Modeling and sensitivity analysis of fire emissions in southern Africa during SAFARI 2000

doi:10.1016/j.rse.2004.06.010

Remote Sensing of Environment

Volume 92, Issue 2, 15 August 2004, Pages 255-275

https://doi.org/10.1016/j.rse.2004.06.010 Get rights and content

Abstract

This paper uses three recently generated southern African satellite burned area products for the month of September 2000 in a sensitivity study of regional biomass burning emissions for a number of trace gases and particulates. Differences in the extent and location of areas burned among products generated from Moderate Resolution Imaging Spectroradiometer (MODIS), Systeme Pour l'Observation de la Terre (SPOT-VEGETATION), and Along Track Scanning Radiometer (ATSR-2) data are significant and result in different emissions estimates for woodland and grassland land cover types. Due to the different emission profiles in woodlands and grasslands, favoring relatively more products of incomplete combustion in woodlands compared with products of complete combustion in grasslands in the late dry season, these changes are not proportional to the differences in the burned area amounts. The importance of accurate burned area information not just in terms of the total area but also in terms of its spatial distribution becomes apparent from our modeling results. This paper highlights the urgent need for satellite data producers to provide accuracy assessments associated with satellite-derived products. Preferably, these accuracy data will be spatially explicit, or defined in a way that can be applied in a spatially explicit modeling context, to enable emissions uncertainties to be defined with respect to different landscape units in support of greenhouse gas emissions reporting.

Introduction

Reliable burned area information is required by a number of users, including global change scientists modeling the source strength, transport, fate, and impacts of trace gas and aerosol emissions from vegetation fires. National governments are required to report their greenhouse gas emissions. Natural resources managers and policy makers can benefit from a synoptic view of fire activity. The savannas of Africa are thought to experience the most extensive biomass burning in the world, for example, contributing an estimated 49% of the carbon lost from fires worldwide Scholes & Andreae, 2000, Dwyer et al., 2000, Van der Werf et al., 2003. Savanna fires in southern Africa play an important role in the regional atmospheric distribution of trace gases and aerosols, ecosystem functioning, and biogeochemical cycling Frost & Robertson, 1987, Scholes et al., 1996.

The Southern African Regional Science Initiative (SAFARI 2000) was an international scientific campaign aimed to study land–atmosphere interactions in southern Africa (Swap et al., 2003). One of the main objectives of SAFARI 2000 was the characterization and quantification of regional emissions sources, including those from savanna burning. Within this context, we have developed a spatially explicit emissions model, which we apply here to estimate the pyrogenic emissions for a number of trace gases and particulates during the dry season field campaign in September 2000. In this paper, we perform a sensitivity analysis to examine the geospatial relationship between burned area and emissions. Three recently developed moderate-resolution satellite burned area products from the Moderate Resolution Imaging Spectroradiometer (MODIS), Systeme Pour l'Observation de la Terre (SPOT-VEGETATION), and Along Track Scanning Radiometer (ATSR-2) sensors provide the basis for generating and comparing new estimates of biomass burning emissions from woodland and grassland fires in southern Africa during September 2000. An emissions sensitivity analysis with respect to the woodland–grassland model parameterization is also performed.

Section snippets

Fire emissions modeling methodology

Estimates of pyrogenic gas emissions require information on burned area, fuel load (the amount of fuel per unit area), completeness of combustion (the proportion of the fuel consumed during the fire), and emission factors (characteristic of the amount of the specific atmospheric species produced during the burning). Emissions are computed using spatially explicit input data following the well-established Seiler and Crutzen (1980) model: $E_{x} = ∑ i,j A_{i,j} F_{i,j} CC_{i,j} (EF_{x})_{i,j}$ where E_x=total pyrogenic

Intercomparison of MODIS, GBA-2000, and GLOBSCAR burned area products, southern Africa, September 2000

Table 1, Table 2 summarize the regional differences between the MODIS, GBA-2000, and GLOBSCAR burned area products for September 2000. The total southern African area considered is 9,269,225 km². MODIS detects a regional September burned area amount of approximately 268,500 km², GBA-2000 detects approximately 173,100 km², and GLOBSCAR detects approximately 59,800 km². The MODIS active fire product detects an approximate area of 126,400 km², although as we noted above, this estimate is expected

Conclusions

This sensitivity study contributes to a better understanding of the relationship between the spatial distribution of burned areas and emissions. The model's predictions, regardless of the burned area product, are consistent with published field studies of emission patterns from woodland and grassland fires Hoffa et al., 1999, Ward et al., 1996. These published studies show that burning in woodlands favors emissions of products of incomplete combustion compared with grasslands. Changes in the

Acknowledgements

This study was carried out as part of SAFARI-2000. Funding in partial support of this research was provided by National Aeronautics and Space Administration (NASA) grants NAS-531365 and NAG-511251. Stefania Korontzi is also supported by the NASA's Earth System Graduate Research Fellowship. The anonymous reviewers are thanked for their helpful comments.

References (47)

B. Guenther et al.
On-orbit performance of the earth observing system Moderate Resolution Imaging Spectroradiometer; first year of data
Remote Sensing of Environment
(2002)
M.C. Hansen et al.
Towards an operational MODIS continuous field of percent tree cover algorithm: Examples using AVHRR and MODIS data
Remote Sensing of Environment
(2002)
C.O. Justice et al.
The MODIS fire products
Remote Sensing of Environment
(2002)
C.O. Justice et al.
An overview of MODIS Land data processing and product status
Remote Sensing of Environment
(2002)
S. Korontzi et al.
Influence of timing and spatial extent of vegetation fires in southern Africa on atmospheric emissions
Journal of Arid Environments
(2003)
J.T. Morisette et al.
A framework for the validation of MODIS land products
Remote Sensing of Environment
(2002)
D.P. Roy et al.
The MODIS land product quality assessment approach
Remote Sensing of Environment
(2002)
D.P. Roy et al.
Burned area mapping using multi-temporal moderate spatial resolution data—A bi-directional reflectance model-based expectation approach
Remote Sensing of Environment
(2002)
D. Stroppiana et al.
Radiometric analysis of SPOT-VEGETATION images for burnt area detection in northern Australia
Remote Sensing of Environment
(2002)
M.O. Andreae
Emissions of trace gases and aerosols from southern African savanna fires

M.O. Andreae et al.

Emissions of trace gases and aerosols from biomass burning

Global Biogeochemical Cycles

(2001)

I. Bertschi et al.

Trace gas and particle emissions from fires in large diameter and belowground biomass fuels

Journal of Geophysical Research

(2003)

B.V. Braatz et al.

African greenhouse gas emission inventories and mitigation options: Forestry, land-use change, and agriculture

Environmental Monitoring and Assessment

(1995)

O. Dikshit et al.

An empirical investigation of image resampling effects upon the spectral and textural information content of a high spatial resolution multispectral image

Photogrammetric Engineering and Remote Sensing

(1996)

E. Dwyer et al.

Characterization of the spatio-temporal patterns of global fire activity using satellite imagery for the period April 1992 to March 1993

Journal of Biogeography

(2000)

H. Eva et al.

Burnt area mapping in Central Africa using ATSR data

International Journal of Remote Sensing

(1998)

P.G.H. Frost

The ecology of miombo woodlands

P.G.H. Frost et al.

The ecological role of fire in savannas

Food and Agriculture Organization of the United Nations

Global forest resources assessment 2000 (FRA 2000): Appendix 2. Terms and definitions

J.-M. Grégoire et al.

The GBA-2000 initiative: Developing a global burned area database from SPOT VEGETATION imagery

International Journal of Remote Sensing

(2003)

M.C. Hansen et al.

The MODIS 500 meter global vegetation continuous field products

Hély, C., Caylor, K. K., Dowty, P. R., Swap, R. J., & Shugart, H. H. (2003a). SAFARI 2000 Modeled Fuel Load in Southern...

C. Hély et al.

Regional fuel load for two climatically contrasting years in southern Africa

Journal of Geophysical Research

(2003)

Cited by (54)

Opportunities and challenges for savanna burning emissions abatement in southern Africa
2021, Journal of Environmental Management
Citation Excerpt :
Under such Australian cured fuel conditions, EF(CH4) and EF(N2O) have been shown to be equivalent in EDS and LDS periods (Hurst et al., 1994, 1996; Meyer et al., 2012), albeit exhibiting substantial variation associated with different vegetation (VFT) and fuel types (FTc) (Meyer et al., 2012). Conversely, various African savanna studies have demonstrated that EF(CH4) from Miombo woodlands and seasonally inundated dambo grasslands are strongly influenced by fuel moisture content (Hoffa et al., 1999; Korontzi et al., 2003a, 2003b, 2004; Korontzi, 2005). Undertaking prescribed EDS fire management under such enhanced moist fuel conditions would necessarily confound the putative emissions abatement benefits of the proposed method.
Savanna fires occurring in sub-Saharan Africa account for over 60% of global fire extent, of which more than half occurs in the Southern Hemisphere contributing ~29% of global fire emissions. Building on experience in reducing savanna fire emissions in fire-prone north Australian savannas through implementation of an internationally accredited ‘savanna burning’ emissions abatement methodology, we explore opportunities and challenges associated with the application of a similar approach to incentivise emissions reduction in fire-prone southern African savannas. We first show that for a focal region covering seven contiguous countries, at least 80% of annual savanna large fire (>250 ha) extent and emissions occur under relatively severe late dry season (LDS) fire-weather conditions, predominantly in sparsely inhabited areas. We then assess the feasibility of adapting the Australian emissions abatement methodology through exploratory field studies at the Tsodilo Hills World Heritage site in north-west Botswana, and the Niassa Special Reserve in northern Mozambique. Our assessment demonstrates that application of a savanna burning emissions abatement method focused on the undertaking of strategically located early dry season (EDS) burning to reduce LDS wildfire extent and resultant emissions meets key technical criteria, including: LDS fine fuels tend to be markedly greater than EDS fuels given seasonal leaf litter inputs; LDS fires tend to be significantly more severe and combust more fuels; methane and nitrous oxide emission factors are essentially equivalent in EDS and LDS periods under cured fuel conditions. In discussion we consider associated key implementation challenges and caveats that need to be addressed for progressing development of savanna burning methods that incentivise sustainable fire management, reduce emissions, and support community livelihoods in wildfire-dominated southern African savannas.
Advances in the estimation of high Spatio-temporal resolution pan-African top-down biomass burning emissions made using geostationary fire radiative power (FRP) and MAIAC aerosol optical depth (AOD) data
2020, Remote Sensing of Environment
Citation Excerpt :
The 2015 European Space Agency (ESA) Climate Change Initiative (CCI) Landcover map (Validated by ESA (2017)) is derived from 300 m spatial resolution PROBA-V observations and comprises 36 landcover type, which we aggregate into five distinct biome classes. Following Mota and Wooster (2018), grassland and woodland savanna are classified as separate biomes as suggested by Korontzi et al. (2004) for fire-related GHG emission reporting. Full details of the CCI land cover class assignment for the FREMv2 biomes can be found in Appendix A.
We provide major updates to the ‘top down’ Fire Radiative Energy Emissions (FREM) approach to biomass burning emissions calculations, bypassing the estimation of fuel consumption that is a major source of uncertainty in widely used ‘bottom up’ approaches. The FREM approach links satellite observations of fire radiative power (FRP) to emission rates of total particulate matter (TPM) via spatially varying smoke emissions coefficients (g.MJ⁻¹) – each derived from matchups of FRP and smoke plume aerosol optical depth (AOD). In the original FREMv1 approach, FRP data came from the geostationary Meteosat satellite and AOD data from the 10 km spatial resolution MODIS MOD04 aerosol product. However, the latter often performs quite poorly close to biomass burning sources due to its large 10 km pixels, bias at high MODIS view zenith angles, and saturation and/or removal of areas of high AOD - limitations introducing bias and uncertainty into the final FREM-derived smoke emissions estimates. We address each of these issues through a series of significant methodological and input data improvements, including exploitation of the 1 km MODIS MAIAC AOD product that performs far better close to fire sources. We use our FREMv2 methodology to generate a new pan-African fire emissions inventory for TPM and the carbonaceous gases CO₂, CO and CH₄, and our annual mean TPM emissions are within 11% of those of the MODIS-based FEER top-down approach, but significantly higher than those of GFASv1.2 and GFEDv4.1s (by 114% and 69% respectively) - agreeing with independent assessments that aerosol emissions of GFASv1.2 require upscaling by a factor of 2 to 3.4 to deliver matching magnitudes between modelled and observed AODs. From our carbonaceous emissions totals we map dry matter consumed (DMC) across Africa, and dividing this by the FireCCISFD11 20 m burned area product we provide one of the first data-driven pan-African maps of fuel consumption per unit area (kg.m⁻²) which in many areas is higher than in GFEDv4.1s. Our estimates represent the highest spatio-temporal resolution biomass burning emissions data yet available over Africa, and significantly advance the aim of a pan-tropical and mid-latitude inventory based on FRP from the global geostationary satellite network (Meteosat, Meteosat IOD, GOES and Himawari).
Detection rates and biases of fire observations from MODIS and agency reports in the conterminous United States
2019, Remote Sensing of Environment
Citation Excerpt :
However, estimates by different satellite fire products of fire counts and area burned vary by thousands of hectares (e.g. Chang and Song, 2009; Loepfe et al., 2012), suggesting these sensors are detecting different types of fire events, or providing different representations of the same events due to omission and commission errors. Attempts have been made to validate MODIS fire products primarily using other remotely sensed data (e.g. Korontzi et al., 2004; Padilla et al., 2015; Schroeder et al., 2008), however, a lack of consistent ground-based data has made independent assessment from ground-based fire observations difficult (e.g. Boschetti et al., 2016; Hawbaker et al., 2008). With wildfires increasing over large portions of the conterminous U.S. (Westerling, 2016; Westerling et al., 2006), a better understanding of the limitations of satellite fire products is needed.
With growing concern about the impacts of fires on ecosystems and economies, satellite products are increasingly being used to understand fire regimes. Concurrently, where available, agency records of fires have also been used to assess fire regimes. Yet, it remains unclear if these independent datasets measure the same fires, which raises concerns about the interpretation and benchmarking of models derived from these products. Here, we present a novel product intercomparison of the MODIS burned area and active fire products across the conterminous United States using nearly 250,000 agency reported wildfires as reference data to model consistencies and inconsistencies between all three datasets. We compared agency reported wildfires from the Fire Program Analysis fire occurrence database to the MODIS products to identify which fires were detected vs. omitted by MODIS products relative to agency fire records, and by agency fire records relative to MODIS. We created generalized linear models as a function of fire attributes (e.g. size) and environmental variables (e.g. cloud cover) to predict MODIS detection of agency wildfires, and anthropogenic variables (e.g. agriculture) to predict agency detection of MODIS fires. We modeled fire detection probability separately for MODIS burned area and active fire products, and for the eastern and western U.S. Overall, we found that MODIS product detection rates ranged from 3.5% to 23.4% of all documented agency wildfires >1 ha, and that likelihood of detection increased with fire size. Agency detection rates ranged from 23.5% to 48% of MODIS burned area and active fires. Under ideal conditions, the MODIS active fire product had a 50% probability of detecting a wildfire that grew to at least 10 ha (eastern U.S.) – 78 ha (western U.S.), while the burned area product had a 50% probability of detecting a wildfire that grew to at least 169 ha (eastern U.S.) – 234 ha (western U.S.). Cloud cover and leaf area index were significant predictors of MODIS fire detection, while state boundaries were significant predictors of agency fire detection. This analysis presents an important assessment of the fire attributes and ground conditions that influence MODIS fire detection relative to extensive and increasingly used ground-based wildfire records. The large discrepancy in records of fire occurrence between MODIS and agency fire datasets highlights the need for this type of analysis into the types of fires likely to be included in each database.
A new top-down approach for directly estimating biomass burning emissions and fuel consumption rates and totals from geostationary satellite fire radiative power (FRP)
2018, Remote Sensing of Environment
Regional to global-scale biomass burning emissions inventories are primarily based on satellite-derived burned area or fire radiative power (FRP), and most rely on conversions to fuel consumption prior to the emissions estimation stage. This is generally considered the step introducing greatest uncertainty, and some apparently discrete inventories are not fully independent, as they have been cross-calibrated to aid this stage. We present a novel emissions inventory approach that bypasses the fuel consumption step, directly linking geostationary FRP measures to emission rates of total particulate matter (TPM), via coefficients derived from observations of smoke plume aerosol optical depth (AOD). The approach is fully ‘top-down’, being based on spaceborne observations alone, is performed at or close to the FRP data's original pixel resolution, and avoids the need to assume or model fuel consumption per unit area prior to the emissions calculation. Rates and totals of trace gas and carbon emission can be inferred from the TPM fluxes, and in combination with satellite burned area (BA) products the approach provides an innovative top down approach to mapping fuel consumption per unit area (kg·m^− 2) as a last step in the calculation. Using this innovative methodology, which we term ‘FRE_emissions’ (FREM), we generate a 2004–2012 fire emissions inventory for southern Africa, based on Meteosat FRP-PIXEL data. We find basic annual average TPM emissions 45% higher than the widely used GFASv1.2 inventory, with our higher totals in line with independent assessments that necessitate a significant upscaling of GFAS TPM emissions to match observed AODs. Our estimates are also 12% higher than GFEDv4.1s, which already includes a substantial upward adjustment for fires too small to be detected by the MODIS MCD64A1 BA product. If we adjust the FREM-derived emissions for SEVIRI's inability to detect the lower FRP component of the regions fire regime then the differences between FREM and GFAS/GFED grow further, to a mean of 64% with respect to GFED4.1s TPM emissions for example. These upwardly adjusted FREM estimates agree very well with FEER, an FRP- and AOD-based inventory driven by polar-orbiting MODIS FRP ‘snapshots’ rather than geostationary observations. Similarly higher totals are seen for FREM's fire-emitted trace gases, derived using the emission factor ratios of gases to particulates. Our exploitation of geostationary FRP requires fewer assumptions than use of polar orbiter FRP measures, avoids biases coming from incomplete sampling of the fire diurnal cycle, and enables the FREM approach to provide fire emissions and fuel consumption estimates at a higher spatio-temporal resolution than any inventory currently available (e.g. 0.05°, and hourly averages or better), including per km² of area burned. The approach offers great potential to generate very high resolution fire emissions datasets for the tropics, sub-tropics and potentially temperate zones, with updates available in near real-time from the global suite of geostationary meteorological satellites operated by organisations such as EUMETSAT (Meteosat), NOAA (GOES) and JMA (Himawari).
A stratified random sampling design in space and time for regional to global scale burned area product validation
2016, Remote Sensing of Environment
The potential research, policy and management applications of global burned area products place a high priority on rigorous, quantitative assessment of their accuracy. Such an assessment can be achieved by implementing validation methods employing design-based inference in which the independent reference data are selected via a probability sampling design. The majority of global burned area validation exercises use Landsat data to derive the independent reference data. This paper presents a three-dimensional sampling grid that allows for probability sampling of Landsat data in both space and time. To sample the globe in the spatial domain with non-overlapping sampling units, the Thiessen Scene Area (TSA) tessellation of the Landsat path/row geometry is used. The TSA grid is combined in time with the 16-day Landsat acquisition calendar to provide three-dimensional elements (voxels). This allows for implementation of stratified random sampling designs, where not only the location but also the time interval of the independent reference data is explicitly drawn by probability sampling. To illustrate this, we use a stratification methodology based on the Olson global ecoregion map and on the MODIS global active fire product. Using the global MODIS burned area product to establish a hypothetical population of reference data, we show that a sampling scheme based on the proposed stratification with equal sample allocation among strata is effective in reducing the standard errors of accuracy and area estimators compared to simple random sampling. Globally, the standard errors were reduced by 63%, 54%, 22% and 53% for overall accuracy, omission error, commission error and total burned area estimates respectively. By incorporating probability sampling in both the spatial and temporal domains, the present study establishes the foundation for rigorous design-based validation of global burned area products and, more generally, of terrestrial thematic products that have high temporal variability.
Ten years of global burned area products from spaceborne remote sensing-A review: Analysis of user needs and recommendations for future developments
2014, International Journal of Applied Earth Observation and Geoinformation
Early global estimates of carbon emissions from biomass burning were based on empirical assumptions of fire return interval in different biomes in the 1980s. Since then, significant improvements of spaceborne remote sensing sensors have resulted in an increasing number of derived products characterizing the detection of active fire or the subsequent burned area (GFED, MODIS MCD45A1, L3JRC, Globcarbon, GBS, GLOBSCAR, GBA2000). When coupled with global land cover and vegetation models allowing for spatially explicit fuel biomass estimates, the use of these products helps to yield important information about the spatial and the temporal variability of emission estimates. The availability of multi-year products (>10 years) leads to a better understanding of uncertainties in addition to increasing accuracy. We surveyed a wide range of users of global fire data products whilst also undertaking a review of the latest scientific literature. Two user groups were identified, the first being global climate and vegetation modellers and the second being regional land managers. Based on this review, we present here the current needs covering the range of end-users. We identified the increasing use of BA products since the year 2000 with an increasing use of MODIS as a reference dataset. Scientific topics using these BA products have increased in diversity and area of application, from global fire emissions (for which BA products were initially developed) to regional studies with increasing use for ecosystem management planning. There is a significant need from the atmospheric science community for low spatial resolution (gridded, 1/2 degree cell) and long time series data characterized with supplementary information concerning the accuracy in timing of the fire and reductions of omission/commission errors. There is also a strong need for precisely characterizing the perimeter and contour of the fire scar for better assimilation with land cover maps and fire intensity. Computer and earth observation facilities remain a significant gap between ideal accuracies and the realistic ones, which must be fully quantified and comprehensive for an actual use in global fire emissions or regional land management studies.

View all citing articles on Scopus

View full text

Modeling and sensitivity analysis of fire emissions in southern Africa during SAFARI 2000

Abstract

Introduction

Section snippets

Fire emissions modeling methodology

Intercomparison of MODIS, GBA-2000, and GLOBSCAR burned area products, southern Africa, September 2000

Conclusions

Acknowledgements

Remote Sensing of Environment

Remote Sensing of Environment

Remote Sensing of Environment

Remote Sensing of Environment

Journal of Arid Environments

Remote Sensing of Environment

Remote Sensing of Environment

Remote Sensing of Environment

Remote Sensing of Environment

Emissions of trace gases and aerosols from southern African savanna fires

Emissions of trace gases and aerosols from biomass burning

Global Biogeochemical Cycles

Trace gas and particle emissions from fires in large diameter and belowground biomass fuels

Journal of Geophysical Research

African greenhouse gas emission inventories and mitigation options: Forestry, land-use change, and agriculture

Environmental Monitoring and Assessment

An empirical investigation of image resampling effects upon the spectral and textural information content of a high spatial resolution multispectral image

Photogrammetric Engineering and Remote Sensing

Characterization of the spatio-temporal patterns of global fire activity using satellite imagery for the period April 1992 to March 1993

Journal of Biogeography

Burnt area mapping in Central Africa using ATSR data

International Journal of Remote Sensing

The ecology of miombo woodlands

The ecological role of fire in savannas

Global forest resources assessment 2000 (FRA 2000): Appendix 2. Terms and definitions

The GBA-2000 initiative: Developing a global burned area database from SPOT VEGETATION imagery

International Journal of Remote Sensing

The MODIS 500 meter global vegetation continuous field products

Regional fuel load for two climatically contrasting years in southern Africa

Journal of Geophysical Research