remote sensing of burned areas
DESCRIPTION
RIEL seminar presentation given by Emilio Chuvieco a Professor of Geography at the University of Alcalá, SpainTRANSCRIPT
![Page 1: Remote sensing of burned areas](https://reader034.vdocuments.site/reader034/viewer/2022052302/568c48601a28ab49168fe73e/html5/thumbnails/1.jpg)
Click to edit Master title style
Earth Observation of Burned Areas
Emilio Chuvieco Department of Geography, University of Alcalá (Spain)
![Page 2: Remote sensing of burned areas](https://reader034.vdocuments.site/reader034/viewer/2022052302/568c48601a28ab49168fe73e/html5/thumbnails/2.jpg)
Click to edit Master title style
Main University Building: 1502
![Page 3: Remote sensing of burned areas](https://reader034.vdocuments.site/reader034/viewer/2022052302/568c48601a28ab49168fe73e/html5/thumbnails/3.jpg)
Click to edit Master title style
Our city
![Page 4: Remote sensing of burned areas](https://reader034.vdocuments.site/reader034/viewer/2022052302/568c48601a28ab49168fe73e/html5/thumbnails/4.jpg)
Click to edit Master title style Our “kangaroos”
![Page 5: Remote sensing of burned areas](https://reader034.vdocuments.site/reader034/viewer/2022052302/568c48601a28ab49168fe73e/html5/thumbnails/5.jpg)
Click to edit Master title style Activities of UAH research group
• Fire risk estimation:
– Generation of input variables: Fuel type classification, Lidar, FMC, Socio-economic variables.
– Integration & validation methods.
• Fire effects assessment:
– Mapping burned areas at global and regional scales.
– Burn severity estimation.
– Input of Burned Area into Global Vegetation Models.
![Page 6: Remote sensing of burned areas](https://reader034.vdocuments.site/reader034/viewer/2022052302/568c48601a28ab49168fe73e/html5/thumbnails/6.jpg)
Click to edit Master title style Outline of the seminar
• Global mapping of burned areas:
– Background
– ESA fire_cci project
• Approaches to map burn severity from RS data.
![Page 7: Remote sensing of burned areas](https://reader034.vdocuments.site/reader034/viewer/2022052302/568c48601a28ab49168fe73e/html5/thumbnails/7.jpg)
Click to edit Master title style ESA-CCI programme
Aerosol cci Cloud cci
Fire cci
GHG cci
Glaciers cci
Land Cover cci
Ocean Colour cci
Ozone cci
Sea-level cci
Sea Ice cci
Sea Surface Temperature cci
Ice Sheets
CMUG
Soil moisture
![Page 8: Remote sensing of burned areas](https://reader034.vdocuments.site/reader034/viewer/2022052302/568c48601a28ab49168fe73e/html5/thumbnails/8.jpg)
Click to edit Master title style Fire_cci science context
• Fire affects:
– GHG and aerosol emissions.
– Carbon budgets and vegetation cycles.
– Land cover change (defforestation)
• Fire is affected by:
– Temperature-rainfall trends, particularly heat waves and “El Niño” episodes (climate prediction)
– Socio-economic changes (land use policy).
![Page 9: Remote sensing of burned areas](https://reader034.vdocuments.site/reader034/viewer/2022052302/568c48601a28ab49168fe73e/html5/thumbnails/9.jpg)
Click to edit Master title style Science questions
• What are the recent trends in fire activity?
• What factors are behind fire occurrence?
• What is the actual magnitude of fire impacts?
– How much area is burned annually?
– How much biomass is actually consumed?
– What is the combustion efficiency (CO/CO2)?
– What is the role of fire in carbon accounting? Is biomass burning “carbon neutral”?
![Page 10: Remote sensing of burned areas](https://reader034.vdocuments.site/reader034/viewer/2022052302/568c48601a28ab49168fe73e/html5/thumbnails/10.jpg)
Click to edit Master title style ¿How much area is burned every year?
• Inconsistencies between RS products and official forest fire statistics.
• Inconsistencies between RS products.
• Internal uncertainty of each RS product.
![Page 11: Remote sensing of burned areas](https://reader034.vdocuments.site/reader034/viewer/2022052302/568c48601a28ab49168fe73e/html5/thumbnails/11.jpg)
Click to edit Master title style
FRA2010 GVED v3
FAO (FRA2010): 0.6 Mkm². Only 78 countries are covered.
Average 4 Mkm²
![Page 12: Remote sensing of burned areas](https://reader034.vdocuments.site/reader034/viewer/2022052302/568c48601a28ab49168fe73e/html5/thumbnails/12.jpg)
Click to edit Master title style Different EO BA estimations
% of BA from different satellite products
Red: over estimation
Blue: under estimation
(Giglio et al., 2010).
-L3JRC: 3.5 - 4.5 Mkm²
(2000-07)
-MCD45 c5: 3.3 - 3.6 Mkm²
(2000–2006)
-GFED v3: 3.39 - 4.31 Mkm²
(1997-2009).
![Page 13: Remote sensing of burned areas](https://reader034.vdocuments.site/reader034/viewer/2022052302/568c48601a28ab49168fe73e/html5/thumbnails/13.jpg)
Click to edit Master title style Inconsistency in derived products
SEVIRI GFEDv2
2004 From FREEVAL final report. Courtesy of Martin Schultz
Comparison between SEVIRI FRP and GFED estimates of combusted biomass
![Page 14: Remote sensing of burned areas](https://reader034.vdocuments.site/reader034/viewer/2022052302/568c48601a28ab49168fe73e/html5/thumbnails/14.jpg)
Click to edit Master title style Uncertainty within a product (GFED v3)
Burned area proportion Uncertainty Giglio, L., J. T. Randerson, G. R. van der Werf, P. S. Kasibhatla, G. J. Collatz, D. C. Morton y R. S. DeFries (2010): Assessing variability and long-term trends in burned area by merging multiple satellite fire products. Biogeosciences Discuss., 7: 1171-1186, doi:10.5194/bg-7-1171-2010
This is very relevant for climate-carbon modelers!
![Page 15: Remote sensing of burned areas](https://reader034.vdocuments.site/reader034/viewer/2022052302/568c48601a28ab49168fe73e/html5/thumbnails/15.jpg)
Click to edit Master title style Scientific goals of fire_CCI
1. Refine definition of user requirements (GCOS are unrealistic).
2. Improve current estimations of global burned area (based on European sensors: VGT-ATSR-MERIS).
3. Validate and intercompare existing BA global products.
4. Test improvements of climate-vegetation-carbon models with new BA data.
![Page 16: Remote sensing of burned areas](https://reader034.vdocuments.site/reader034/viewer/2022052302/568c48601a28ab49168fe73e/html5/thumbnails/16.jpg)
Click to edit Master title style Consortium composition
Climate Modeling
User Group (CMUG)
International Science
Working Group:
•UMD – MODIS team
•FAO REDD
•JRC EFFIS
•NGO, CI
Science
Coordinator
Project
Manager
EO Science
Team
Algorithm Development
& Intercomparison
Validation
Data pre-
processing
Climate Modelers
System Engineering
![Page 17: Remote sensing of burned areas](https://reader034.vdocuments.site/reader034/viewer/2022052302/568c48601a28ab49168fe73e/html5/thumbnails/17.jpg)
Click to edit Master title style fire_cci production targets
• Temporal series of BA over 10 selected study sites (500x500 km) (1995-2009):
– Assure spatial accuracy and stability.
– Consistency across multiple satellites
– Demonstrate full-time series available.
• Global coverage for five years (1999, 2000, 2003, 2005 and 2008):
– Demonstrate the semi-operational processing.
– Ensemble chain, bulk processing of data.
![Page 18: Remote sensing of burned areas](https://reader034.vdocuments.site/reader034/viewer/2022052302/568c48601a28ab49168fe73e/html5/thumbnails/18.jpg)
Click to edit Master title style Study sites
![Page 19: Remote sensing of burned areas](https://reader034.vdocuments.site/reader034/viewer/2022052302/568c48601a28ab49168fe73e/html5/thumbnails/19.jpg)
Click to edit Master title style Target products
• Burned pixels (mixing all three sensors whenever possible): – Monthly files with date of detection.
– Minimum Mapping Unit (MMU) is under discussion.
– GeoTiff format
• Grid product: – 0.5 x 0.5 degree (CGM) / improvements
to 0.25 or 0.1 degrees are foreseen.
– NetCDF format.
![Page 20: Remote sensing of burned areas](https://reader034.vdocuments.site/reader034/viewer/2022052302/568c48601a28ab49168fe73e/html5/thumbnails/20.jpg)
Click to edit Master title style Tiles for the pixel product
In addition to standard tiles, the user will have a web tool to interactively select his/her target site and apply for personal downloads
![Page 21: Remote sensing of burned areas](https://reader034.vdocuments.site/reader034/viewer/2022052302/568c48601a28ab49168fe73e/html5/thumbnails/21.jpg)
Click to edit Master title style Project status
User requirements
External BA
algorithms
Product
specifications Geometric correction
Calibrated reflectances
Validation
Topographic shadow
correction
Atmospheric correction
Global BA
production
Raw Data:
ATSR, VGT, MERIS
Water-snow-
cloud masking
Round robin
Development of BA
algorithms
DEM
Testing of models
BA reference data
Merging algorithm
Done
In process
BA
reference
data
![Page 22: Remote sensing of burned areas](https://reader034.vdocuments.site/reader034/viewer/2022052302/568c48601a28ab49168fe73e/html5/thumbnails/22.jpg)
Click to edit Master title style Major deliverables
• User Requirement Document (URD).
• Product Specification Document (PSD).
• Product Validation Plan (PVP).
• Comprehensive Error Characterisation Report (CERC).
• ATBDs (Pre-processing, BA algorithms, Merging).
• System Requirement Document (SRD)
• System Specification Document (SSD).
![Page 23: Remote sensing of burned areas](https://reader034.vdocuments.site/reader034/viewer/2022052302/568c48601a28ab49168fe73e/html5/thumbnails/23.jpg)
Click to edit Master title style Pre-processing
• Geometric correction.
• Masking (cloud, haze, snow, water).
• Atmospheric correction (ATCOR).
• 10 sites x 3 sensors x (12-9-5) years: more than 70,000 corrected reflectance images + masks have been processed.
• Global processor is being implemented.
![Page 24: Remote sensing of burned areas](https://reader034.vdocuments.site/reader034/viewer/2022052302/568c48601a28ab49168fe73e/html5/thumbnails/24.jpg)
Click to edit Master title style BRDF: TOA/BOA time series
• Time series RGB (Meris bands 7, 5, 3):
Meris-FRS, Australia, Pixel 844,555 after atmospheric correction (BOA)
Meris-FRS, Australia, Pixel 844,555 before atmospheric correction (TOA)
Meris-FRS: Blue 1-4, Green 5, Red 6-6, Red-Edge 8, NIR 9-15, O2-Absorption 11, Water vapour 15
![Page 25: Remote sensing of burned areas](https://reader034.vdocuments.site/reader034/viewer/2022052302/568c48601a28ab49168fe73e/html5/thumbnails/25.jpg)
Click to edit Master title style BA VGT Algorithm (ISA, Portugal) Pereira and Mota, 2012
![Page 26: Remote sensing of burned areas](https://reader034.vdocuments.site/reader034/viewer/2022052302/568c48601a28ab49168fe73e/html5/thumbnails/26.jpg)
Click to edit Master title style BA Algorithm VGT Results
VGT vs MODIS VGT detection dates
AUSTRALIA
![Page 27: Remote sensing of burned areas](https://reader034.vdocuments.site/reader034/viewer/2022052302/568c48601a28ab49168fe73e/html5/thumbnails/27.jpg)
Click to edit Master title style MERIS algorithm (UAH, Spain)
![Page 28: Remote sensing of burned areas](https://reader034.vdocuments.site/reader034/viewer/2022052302/568c48601a28ab49168fe73e/html5/thumbnails/28.jpg)
Click to edit Master title style
6 layers for a monthly product: example from June 2005 Australian study site: BA, CL, days between burn date and last valid bservation before that date, valid obs, all obs, cloud obs
Auxiliary layers
![Page 29: Remote sensing of burned areas](https://reader034.vdocuments.site/reader034/viewer/2022052302/568c48601a28ab49168fe73e/html5/thumbnails/29.jpg)
Click to edit Master title style MERIS results: Australian site
2005 2006
![Page 30: Remote sensing of burned areas](https://reader034.vdocuments.site/reader034/viewer/2022052302/568c48601a28ab49168fe73e/html5/thumbnails/30.jpg)
Click to edit Master title style MERIS results: Canadian site
2005 2006
![Page 31: Remote sensing of burned areas](https://reader034.vdocuments.site/reader034/viewer/2022052302/568c48601a28ab49168fe73e/html5/thumbnails/31.jpg)
Click to edit Master title style MERIS results: Kazakhstan site
2005 2006
![Page 32: Remote sensing of burned areas](https://reader034.vdocuments.site/reader034/viewer/2022052302/568c48601a28ab49168fe73e/html5/thumbnails/32.jpg)
Click to edit Master title style Sensor Combinations (PL1) ATSR Algo1 (1km)
VGT Algo1 (1km)
MERIS Algo2 (300m)
Sensor combo (1km)
Uncertainty reduction: Multiple observation of same burn
500km2 Area stats
ATSR – 3691km2
– core burns
– overlap 89.9%
VEGETATION – 4211km2
– core burns
– overlap 86.7%
MERIS – 6977km2
– cores, detail, other events
– overlap 56.9%
![Page 33: Remote sensing of burned areas](https://reader034.vdocuments.site/reader034/viewer/2022052302/568c48601a28ab49168fe73e/html5/thumbnails/33.jpg)
Click to edit Master title style Date of detection PL2 ATSR Algo1 (JD)
VGT Algo1 (JD)
MERIS Algo2 (JD)
Earliest detection (JD) • Dates brought back
• Patch progression
Julian Day (August)
Uncertainty reduction: Reduced time lapse of observation
![Page 34: Remote sensing of burned areas](https://reader034.vdocuments.site/reader034/viewer/2022052302/568c48601a28ab49168fe73e/html5/thumbnails/34.jpg)
Click to edit Master title style Grid output (1)
Burned pixels (1km) Sum burned m2(0.5 degree) Unsuitable obs %. (0.5 deg)
Jan Feb Mar
Apr May Jun
Jul Aug Sep
Oct Nov Dec
![Page 35: Remote sensing of burned areas](https://reader034.vdocuments.site/reader034/viewer/2022052302/568c48601a28ab49168fe73e/html5/thumbnails/35.jpg)
Click to edit Master title style
Burned pixels (1km)
Jan Feb Mar
Apr May Jun
Jul Aug Sep
Oct Nov Dec
Confidence (0.5 degree) Conf sd (0.5 degree)
Apr May Jun
Jul Aug Sep
Grid output (2)
![Page 36: Remote sensing of burned areas](https://reader034.vdocuments.site/reader034/viewer/2022052302/568c48601a28ab49168fe73e/html5/thumbnails/36.jpg)
Click to edit Master title style
Burned pixels (1km)
Jan Feb Mar
Apr May Jun
Jul Aug Sep
Oct Nov Dec
Homog. ind (0.5 degree) Land cover (0.5 degree)
Grid output (3)
![Page 37: Remote sensing of burned areas](https://reader034.vdocuments.site/reader034/viewer/2022052302/568c48601a28ab49168fe73e/html5/thumbnails/37.jpg)
Click to edit Master title style Validation (UAH – GAF)
• Standard CEOS Validation protocol.
• 250 Landsat-TM/ETM+ multitemporal pairs are being processed:
– Temporal validation: study sites.
– Spatial validation: stratified random sampling.
• Validation metrics:
– Accuracy (agreement global-reference data).
– Error balance (over-under estimation).
– Temporal consistency.
![Page 38: Remote sensing of burned areas](https://reader034.vdocuments.site/reader034/viewer/2022052302/568c48601a28ab49168fe73e/html5/thumbnails/38.jpg)
Click to edit Master title style Temporal validation
![Page 39: Remote sensing of burned areas](https://reader034.vdocuments.site/reader034/viewer/2022052302/568c48601a28ab49168fe73e/html5/thumbnails/39.jpg)
Click to edit Master title style Temporal validation
Canada Colombia Brazil Portugal Angola South Africa Kazachstan Russia Australia Borneo
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
Light green, SLC-OFF
![Page 40: Remote sensing of burned areas](https://reader034.vdocuments.site/reader034/viewer/2022052302/568c48601a28ab49168fe73e/html5/thumbnails/40.jpg)
Click to edit Master title style Spatial validation
![Page 41: Remote sensing of burned areas](https://reader034.vdocuments.site/reader034/viewer/2022052302/568c48601a28ab49168fe73e/html5/thumbnails/41.jpg)
Click to edit Master title style Generation of reference perimeters
• ABAMS: (Bastarika et al. 2011). Based on a two-phase algorithm:
– Seed detection.
– Region-growing algorithm.
– Includes multitemporal images.
• Results are visually reviewed and cross-check with another interpreter.
• Standard documentation protocol (CEOS).
![Page 42: Remote sensing of burned areas](https://reader034.vdocuments.site/reader034/viewer/2022052302/568c48601a28ab49168fe73e/html5/thumbnails/42.jpg)
Click to edit Master title style Examples of fire reference data
•
Canada
![Page 43: Remote sensing of burned areas](https://reader034.vdocuments.site/reader034/viewer/2022052302/568c48601a28ab49168fe73e/html5/thumbnails/43.jpg)
Click to edit Master title style Examples
• Brasil
![Page 44: Remote sensing of burned areas](https://reader034.vdocuments.site/reader034/viewer/2022052302/568c48601a28ab49168fe73e/html5/thumbnails/44.jpg)
Click to edit Master title style Examples
pre post Angola
![Page 45: Remote sensing of burned areas](https://reader034.vdocuments.site/reader034/viewer/2022052302/568c48601a28ab49168fe73e/html5/thumbnails/45.jpg)
Click to edit Master title style Fuzzy error matrix
Reference data
Global
product Burned Unburned
Global
total
Burned p11 p12 p1+
Unburned p21 p22 p2+
Reference Total
p+1 p+2 p=1
Error matrix
commission
true burned
omission
true unburned
![Page 46: Remote sensing of burned areas](https://reader034.vdocuments.site/reader034/viewer/2022052302/568c48601a28ab49168fe73e/html5/thumbnails/46.jpg)
Click to edit Master title style Round Robin results
• BA algorithms/products tend to underestimate (red areas), with exceptions (green areas)
![Page 47: Remote sensing of burned areas](https://reader034.vdocuments.site/reader034/viewer/2022052302/568c48601a28ab49168fe73e/html5/thumbnails/47.jpg)
Click to edit Master title style Modeling exercises with BA data
• Monthly C emissions from biomass burning for the period of the ESA fire product
• Carbon budgets and Vegetation dynamics (Orchidee).
• Update Mouillot & Field 2005 historical database.
• Estimating errors in existing historical C emissions reconstructions
• Comparing regional and global estimations.
![Page 48: Remote sensing of burned areas](https://reader034.vdocuments.site/reader034/viewer/2022052302/568c48601a28ab49168fe73e/html5/thumbnails/48.jpg)
Click to edit Master title style International scope
• Critical phases are monitored by Key Science bodies.
• Close connection with GOFC-GOLD Fire IT and CEOS Cal-Val.
• Openness: Round Robin exercise.
• Regional validation workshops:
– Tropical: Brazil.
– Boreal: ¿Russia?
![Page 49: Remote sensing of burned areas](https://reader034.vdocuments.site/reader034/viewer/2022052302/568c48601a28ab49168fe73e/html5/thumbnails/49.jpg)
Click to edit Master title style Stresa R-R workshop (17-18 October, 2011)
![Page 50: Remote sensing of burned areas](https://reader034.vdocuments.site/reader034/viewer/2022052302/568c48601a28ab49168fe73e/html5/thumbnails/50.jpg)
Click to edit Master title style Main challenges of fire_CCI
• GCOS requirements very demanding.
• Input data for BA mapping:
– None of the input sensors (ATSR, VGT, MERIS) was designed for BA mapping.
– Little experience with ESA sensors.
• None for MERIS
• Limited for VGT and ATSR (Globcarbon and L3JRC)
– Existing MODIS BA products (2000-2011).
• Time constrains, particularly for BA algorithms.
![Page 51: Remote sensing of burned areas](https://reader034.vdocuments.site/reader034/viewer/2022052302/568c48601a28ab49168fe73e/html5/thumbnails/51.jpg)
Click to edit Master title style Fortitudes of fire_CCI
• Output product will combine three input sensors.
• Validation, temporal and spatial datasets.
• Uncertainty and error characterization.
• Strong connections with climate and international science community (GOFC-GOLD Fire IT).
![Page 52: Remote sensing of burned areas](https://reader034.vdocuments.site/reader034/viewer/2022052302/568c48601a28ab49168fe73e/html5/thumbnails/52.jpg)
Click to edit Master title style http://www.esa-fire-cci.org/
![Page 53: Remote sensing of burned areas](https://reader034.vdocuments.site/reader034/viewer/2022052302/568c48601a28ab49168fe73e/html5/thumbnails/53.jpg)
Click to edit Master title style Burn severity
• Degree of post-fire disturbance.
• Factors:
– Previous biomass loads.
– Fire behaviour (intensity and duration).
• Importance:
– BS is critical for post-fire regeneration and soil degradation.
– BS is an key factor in estimating gas emissions.
![Page 54: Remote sensing of burned areas](https://reader034.vdocuments.site/reader034/viewer/2022052302/568c48601a28ab49168fe73e/html5/thumbnails/54.jpg)
Click to edit Master title style
ECOLOGICAL CONSECUENCES ON BIOPHYSICAL PRE-FIRE
COMPONENTS IMMEDIATELY AFTER THE FIRE
DESCRIBE THE RECOVERY OF THE ECOSYSTEM FROM FIRE IMPACTS
From Key (2006)
Temporal scales
![Page 55: Remote sensing of burned areas](https://reader034.vdocuments.site/reader034/viewer/2022052302/568c48601a28ab49168fe73e/html5/thumbnails/55.jpg)
Click to edit Master title style Fire behaviour variations
![Page 56: Remote sensing of burned areas](https://reader034.vdocuments.site/reader034/viewer/2022052302/568c48601a28ab49168fe73e/html5/thumbnails/56.jpg)
Click to edit Master title style Field methods
• Quantitative measurements:
– Depth of charcoal layer.
– Ash/charcoal proportion.
– Amount of dead species.
– Depth of soil organic layer affected.
– Thickness of the minimum branch left.
• Qualitative observations:
– Visual estimations: ordinal ranks: low, medium, high.
– Quantitative ranges, visually estimated: CBI (Key and Benson, 2005).
![Page 57: Remote sensing of burned areas](https://reader034.vdocuments.site/reader034/viewer/2022052302/568c48601a28ab49168fe73e/html5/thumbnails/57.jpg)
Click to edit Master title style Remote Sensing methods
• Empirical models: – Collection of field samples.
– Extraction of satellite information (after calibration).
– Generation of statistical fittings.
• Simulation models: – Find a good model.
– Provide sound input parameters (realistic scenarios).
– Find a good inversion method.
![Page 58: Remote sensing of burned areas](https://reader034.vdocuments.site/reader034/viewer/2022052302/568c48601a28ab49168fe73e/html5/thumbnails/58.jpg)
Click to edit Master title style • Empirical models
• Simulation models
N
Cab
Cw
Cm
DIRECT
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
B3 B4 B1 B2 B5 B6 B7
INVERSE
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
B3 B4 B1 B2 B5 B6 B7
SdNDVICBI 574.983.2679.1
![Page 59: Remote sensing of burned areas](https://reader034.vdocuments.site/reader034/viewer/2022052302/568c48601a28ab49168fe73e/html5/thumbnails/59.jpg)
Click to edit Master title style
Study case:
Fire started on July 16,
2005 and it was caused by
careslessness
![Page 60: Remote sensing of burned areas](https://reader034.vdocuments.site/reader034/viewer/2022052302/568c48601a28ab49168fe73e/html5/thumbnails/60.jpg)
Click to edit Master title style
![Page 61: Remote sensing of burned areas](https://reader034.vdocuments.site/reader034/viewer/2022052302/568c48601a28ab49168fe73e/html5/thumbnails/61.jpg)
Click to edit Master title style U
nd
ers
tory
C
an
op
y
Plo
t CBI method: strata
![Page 62: Remote sensing of burned areas](https://reader034.vdocuments.site/reader034/viewer/2022052302/568c48601a28ab49168fe73e/html5/thumbnails/62.jpg)
Click to edit Master title style
B+C
1-5 m
A
Substrate
D+E
5-30 m
![Page 63: Remote sensing of burned areas](https://reader034.vdocuments.site/reader034/viewer/2022052302/568c48601a28ab49168fe73e/html5/thumbnails/63.jpg)
Click to edit Master title style Models selected
• Leaf level: PROSPECT.
• Canopy level: Kuusk.
– Includes two vegetation layer + background.
– Vegetation is assumed to be distributed homogeneously.
• Reference: Kuusk, Journal of Quantitative Spectroscopy & Radiative Transfer 71 (2001): 1 –9
![Page 64: Remote sensing of burned areas](https://reader034.vdocuments.site/reader034/viewer/2022052302/568c48601a28ab49168fe73e/html5/thumbnails/64.jpg)
Click to edit Master title style
LAI
s()
l
sl
a
s
v
v
Direct mode
R(,s,v,v) PROSPECT Kuusk 2 Layer RTM
N
Cab
Cw
Cm
LAI
s()
l
sl
a
s
v
v
()
()
Model overview
Target CBI
Inverse mode
Fixed parameters
Variables
Model outputs
PFA PCC
SB
![Page 65: Remote sensing of burned areas](https://reader034.vdocuments.site/reader034/viewer/2022052302/568c48601a28ab49168fe73e/html5/thumbnails/65.jpg)
Click to edit Master title style
Variables Strata
Substrate
Lineal Mixture of soil
and char + ash
spectra (SB)
Canopy
Understory Percentage in Cover
Change (PCC). Linear
change of LAI
Percentage of Foliage
Altered (PFA). Linear
change from green to
brown leaves
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45
400 600 800 1000 1200 1400 1600 1800 2000 2200 2400
longitud de onda (nm)
Refl
ecti
vid
ad
0
20
40
65
80
Variables in the simulation
![Page 66: Remote sensing of burned areas](https://reader034.vdocuments.site/reader034/viewer/2022052302/568c48601a28ab49168fe73e/html5/thumbnails/66.jpg)
Click to edit Master title style Model implementation: LUT
Substrate
Canopy
Understory
Look up
table
Simulated Spectra
Final
Spectra
CBI Filters
CBI B+C > CBI D+E: No canopy fires are more severe than understory fires.
If CBI A = 3 then CBI B+C>2 regardless CBI D+E (Severe fires in the soil imply medium to heavy understory fires)
If CBI A < 3 then CBI B+C>CBI A (most commonly severity is higher in the understory than substrate)
![Page 67: Remote sensing of burned areas](https://reader034.vdocuments.site/reader034/viewer/2022052302/568c48601a28ab49168fe73e/html5/thumbnails/67.jpg)
Click to edit Master title style
LUT 201 spectral bands Convolution to fit target
sensor (Landsat-TM)
LUT to Spectral Library
Spectral signatures for different CBI values
![Page 68: Remote sensing of burned areas](https://reader034.vdocuments.site/reader034/viewer/2022052302/568c48601a28ab49168fe73e/html5/thumbnails/68.jpg)
Click to edit Master title style
BAND 1
BA
ND
2
Spectral
angle Reference spectrum
Inversion criterion: SAM
• Selects the LUT spectrum with the minimum angle to the target and assigns it the correspondent CBI value.
• It is less sensitive to albedo variations than minimum distance.
![Page 69: Remote sensing of burned areas](https://reader034.vdocuments.site/reader034/viewer/2022052302/568c48601a28ab49168fe73e/html5/thumbnails/69.jpg)
Click to edit Master title style Comparison with empirical model results (De Santis and Chuvieco, 2007)
R2= 0.66 Variables included: dNDVI+Sat Tendency to smooth CBI values
R2= 0.63 Supervised simulation
Dr. Viegas
![Page 70: Remote sensing of burned areas](https://reader034.vdocuments.site/reader034/viewer/2022052302/568c48601a28ab49168fe73e/html5/thumbnails/70.jpg)
Click to edit Master title style
0.0
0.5
1.0
1.5
2.0
2.5
3.0
P-N7 P59 P46 P12 P86 P5 P11 P73 P45 P16 P32 P71 P49 P17 P96 P68
CBI plot Supervised Empirical
Empirical - Simulation model
Empirical fitting tends to smooth CBI range
![Page 71: Remote sensing of burned areas](https://reader034.vdocuments.site/reader034/viewer/2022052302/568c48601a28ab49168fe73e/html5/thumbnails/71.jpg)
Click to edit Master title style
CBI
DOES NOT
TAKE INTO
ACCOUNT
% OF DEAD LEAVES LITTER
FCOV OF EACH VEGETATION STRATUM
The results of RTM inversion suggest that both variables and their mixing effects are key factors of burn severity estimation
from remotely sensed data.
Problems with CBI
![Page 72: Remote sensing of burned areas](https://reader034.vdocuments.site/reader034/viewer/2022052302/568c48601a28ab49168fe73e/html5/thumbnails/72.jpg)
Click to edit Master title style
2 NEW VARIABLES PER STRATUM :
n
n
m
m
m
m
m
mm
FCOV
FCOVCBI
GeoCBI
1
1
)*(
FIELD EXPERIENCE SIMULATION ANALYSIS RESULTS
The new version of CBI proposed, called GeoCBI (which is short for Geometrically structured
Composite Burn Index), was computed as follows:
% OF CHANGES IN THE LAI
FCOV OF VEGETATION STRATA
from 0 to 3 (for strata C, D and E ), as the original variables
0 to 1 (for strata B, C, D and E )
WEIGHTING FACTOR
where m is the identification of each stratum and n is the number of strata.
GeoCBI
![Page 73: Remote sensing of burned areas](https://reader034.vdocuments.site/reader034/viewer/2022052302/568c48601a28ab49168fe73e/html5/thumbnails/73.jpg)
Click to edit Master title style FIELD PLOTS CBI GeoCBI SPECTRAL SIGNATURES(TM)
P95
2.58
2.7
P86 2.85
P23 2.8
0
0.05
0.1
0.15
0.2
0.25
400 900 1400 1900 2400R
EFLE
CTA
NC
E
WAVELENGTH (nm)
P95
P86
P23
![Page 74: Remote sensing of burned areas](https://reader034.vdocuments.site/reader034/viewer/2022052302/568c48601a28ab49168fe73e/html5/thumbnails/74.jpg)
Click to edit Master title style
Previous model at canopy level: Kuusk
New model at canopy level: GeoSail
SUBSTRATUM
UNDERSTORY
OVERSTORY Model at leaf
level: PROSPECT
EASY TO COMPUTE (few inputs)
SUCCESSFULLY APPLIED TO CONIFERS (Cheng et al., 2006; Kötz et al, 2003 y 2004; Zarco-Tejada et al., 2004) EASY TO COMPUTE(few inputs)
IT CAN SIMULATE SEVERAL VEG. LAYERS FITS WELL INTO THE STRUCTURE OF GeoCBI
New RTM simulation tools
![Page 75: Remote sensing of burned areas](https://reader034.vdocuments.site/reader034/viewer/2022052302/568c48601a28ab49168fe73e/html5/thumbnails/75.jpg)
Click to edit Master title style Advantages of simulation models
• Interpretation is based on physical roots.
• They are applicable everywhere (properly parametrized).
• They can simulate a wide range of conditions (difficult to find in a single fire).
![Page 76: Remote sensing of burned areas](https://reader034.vdocuments.site/reader034/viewer/2022052302/568c48601a28ab49168fe73e/html5/thumbnails/76.jpg)
Click to edit Master title style Other sites
De Santis and Chuvieco, 2009, RSE
![Page 77: Remote sensing of burned areas](https://reader034.vdocuments.site/reader034/viewer/2022052302/568c48601a28ab49168fe73e/html5/thumbnails/77.jpg)
Click to edit Master title style From Burn Severity to Burning efficiency Seiler and Crutzen [1980] model Mk,i = (BLi * BEi * BSi * AEk)*10-15 Mk,I = Emissions of gas k (Tg) BLi= Biomass loads (gr/m2) BEi = Burning efficiency (combustion completness) (0/1) BSi= Burned surface(m2) AEk= Emission factors (gr gas / Kg biomass)
BE is estimated from: • Standard coeffcients. • Fuel moisture content. • Remote sensing.
![Page 78: Remote sensing of burned areas](https://reader034.vdocuments.site/reader034/viewer/2022052302/568c48601a28ab49168fe73e/html5/thumbnails/78.jpg)
Click to edit Master title style BE standard values
ID Land Cover Category BE
1 Evergreen Needleleaf Woods 0.250
2 Evergreen Broadleaf Forest 0.250
3 Deciduous Needleleaf Wood 0.250
4 Deciduous Broadleaf Woods 0.250
5 Mixed Forest 0.250
6 Woody Savanna (30-60% >2m) 0.350
7 Savanna Trees (10-30%) 0.400
8 Closed Shrubland (scrub) 0.500
9 Open Shrubland (semidesert) 0.950
10 Grassland 0.950
11 Cropland Herbaceous & villages 0.800
12 Barren deserts volcanos 0.200
13 Urban/ suburban built-up 0.100
14 Water +/- coastal 0.100
15 Permanent wetlands 0.500
16 Cropland/ grass-woods(Field-woods) 0.400
17 Snow& Ice 0.000
18 Tundra/ Paramo 0.300
19 Woodlands trees (40-60%> 5)m 0.400
20 Forest-Field Mix (40-60% woods) 0.300
21 Mediterranean scrub 0.700
![Page 79: Remote sensing of burned areas](https://reader034.vdocuments.site/reader034/viewer/2022052302/568c48601a28ab49168fe73e/html5/thumbnails/79.jpg)
Click to edit Master title style Emission estimate formula’s
• Emission= Aburned x C x Eeff x Fload
– Aburned is the area burned (retrieved)
– C is the combustion completeness (guessed)
– Eeff Emission Efficiency (guessed)
– Fload is the fuel load (computed with Biomass model)
• Emission = Efactor x
dtEnergyRadiativeFire ).__(
![Page 80: Remote sensing of burned areas](https://reader034.vdocuments.site/reader034/viewer/2022052302/568c48601a28ab49168fe73e/html5/thumbnails/80.jpg)
Click to edit Master title style
Martin J. Wooster’s results (in press)
![Page 81: Remote sensing of burned areas](https://reader034.vdocuments.site/reader034/viewer/2022052302/568c48601a28ab49168fe73e/html5/thumbnails/81.jpg)
Click to edit Master title style BE from RS methods
• Post-fire reflectance analysis:
– BE from Burn Severity.
– BE from simulation models.
• Energy released by the fire (FRP):
– Instantaneous to total FRP.
– Relation of FRP to biomass consumption.
![Page 82: Remote sensing of burned areas](https://reader034.vdocuments.site/reader034/viewer/2022052302/568c48601a28ab49168fe73e/html5/thumbnails/82.jpg)
Click to edit Master title style BE from BS
Local estimation of burned severity (RTM
– CBI)
Regional estimation
Burning efficiency
Validation with Landsat TM
BA map
Vegetation cover Max/Min values
of BE Burning Severity
BE
Low Med Severe
Grass 0.85 0.9 0.98
Shrub 0.7 0.85 0.95
Conifer 0.25 0.42 0.57
Deciduous 0.25 0.4 0.56 Oliva and Chuvieco, 2012
![Page 83: Remote sensing of burned areas](https://reader034.vdocuments.site/reader034/viewer/2022052302/568c48601a28ab49168fe73e/html5/thumbnails/83.jpg)
Click to edit Master title style Results
Oliva and Chuvieco, 2012
![Page 84: Remote sensing of burned areas](https://reader034.vdocuments.site/reader034/viewer/2022052302/568c48601a28ab49168fe73e/html5/thumbnails/84.jpg)
Click to edit Master title style Thank you!!