Analysis of Forest Fire Disturbance in the Western US Using Landsat Time Series Images: 1985-2005

1. Analysis of Forest Fire Disturbance in the Western US Using Landsat Time Series Images: 1985-2005 Haley Wicklein Advisors: Jim Collatz and Jeff Masek, Goddard Space Flight Center, NASA Research and Discover Program 2008

2. Motivation • Forest carbon sink: ~ 270 ± 130 Mt C / yr • Debate about relative importance of causal factors (disturbance, land use, climate change, etc) • Future of sink is highly uncertain • Landsat is an important tool for environmental monitoring The First State of the Carbon Cycle Report (2007) King, Dilling, many others

3. Objectives • Compare two disturbance products, identify fraction of disturbance that is fire • Use products to understand meteorological drivers of fire • Examine regrowth dynamics for fire vs. other types of disturbance

4. Objective 1: Two Disturbance Maps Monitoring Trends in Burn Severity (MTBS) North American Forest Dynamics Project (NAFD) Joint project by USGS and USFS Maps burn severity and fire perimeter across the United States Uses before and after Landsat images to determine severity for all fires > 1000 acres Currently available for Pacific Northwest, CA, and parts of the Southwest Joint project by Goddard, UMD, and USFS Landsat based data cubes (30m res) Annual record of forest disturbance events (logging, fire, insects, etc) from 1984-2005 Don’t distinguish between different types of disturbance Year since disturbance Oregon Landsat cube Fire severity

5. Disagreement between MTBS and NAFD High 3% Moderate Disagreement 33% 6% Low 8% Unburned - Low Agreement 67% 16% Hash Rock Fire (2000) MTBS Ex: Oregon (1985-2005) NAFD disturbance map

6. Cube Statistics

7. Forest Disturbance Trends Total Area Disturbed Area Burned

8. Objective 2: Climate Correlations Westerling et al. (2006) found Occurrence of Western wildfires largely governed by climate Attributed fire to: Warmer spring and summer temperature Reduced precipitation Reduced snowpack/ earlier spring snowmelt All lead to… Longer, drier summer fire season

9. Fire Trends Correlate with Aridity Index 1=dry, 0=wet 1- (Precip/ PET) Fire

10. Objective 3: Regrowth dynamics

11. Spectral Indices: NBR and NDVI Normalized Burn Ratio (NBR) NIR – MIR = NIR + MIR [MTBS uses dNBR = pre NBR – post NBR] VIS NIR MIR WAVELENGTH NIR – VIS From: rangeview.arizona.edu = NIR + VIS Normalized Difference Vegetation Index (NDVI) NBR values range from -1 (bare soil) to 1 (very green) NDVI values range from 0 (bare soil) to 1 (very green)

12. NBR and NDVI by Severity Class Severity Class High Moderate Low Unburned to Low Jefferson Fire, 1996 Oregon (p45r29) 3922 acres NBR NDVI *NBR and NDVI values from NAFD maps

13. Spectral Recovery from Fire and Logging Disturbance Jefferson Fire Oregon (p45r29) 1996 3922 acres Fire Logging Logging disturbance 1985

14. Do NBR and NDVI Provide Useful Ecological Information? 1995 (year before fire) 1996 (year of fire) 2005(from: google earth)

15. Conclusions • Only 7-30% of total disturbance in Western US is fire • Hot and dry conditions necessary, but not sufficient for fire activity • NDVI and NBR are good measures of timing and magnitude of disturbance, but canopy structure and height data (ex: LIDAR) is needed for useful information on ecological variables.

16. Questions? Special Thanks to: Jim Collatz Jeff Masek Chris Williams The Research and Discover Program