Detecting and Parameterizing Wildfire Induced Land-surface Changes for Earth System Models

73 %
27 %
Information about Detecting and Parameterizing Wildfire Induced Land-surface Changes for...

Published on December 23, 2016



1. Detecting and Parameterizing Wildfire Induced Land-surface Changes for Earth System Models Yongqiang Liu USDA Forest Service, Athens, GA Xianjun Hao, John Qu George Mason University, Fairfax, VA Agroclimatology Project Directors Meeting December 16-18, 2016 JW Marriott, San Francisco (NIFA award #: 2013-35100-20516)

2. Climate impacts of wildfire smoke Southern California fires in fall, 2003 (Image from NASA)

3. Fire Vegetation coverage Stomata Albedo Radiation Roughness Wind Evapotrans- piration Sensible heat flux Climate - Large vegetation removal Díaz-Delgado et al. 2001, Díaz-Delgado et al. 2003, Li et al. 2008, van Leeuwen 2008, 2010. - Post-fire surface albedo recovers quickly after initial drop, and even exceeds pre-fire values when char materials are removed and vegetation starts to regenerate Amiro et al. 2006, Randerson et al. 2006, Lyons et al. 2008, Veraverbeke et al. 2012 - Immediate post-fire day-time surface temperature increase Lyons et al. 2008, Veraverbeke et al. 2012

4. Fire-climate interactions in Earth System Modeling (from Spessa et al., 2015)

5. Issues  Complex land-surface property changes due to variations in fire intensity, region, fuel type, and many other factors.  Quantitative description of instant land cover change and subsequent variations with DGVM modeling Objectives  Obtain quantitative estimation of land-surface property changes through detecting and analyzing multiple mega- fires in the U.S.  Develop an observation based scheme to parameterize the land-surface property changes to be used in CESM

6. Remote Sensing Data MTBS Wildland Fire Data GHCN Weather Data RS Data Processing Rasterization Extract Ground Weather Data Time Series of Land Properties Wildland Fire Burned Area Map Time Series of Weather Data Surface Vegetative Changes Surface Radiative Changes Surface Hydrological Changes Statistical Analysis and Inference Impacts of Large Fires Spatio- Temporal Data Integration and Analysis Detection ProcedureData collection Data analysis Surface change Statistics MODIS: Global, 500 m and 1000 m resolutions, 2000-present, 8-day products, Land cover type data (MCD12), Surface Albedo Data (MCD43A3), Surface Reflectance Data (MOD09A1), Leaf Area Index (LAI) data (MOD15A2), Land Surface Temperature Data (MOD11A2) Monitoring Trends in Burning Severity (MTBS) (USGS and USFS) Burned area and severity map. CONUS, 30 m resolution, 1984-present

7.  Spatial: comparison between burned and control pixels - Select burned pixels and control pixels with same prefire land cover type and very close surface albedo. - Compare for fire year and postfire years. Temporal: comparison before and after a fire - Significance test of differences between fire year and prior year T-1 T T+1 Change Difference X X+1

8. Fire events Group No. Fire Name Time State Lat Long Burned Acres Climate 1 1 Cedar 10/25-11/5, 2003 CA 33.30 -116.80 280k Warm 2 Zaca 7/4-9/4, 2007 CA 34.60 -119.70 240k Warm 3 Egley 7/6-7/25, 2007 OR 43.49 -119.23 140k Cool 4 Rodeo 6/8-9/1, 2002 AZ 34.11 -110.49 260k Warm 5 Derby 8/21-11/7, 2006 MT 45.54 -109.93 210k Cool 6 Big Turnaround 5/5-10/10, 2007 GA 30.77 -82.27 305k Warm 7 Biscuit 7/12-7/15, 2002 OR 42.27 -123.54 495k Cool 2 8 Crystal 8/15-11/9, 2006 ID 42.96 -113.24 220k Cool 9 Winters 7/25-9/10, 2006 NV 41.36 -116.91 240k Cool 10 S Nevada 6/22-7/10, 2005 NV 37.20 -114.38 240k Warm 11 Milford Flat 7/6-8/3, 2007 UT 38.58 -112.97 360k Cool 12 Cave Creek 6/21-8/1, 2005 AZ 33.98 -111.82 250k Warm

9. LAI – Group 1 (≥ 1.0) Cedar Fire Zaca Fire Egley Fire

10. LAI – Group 2 (<1.0)

11. LAI change – Group 1 (LAI≥1.0) Cedar Fire LAI change – Group 2 (LAI<1.0)

12. NDVI change – Group 1 (NDVI ≥ 0.5) Cedar Fire

13. Day temperature change – Group 1

14. Albedo change – Group 1 Cedar Fire Egley Fire

15. Forest vs. shrub/grass (Egley fire) fire period

16. Scheme development procedure Detected change Fit with natural exponential function Trend Remove trend from detected change Fluctuation Fourier analysis Periodic variation Amplitude of the year Simulated change Max – Min within a year

17. Before and after fire termination t0 Cedar fire Cedar fire case

18. Separate into trend and fluctuation F(t)= a0=-0.597 a1=-0.0073 a2=0.00039

19. Express fluctuation as multiple of periodic variation and amplitude for fluctuation b0=-1.851 b1=-0.444 b2=0.0403

20. Simulation scheme F(t)= + x

21. Summary Land-surface changes are remarkable only if land coverage meets certain thresholds. Large changes over 4-10 years, longer in cool regions, larger for forested lands, and opposite phases for albedo. Post-fire changes are featured by long-term deceasing trend and fluctuation, which contains periodic variations and decaying amplitude. The scheme developed based on these features would provide a tool for simulating the regional climate effects of wildfires with CESM.

22. Plan for next year • Detect more fires • Detect atmospheric properties for evaluating simulation of climate impacts • Couple with CESM • Simulate the climate impacts of fires

23. Thanks! Funding support from USDA NIFA

Add a comment