36
Fire From the Sky: Remote Sensing in Wildfire Management Leda Kobziar, School of Forest Resources and Conservation University of Florida Graduate Student Researchers: David Godwin, Sparkle Malone, Johanna Freeman

Fire From the Sky: Remote Sensing in Wildfire …abe.ufl.edu/research/CRS/seminar/20120405_Kobziar_Seminar.pdfFire From the Sky: Remote Sensing in Wildfire Management ... Including

  • Upload
    buingoc

  • View
    216

  • Download
    3

Embed Size (px)

Citation preview

Page 1: Fire From the Sky: Remote Sensing in Wildfire …abe.ufl.edu/research/CRS/seminar/20120405_Kobziar_Seminar.pdfFire From the Sky: Remote Sensing in Wildfire Management ... Including

Fire From the Sky: Remote Sensing in Wildfire Management

Leda Kobziar, School of Forest Resources and Conservation University of Florida Graduate Student Researchers: David Godwin, Sparkle Malone, Johanna Freeman

Page 2: Fire From the Sky: Remote Sensing in Wildfire …abe.ufl.edu/research/CRS/seminar/20120405_Kobziar_Seminar.pdfFire From the Sky: Remote Sensing in Wildfire Management ... Including
Page 3: Fire From the Sky: Remote Sensing in Wildfire …abe.ufl.edu/research/CRS/seminar/20120405_Kobziar_Seminar.pdfFire From the Sky: Remote Sensing in Wildfire Management ... Including

www.pnas.org/cgi/doi/10.1073/pnas.1003669107

Page 4: Fire From the Sky: Remote Sensing in Wildfire …abe.ufl.edu/research/CRS/seminar/20120405_Kobziar_Seminar.pdfFire From the Sky: Remote Sensing in Wildfire Management ... Including
Page 5: Fire From the Sky: Remote Sensing in Wildfire …abe.ufl.edu/research/CRS/seminar/20120405_Kobziar_Seminar.pdfFire From the Sky: Remote Sensing in Wildfire Management ... Including

NOAA’s Hazard Mapping System combines RS data from satellite sources to detect fire/ smoke plumes: Including Geostationary Operational Environmental Satellite (GOES), Moderate-Resolution Imaging Spectroradiometer (MODIS), and Advanced Very High Resolution Radiometer (AVHRR)

Issues: scale of detection, atmospheric interference, timing of satellite pass bys

Page 6: Fire From the Sky: Remote Sensing in Wildfire …abe.ufl.edu/research/CRS/seminar/20120405_Kobziar_Seminar.pdfFire From the Sky: Remote Sensing in Wildfire Management ... Including

fire behavior

Weather: varies across space and time, inherently unpredictable at the scale of fire behavior

Fuels: varies with space & time, predictable

Topography: varies with space, predictable

Can be manipulated!

Page 7: Fire From the Sky: Remote Sensing in Wildfire …abe.ufl.edu/research/CRS/seminar/20120405_Kobziar_Seminar.pdfFire From the Sky: Remote Sensing in Wildfire Management ... Including

To reduce horizontal or vertical continuity of fuelbeds

To reduce oxygen supply to fuels To reduce fuel height To decrease fuel loads

Page 8: Fire From the Sky: Remote Sensing in Wildfire …abe.ufl.edu/research/CRS/seminar/20120405_Kobziar_Seminar.pdfFire From the Sky: Remote Sensing in Wildfire Management ... Including

BEFORE AFTER

Page 9: Fire From the Sky: Remote Sensing in Wildfire …abe.ufl.edu/research/CRS/seminar/20120405_Kobziar_Seminar.pdfFire From the Sky: Remote Sensing in Wildfire Management ... Including

Thinning from below (small diamter, fire intolerant trees) and burning piles during the winter. Common in western US

Page 10: Fire From the Sky: Remote Sensing in Wildfire …abe.ufl.edu/research/CRS/seminar/20120405_Kobziar_Seminar.pdfFire From the Sky: Remote Sensing in Wildfire Management ... Including

UF Statistics staff burning pine flatwoods at the Austin Cary Memorial Forest

Page 11: Fire From the Sky: Remote Sensing in Wildfire …abe.ufl.edu/research/CRS/seminar/20120405_Kobziar_Seminar.pdfFire From the Sky: Remote Sensing in Wildfire Management ... Including

Western US: over 12 M acres in highest category of need Avg. cost ~$450/ acre

Southern Region: Largely focused on prescribed burning

▪ Approx. $25/acre Region-wide, over 8 million acres

burned/year ▪ ~$200 M

Florida: 2 million acres burned annually Suppression effect- 12 cents per dollar

invested Prevention via fuels treatments- ~$3.76/

dollar invested (Lankoande et al. 2006)

A Wiser Smokey

Page 12: Fire From the Sky: Remote Sensing in Wildfire …abe.ufl.edu/research/CRS/seminar/20120405_Kobziar_Seminar.pdfFire From the Sky: Remote Sensing in Wildfire Management ... Including

Weather: varies across space and time, inherently unpredictable at the scale of fire behavior

Fuels: varies with space & time, predictable and can be managed

Topography: varies with space, predictable

Do fuels reduction work? Check the severity of subsequent fires

Page 13: Fire From the Sky: Remote Sensing in Wildfire …abe.ufl.edu/research/CRS/seminar/20120405_Kobziar_Seminar.pdfFire From the Sky: Remote Sensing in Wildfire Management ... Including

High spatial resolution Low spectral resolution High temporal resolution Low spatial inference High cost (personnel, time)

Page 14: Fire From the Sky: Remote Sensing in Wildfire …abe.ufl.edu/research/CRS/seminar/20120405_Kobziar_Seminar.pdfFire From the Sky: Remote Sensing in Wildfire Management ... Including

Med. Spatial & temporal resolution Low spectral resolution (unless MS or lidar is used) Med. To high cost Quality issues (overlap, cloud cover, georectifying can be a challenge), HUD

Page 15: Fire From the Sky: Remote Sensing in Wildfire …abe.ufl.edu/research/CRS/seminar/20120405_Kobziar_Seminar.pdfFire From the Sky: Remote Sensing in Wildfire Management ... Including
Page 16: Fire From the Sky: Remote Sensing in Wildfire …abe.ufl.edu/research/CRS/seminar/20120405_Kobziar_Seminar.pdfFire From the Sky: Remote Sensing in Wildfire Management ... Including

Low cost High spectral, lower spatial

and temporal resolutions Expands the scale of inference Spatial Temporal

Active fires: Fire perimeter, risk to communities & roads, smoke direction and intensity, fire severity

Past fires: Landscape scale effectiveness of fuels treatments over time Dependent variables: area

burned, fire severity

ETM+ bands 7, 5, and 3.

Page 17: Fire From the Sky: Remote Sensing in Wildfire …abe.ufl.edu/research/CRS/seminar/20120405_Kobziar_Seminar.pdfFire From the Sky: Remote Sensing in Wildfire Management ... Including

• 30-80 year fire return interval • Late spring / early summer fires

associated w/ drought, high winds, low RH, high temperature

• High-intensity, large scale, stand-replacing crown fires

• Auto-successional (fire climax) ecosystem

• Pinus clausa (sand pine) is serotinous

Fire Regime Without fire, sand pine scrub will likely succeed to xeric oak/hickory scrub.

Godwin and Kobziar, 2011

Page 18: Fire From the Sky: Remote Sensing in Wildfire …abe.ufl.edu/research/CRS/seminar/20120405_Kobziar_Seminar.pdfFire From the Sky: Remote Sensing in Wildfire Management ... Including

2006 Prescribed

Fire 2009 Wildfire

5,900 ha.

Wilderness

3800 ha.

Burned

Godwin and Kobziar, 2011

Page 19: Fire From the Sky: Remote Sensing in Wildfire …abe.ufl.edu/research/CRS/seminar/20120405_Kobziar_Seminar.pdfFire From the Sky: Remote Sensing in Wildfire Management ... Including

Issues: Need to conducted ground sampling to “train” RS evaluations of breakpoints between burn severity levels Timing of images must coincide with consistent veg. phenology phase Specific to ecosystem (“function of place”)

Landsat-5 TM •Four Scenes Used (2 pre-fire vs. 2 post-fire) •30 m resolution •Image classification: supervised worked best

Index used to determine fire severity • Normalized Burn Ratio (Key and Benson 2006)

• NBR = Band 4 – Band 7 / Band 4 + Band 7 • Band 4 = Near Infrared (veg. decreases following fire) • Band 7 = Mid Infrared (soil increases following fire) • dNBR = NBR pre fire – NBR post fire

Page 20: Fire From the Sky: Remote Sensing in Wildfire …abe.ufl.edu/research/CRS/seminar/20120405_Kobziar_Seminar.pdfFire From the Sky: Remote Sensing in Wildfire Management ... Including

(Lentile et. al, 2006)

Function of Place

BU

RN

SE

VE

RIT

Y

LOW

MODERATE

HIGH

CA MT AK

Page 21: Fire From the Sky: Remote Sensing in Wildfire …abe.ufl.edu/research/CRS/seminar/20120405_Kobziar_Seminar.pdfFire From the Sky: Remote Sensing in Wildfire Management ... Including
Page 22: Fire From the Sky: Remote Sensing in Wildfire …abe.ufl.edu/research/CRS/seminar/20120405_Kobziar_Seminar.pdfFire From the Sky: Remote Sensing in Wildfire Management ... Including

Is this simply a stand-replacing system, or is there spatial variability in fire effects? How does fire severity relate to pre-burn conditions (stand

age, density, topography, etc.)? How does an initial fire affect subsequent fires? What are the consequences for the conservation of

the sand pine ecosystem?

Page 23: Fire From the Sky: Remote Sensing in Wildfire …abe.ufl.edu/research/CRS/seminar/20120405_Kobziar_Seminar.pdfFire From the Sky: Remote Sensing in Wildfire Management ... Including

Landsat 5 supervised classification •~ 68% overall accuracy / ground truthing • High spatial variability (Godwin and Kobziar 2011)

Results: Accuracy of dNBR Severity Δ NBR Threshold Unburned -100- 57 Low 58 - 382 Moderate 383 - 596 High > 597

“Vaporized pre-cooked” sapling stand (mod. severity 2006, high 2009)

Page 24: Fire From the Sky: Remote Sensing in Wildfire …abe.ufl.edu/research/CRS/seminar/20120405_Kobziar_Seminar.pdfFire From the Sky: Remote Sensing in Wildfire Management ... Including

76% burned 45% burned

Page 25: Fire From the Sky: Remote Sensing in Wildfire …abe.ufl.edu/research/CRS/seminar/20120405_Kobziar_Seminar.pdfFire From the Sky: Remote Sensing in Wildfire Management ... Including

-26x -139x

-247x -314x

Page 26: Fire From the Sky: Remote Sensing in Wildfire …abe.ufl.edu/research/CRS/seminar/20120405_Kobziar_Seminar.pdfFire From the Sky: Remote Sensing in Wildfire Management ... Including

Stand Age, Fire Frequency

Fire

Sev

erit

y

Senescent, freq. low Sapling, freq. high

Low

High Sand pine scrub High BA

Sand pine scrub Low BA

Oak Palmetto Scrub

Sand pine scrub Low BA

Oak Hammock

Oak Palmetto Scrub

Oak Palmetto Scrub

Sand pine scrub High BA

Oak Hammock

Freeman and Kobziar, 2011

Page 27: Fire From the Sky: Remote Sensing in Wildfire …abe.ufl.edu/research/CRS/seminar/20120405_Kobziar_Seminar.pdfFire From the Sky: Remote Sensing in Wildfire Management ... Including

Severity mapping can target sampling strategies for tracking ecological responses across landscapes; these can then be projected to make spatially explicit predictions Only 18% of burned wilderness area is likely to return to sand pine

scrub We now know where these areas are, so they can be managed

appropriately Prescribed burning helps reduce wildfire area burned (-31%),

but effectiveness depends on fire severity (low severity 36% prob. of being unburned vs. high severity 67% unburned)

Page 28: Fire From the Sky: Remote Sensing in Wildfire …abe.ufl.edu/research/CRS/seminar/20120405_Kobziar_Seminar.pdfFire From the Sky: Remote Sensing in Wildfire Management ... Including

• Osceola National Forest (FL): Pine

flatwoods forest, FRI 3-5 yrs., 30,000 of 230,000 acres prescribed burned annually

• Each fire assessed for severity using dNBR,

plus time since last fire, fire frequency, veg. type, soil type, and drought index

• Logistic regression modeling analysis based on a decade (217 fires) of RS imagery to determine historical fuel treatment effects (1998-2009)

• How do compounded prescribed burn fuel treatments influence fire severity?

Malone, S., Kobziar, L. N., Staudhammer, C. L., Abd-Elrahman, A. 2011. Using 217 individual fire severity analyses to model subsequent fire severity in southern pine forests. Remote Sensing 3: 2005-2028.

Page 29: Fire From the Sky: Remote Sensing in Wildfire …abe.ufl.edu/research/CRS/seminar/20120405_Kobziar_Seminar.pdfFire From the Sky: Remote Sensing in Wildfire Management ... Including

Severity level ( αi) of the first fire Time interval between the first

and second fire (βj ) Type of fire (γk ) PDSI for the year before and the

year of each fire event (τ1-4)

Page 30: Fire From the Sky: Remote Sensing in Wildfire …abe.ufl.edu/research/CRS/seminar/20120405_Kobziar_Seminar.pdfFire From the Sky: Remote Sensing in Wildfire Management ... Including

• - Fire Frequency X1ij • + Time since last fire X2ij • Interaction

Page 31: Fire From the Sky: Remote Sensing in Wildfire …abe.ufl.edu/research/CRS/seminar/20120405_Kobziar_Seminar.pdfFire From the Sky: Remote Sensing in Wildfire Management ... Including
Page 32: Fire From the Sky: Remote Sensing in Wildfire …abe.ufl.edu/research/CRS/seminar/20120405_Kobziar_Seminar.pdfFire From the Sky: Remote Sensing in Wildfire Management ... Including
Page 33: Fire From the Sky: Remote Sensing in Wildfire …abe.ufl.edu/research/CRS/seminar/20120405_Kobziar_Seminar.pdfFire From the Sky: Remote Sensing in Wildfire Management ... Including

• Given an ignition, this enables managers to make informed choices about allocation of suppression efforts • Firefighter safety • Suppression effectiveness • Encourages consideration of

let-burn options where severity is unlikely to be high

• Prior to ignition, targeted fuels treatments can be enacted

• Their effectiveness can then be traced over time using RS

Page 34: Fire From the Sky: Remote Sensing in Wildfire …abe.ufl.edu/research/CRS/seminar/20120405_Kobziar_Seminar.pdfFire From the Sky: Remote Sensing in Wildfire Management ... Including
Page 35: Fire From the Sky: Remote Sensing in Wildfire …abe.ufl.edu/research/CRS/seminar/20120405_Kobziar_Seminar.pdfFire From the Sky: Remote Sensing in Wildfire Management ... Including

Fire Science Lab Members: David Godwin, Jesse Kreye, Adam Watts, Brenda Thomas, Mike Camp, Eric Carvalho, Dawn McKinstry, Leland Taylor, Johanna Freeman, Kathryn King, Alex Kattan, Jared Beauchamp, Steve Miller, Nichole Strickler, Terri Mashour

Page 36: Fire From the Sky: Remote Sensing in Wildfire …abe.ufl.edu/research/CRS/seminar/20120405_Kobziar_Seminar.pdfFire From the Sky: Remote Sensing in Wildfire Management ... Including

Lankoande, M., and J. Yoder, 2006. “An Econometric Model of Wildfire Suppression Productivity.” Working Paper, School of Economic Sciences, Washington State University (2006): 40 pp.

Key CH, Benson NC, 2006. Landscape Assessment: Ground measure of severity, the Composite Burn Index; and Remote sensing of severity, the Normalized Burn Ratio. In ‘FIREMON: Fire Effects Monitoring andInventory System’. (Eds DC Lutes, RE Keane, JF Caratti, CH Key, NC Benson, S Sutherland, LJ Gangi) USDAForest Service, Rocky Mountain

Research Station, GeneralTechnical ReportRMRS-GTR-164-CD: LA1-51. (Ogden, UT) Godwin, D.R., Kobziar, L.N., 2011. Comparison of burn severities of consecutive large-scale

fires in Florida sand pine scrub using satellite imagery analysis. Fire Ecology 7, 99-113

Malone, S., Kobziar, L. N., Staudhammer, C. L., Abd-Elrahman, A., 2011. Using 217 individual fire severity analyses to model subsequent fire severity in southern pine forests. Remote Sensing 3: 2005-2028.

Freeman, J., Kobziar, L. N. 2011. Tracking postfire successional trajectories in a plant community prone to high-intensity fire. Ecological Applications 21: 61-74.