will: 1) present these global data sets, their characteristics, as well as known strengths and limitations, 2) show comparisons to atmospheric inversions and other independent data sets, 3) inform on current efforts to further improve and extend the data-driven CO2 flux estimates in terms of quality, spatiotemporal resolution and driving variables.




[1] Le Quéré et al. 2017, Global Carbon Budget 2017, Earth System Science Data Discussions, 2017 1–79, doi:10.5194/essd-2017-123
[2] Keeling et al.1995, Interannual extremes in the rate of rise of atmospheric carbon dioxide since 1980, Nature, 375 666– 670, doi:10.1038/375666a0
[3] Reuter et al. 2017, How Much CO2 Is Taken Up by the European Terrestrial Biosphere?, Bulletin of the American Meteorological Society, 98(4) 665–671, doi:10.1175/BAMS-D-15-00310.1
[4] Ciais et al. 2013, Carbon and Other Biogeochemical Cycles, in Climate Change 2013: The Physical Science Basis. Contribution of Working Group I to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change [Stocker, T.F., D. Qin, G.- K. Plattner, M. Tignor, S.K. Allen, J. Boschung, A. Nauels, Y. Xia, V. Bex and P.M. Midgley (eds.), 465–570, Cambridge University Press, Cambridge, United Kingdom and New York, NY, USA, doi:10.1017/CBO9781107415324.015
[5] Schimel et al. 2001, Recent patterns and mechanisms of carbon exchange by terrestrial ecosystems, Nature, 414, 10.1038/35102500
[6] Jung et al. 2009, Towards global empirical upscaling of FLUXNET eddy covariance observations: validation of a model tree ensemble approach using a biosphere model, Biogeosciences, 6(10) 2001–2013, doi:10.5194/bg-6-2001-2009
[7] Beer et al. 2010, Terrestrial Gross Carbon Dioxide Uptake: Global Distribution and Covariation with Climate, Science, 329(5993) 834–838, doi:10.1126/science.1184984
[8] Jung et al. 2011, Global patterns of land-atmosphere fluxes of carbon dioxide, latent heat, and sensible heat derived from eddy covariance, satellite, and meteorological observations, Journal of Geophysical Research:Biogeosciences, 116(G3), doi:10.1029/2010JG001566, g00J07
[9] Tramontana et al. 2016, Predicting carbon dioxide and energy fluxes across global FLUXNET sites with regression algorithms, Biogeosciences, 13(14) 4291–4313, doi:10.5194/bg-13-4291-2016
[10] Bodesheim et al. 2018, Upscaled diurnal cycles of land-atmosphere fluxes: a new global half- hourly data product, Earth System Science Data Discussions, 2018 1–47, doi:10.5194/essd- 2017-130.

Area A - Board 37

Novel Earth Observations of atmospheric greenhouse gases and the cryosphere have the potential to fundamentally increase our understanding of the carbon cycle specifically at high Northern latitudes. In this work, we present our newly started ESA project “MethEO” on quantifying methane (CH4) emissions in the Northern Hemisphere, and investigating the connection of the fluxes to the soil freezing and thawing at boreal and Arctic latitudes. We combine methods for the quantification of CH4 emissions by applying data from several Earth Observing (EO) satellites, ground-based retrievals, in-situ measurements, and global atmospheric methane inversion model estimates. The EO data consist of a global soil freeze/thaw estimate obtained from the ESA Soil Moisture and Ocean Salinity (SMOS) mission as well as retrievals of atmospheric column-averaged methane obtained from the Greenhouse Gases Observing Satellite (GOSAT) and Sentinel 5 Precursor TROPOMI (S5P-TROPOMI) observations. These EO data will be used in global atmospheric methane inversion model, CarbonTracker Europe – CH4, simulations, focusing on (1) the identification of CH4 sources in the Northern Hemisphere and (2) providing trend analysis on the total methane emissions in the past two decades. EO data will be used to assess the spatial variability of the emissions and to better quantify the contributions from regions dominated by anthropogenic emissions and by natural emissions, especially those from wetlands. Further, EO data will be used to create proxies of the seasonality of the natural methane emissions, focusing on the timing and length of the autumn freezing period and springtime melting period. The combined use of several Earth Observation data sets together with inverse modeling will increase the accuracy of the methane flux estimates at the remote Arctic and boreal regions where in-situ measurements are still rare, and also lead to a better understanding of the longer-term link between the changes in the soil conditions and the resulting methane fluxes.

Area A - Board 38

Limiting global warming to below 2 degrees Celsius as agreed upon in the Paris agreement requires reduction in fossil fuel emissions, however, the magnitude of 'allowable' emissions is uncertain. The reason for this being the uncertainty in the uptake of atmospheric CO₂ by the terrestrial biosphere, and its response to climate change. Process-based terrestrial carbon cycle models are the most sophisticated tools for quantifying this uptake but they suffer from, among others, parametric uncertainty limiting our ability to provide reliable simulations of the carbon uptake.

The ESA-funded 'SMOS + Vegetation' project combines a retrieval component that aims at further improving the Soil Moisture and Ocean Salinity (SMOS) Vegetation Optical Depth (VOD) product with a data assimilation component that demonstrates the added value of this product, in combination with the SMOS soil moisture product and atmospheric CO₂ observations, in constraining simulated land surface fluxes of carbon dioxide. This contribution focuses on the project's modelling and assimilation component.

The simultaneous assimilation of multiple data streams of such different nature constitutes a significant scientific and technological challenge. We describe the construction of dedicated observation operators that link the state of the terrestrial biosphere model to simulated VOD, surface layer soil moisture and atmospheric CO₂. We present our Carbon Cycle Data Assimilation System (CCDAS) around the terrestrial biosphere model BETHY and demonstrate its operation through simultaneous assimilation of the SMOS products together with atmospheric CO₂ at site and global scale. We use soil moisture and L-VOD products that are jointly retrieved from SMOS observations using the SMOS INRA-CESBIO (SMOS-IC) algorithm.

CCDAS operates through minimisation of a cost function that achieves a balance between the observational constraints and the prior information. The minimum is determined by varying, in an iterative minimisation procedure, a control vector composed of the model's process parameters and initial conditions. Posterior uncertainties on parameters are derived from an approximation of the cost function's second derivative (Hessian) and then propagated onto simulated net and gross carbon fluxes through the linearised model.

We show that, with suitable observation operators, the BETHY model is capable of absorbing the information content in the SMOS IC products and that its assimilation tends to improve the consistency with independent observations such as atmospheric CO₂ withheld from the assimilation, fraction of absorbed photosynthetically active radiation, or in-situ observations of soil moisture and carbon fluxes.

Among others, the assimilation system delivers (as a SMOS Level 4 product) monthly fields of net and gross terrestrial carbon fluxes with uncertainty ranges at a spatial resolution of 0.25°. The impact of SMOS is illustrated by Figure 1, which compares preliminary results of net terrestrial carbon fluxes after assimilation of atmospheric CO₂ only and combined with SMOS L-VOD and soil moisture.

Area A - Board 39

Forest Flux is an Innovation Action project of Horizon 2020 research program of the European Union. The outcome of the project will revolutionize forestry value added services in Earth Observation (EO) by creating and piloting a cloud-based service for committed users on forest carbon assimilation and structural variable prediction using Copernicus data. The revolution is driven by sustainable forest management and anticipated by EU forest strategy, the Bioeconomy Action Pan, and the demands of environmentally aware end-users of wood industry products.

Until recently, detailed information on forest carbon cycle has not been available due to a lack of scientific understanding, spatial
data availability, limited processing capacity, and the complexity of implementing this information in business processes. Forest Flux will use the
explosive increase in high-resolution EO data by Copernicus program and developments of cloud computing technology. It will implement worldfirst
service platform for high-resolution maps of traditional forestry variables together with forest carbon fluxes using a holistic approach in a single
processing chain. By the end of the project, forestry and carbon data will be completely integrated into the decision-making processes of selected
core users and establish the leadership of European industry in sustainable utilization of forest resources. The relevant algorithms have been
recently demonstrated by consortium partners. The computing infrastructure is specifically targeted for EO data and forestry users, and will be fully
functional by the end of the project. The web-based human and machine interfaces will allow market access unrestricted by country boundaries
and facilitate easy commercial interactions of players of different sizes and backgrounds. Dedicated user involvement, strong commercial
interests, rapidly developing online markets, and demonstrated excellence of the consortium make the Forest Flux service platform sustainable
beyond the end of the project.

The Forest Flux system will be implemented of the Forest Thematic Exploitation Platform F-TEP that is using a DIAS infrastructure.

Area A - Board 40

Fire is an ubiquitous form of disturbance, playing an important role in the carbon cycle. Frequent observations from Sentinel-2 (S2) and/or Landsat-8 (LC8) provide an opportunity to observe the change of reflectance due to the fire. The goal of this contribution is to introduce a linear model of reflectance change due to fire, coupled with a spectral model of burnt surfaces, that quantifies the change of reflectance due to short-term fire effects.

Starting from a statement of the post-fire reflectance being a linear combination of a burnt and an unburnt signal, modulated by the fraction of the fire affected by the pixel (f) as well as by the radiometric combustion completeness (cc), we arrive at a very simple model of reflectance change as a function of f*cc and the spectral burnt and unburnt signals.

Assuming the pre-fire reflectance can be used as the unburnt signal, we develop a spectral model of the burnt signal, derived from measurements of typical elements in burnt scenes: char, ash, exposed (dry) soils, as well as charred wood, leaves, etc. The proposed spectral model can be expressed simply by a sum of two basis functions: a0, a char/ash spectral component and a1, a dry soil component. Plugging the spectral model in the reflectance change model results in a linear model of reflectance controlled by fcc, a0 and a1.

The fcc model is illustrated with data from MODIS, Sentinel-2 and Landsat 8 over some fires, showing that the retrieved patterns of fcc, a0 and a1 are comparable. We further show time series of fcc over different fire-affected regions, showing that fcc has a clear seasonality that is strongly correlated with fuel moisture dynamics. Further, over all biomes, we show a clear link between the relative drop in effective leaf area index (eLAI) due to the fire (obtained by inverting a radiative transfer model before and after the fire) and fcc for MODIS data, paving the way for including this metric in e.g. fire models embedded in dynamic global vegetation models (DGVMs).

As a form of validation, we compare the MODIS retrieved fcc with the fraction of the MODIS pixel affected by the fire (f). The latter dataset is produced from a large number of manually-interpreted burnscars derived from Landsat data, covering a large number of fire systems. Results that fcc from MODIS is highly correlated with the underlying f, consistently more than other popular metrics such as deltaNBR or deltaNDVI.

Area A - Board 41