Click to edit Master title style Pro s and Con s of using remote sensing in fire research Emilio Chuvieco Environmental Remote Sensing Research Group University of Alcalá, Spain emilio.chuvieco@uah.es
Initial Click thought to edit Master title style Sophisticated Innovative Accurate Accesible
RS is a basic tool to retrieve fire Click to edit Master title style information Fuel moisture Soils Elevation / DTM Fuel Types Meteorology / climate Density Duration / size Recurrence Intensity Socio-economic Data / Trends
RS information to answer Click to edit Master title style these questions: How much fuel is available to burn? Is it dry enough? Is there an active fire? Where? When did it start? How is it growing? How much energy? How much area is burned? How often? When in the year? How much biomass is consumed? Are fire characteristics changing?
Pros Click to edit Master Cons title style Multi variable Multi parameter. Multi sensor. Multi temporal. Multi scale. Products methods Analysis Data generation Validation & Uncertainty charact.
Pros Click to edit Master Cons title style Multi variable Multi parameter. Multi sensor. Multi temporal. Multi scale. Products methods Analysis Data generation Validation & Uncertainty charact.
Multivariable: Click to edit Fire Master risk title style Fire risk = Danger * Vulnerability Human Ignition Cause Lightning Danger Fuel moisture content Live Risk Propagation Fuel types Slope weather conditions Dead Fireglobe project Chuvieco et al., 2014, IJWF Vulnerability Ecological value Socio economic value Total value of environmental services Houses and infraestructure Recovery time Actual value of environmental services 7
Multivariable: Click to edit Fire Master risk title style Fire risk = Danger * Vulnerability Human Ignition Cause Lightning Danger Fuel moisture content Live Risk Propagation Fuel types Slope weather conditions Dead Fireglobe project Chuvieco et al., 2014, IJWF Vulnerability Ecological value Socio economic value Total value of environmental services Houses and infraestructure Recovery time Actual value of environmental services 8
National scale Click to edit results Master title style Value of ecosystem services Chuvieco et al., 2014, IJWF
Global results: Click to edit Master title style Loss of Ecological Values Chuvieco et al., 2014, GEB
Multi parameter: Click to edit Master fuelstitle style Passive Lidar Radar optical Horizontal continuity some some some Vertical distribution no yes some Biomass loads no yes some Surface conditions no some some Crown bulk density no yes no 11
Multi parameter: Click to edit Master CBD title style Lidar flight line Colour Infrared Aerial photo 0 100 m N CBD (kg/m 3 ) 1.2 Riaño et al., 2004, RSE 0 12
Multi sensor: Click to edit Fuel Master classification title style
Multi temporal: Click to edit Master finding title trends style
Burned area time seriestitle style Click to edit Master 30000 25000 9/11 9/1 8/22 20000 8/11 8/1 7/22 15000 7/11 7/1 6/21 6/11 10000 6/1 5/22 5/11 5000 2006 2005 2004 2003 2002 2001 2000 1999 1998 1997 1996 1995 1994 1993 1992 1991 1990 1989 1988 1987 1986 1985 1984 0 Chuvieco et al., 2008, RSE
Multi temporal Click to edit Master medium scale title style Rapid Eye: 5 bands: 6.5 m at nadir. 5 satellites: up to 1 day revisiting time. Sentinel 2 (MSI): 13 bands (10, 20, 60 m) 5 to 2 days temporal frequency.
Earth Click engine to edit Master title style https://earthengine.google.org
Multi scale: Click to edit fuel Master parameters title style Ground Airborne Garcia et al. 2011, IJAEO Garcia et al., 2010, RSE 18
Multi scale: Click to edit Master title style fuel parameters Global Fuel Map Pettinari et al. 2014, IJWF 19
Multi scale: Click to edit post fire Master assessment title style ground surveys
Multi scale: Click to edit medium scale Master title sensors style De Santis and Chuvieco, 2009, RSE
Multi scale: Click to edit global Master scale title style Averaged Burned Area (2006 2008) from Fire_CCI Burned Area
Pros Click to edit Master Cons title style Multi variable Multi parameter. Multi sensor. Multi temporal. Multi scale. Products methods Analysis Data generation Validation & Uncertainty charact.
Products Click to edit Methods Master title style N C ab C w C m DIRECT 0.35 0.35 0.3 0.3 0.25 0.2 0.15 0.1 0.05 INVERSE 0.25 0.2 0.15 0.1 0.05 0 B3 B4 B1 B2 B5 B6 B7 0 B3 B4 B1 B2 B5 B6 B7
Examples Click to of edit FMC Master mapstitle style 10th June FMC % of dry weight 28th August geogra.uah.es/fireglobe
Fire Click propagation to edit Master title style Chuvieco and Martin, 1994, IJRS Veraberveke et al., 2014, IJWF
Analysis Click to edit Data Master generation title style Pereira et al., 2015, PLOS
Burned Click to patch edit analysis Master title style Mouillot et al., 2015
FireClick sizeto distribution edit Master title style Hantson et al., 2015, GEB
Validation Click to edit & Uncertainty Master title style Saunders, 2010
Click to edit Master title style Uncertainty characterization: Monthly confidence level (%)
Click to edit Master title style The need for product intercomparision
Seasonal Click to trends edit in Master carbon title emissions style Yue et al., 2015
Validation Click to edit metrics Master title style Accuracy: Interval scale data: RMSE / R 2. Categorical scale: confusion matrix (OA, OE, CE, DC, Kappa ). Bias: Over / underestimation. Stability: Non parametric Friedman test of variance Wilcoxon t trends.
Validation Click to edit sample Master title style 130 Pairs of Landsat images for spatial validation 110 Pairs for temporal validation
Reference Click to edit filesmaster title style Generated from a semiautomatic algorithm (BAMS) over a pair of Landsat images July 18, 2008 Septembre 20, 2008 False color composition (RGB 743)
Click to edit Master title style BAMS https://bastarrika.wordpress.com/
Spatial Click variation to edit Master of accuracy title (2008) style Padilla et al., 2015, RSE
Temporal variation of accuracy Click to edit Master title style (2001 2007) Padilla et al., 2014, RS
Limitations Click to edit of HS Master products title style MODIS HS is a great product, but it has limitations caused by fire size, clouds, How much burned area is not accounted for in current BA products? # Fires BA (Km 2 ) Undetected Detected Omission Undetected Detected Omission <50Ha 41.521 17.357 0,71 5.470,99 3.040,15 0,64 50Ha 100Ha 1.622 1.842 0,47 1.123,46 1.394,04 0,45 100Ha 500Ha 1.036 2.448 0,30 1.655,02 4.833,27 0,26 >500Ha 81 810 0,09 432,68 13.628,95 0,03 Total 44.260 22.457 0,66 8.682,15 22.896,41 0,27 Total of 66,717 burned patches, affecting to 31,578 km 2 Basedon130 Landsat images for 2008 : Rodríguez & Chuvieco, 2015
RS Click = (Good) to edit Information Master title style Information is the resolution of uncertainty (Claude Shannon). Information is not knowledge (Albert Einstein) Good informa on Good knowledge Bad Information = Bad knowledge. Good knowledge = Good information Thank you! emilio.chuvieco@uah.es