Multiscale fire modeling with WRF-Sfire Adam Kochanski, M. A. - PowerPoint PPT Presentation

Multiscale fire modeling with WRF-Sfire Adam Kochanski, M. A. Jenkins, J. Mandel, J. D. Beezley, K. Yedinak, and B. K. Lamb 1

Introduction Outline:  Range of scales associated with wildland fires  Modeling of Fire-Atmosphere interactions in WRF-Sfire  Idealized LES simulations of prescribed burns • plume dynamics • thermal structure  Wildland fire smoke modeling in a coupled fire-atmosphere framework • Levels of coupling and role of fuel moisture • Plume rise and smoke dispersion forecasting • Simulating air quality impacts of wildland fires 2 2

Range of scales affecting fires • Atmospheric and fire scales Range of scales that WRF 1 m 10 cm Structural Fires Flames Flamelets Wildland Fires Global weather Mesoscale Large Eddy FDS model weather model Simulator (LES) boundary boundary boundary conditions conditions conditions 3 3

Modeling of Fire-Atmosphere interactions WRF-Sfire DATA 4 4

Idealized LES simulation of a small-scale prescribed burn (FireFlux experiment) • FireFlux prescribed burn of 155 acres (0.63 km 2 ) prairie • Model setup: • 1 domain, 1000m x 1600m, 10m horizontal resolution • 80 vertical levels from 2-1200m AGL • Fire grid resolution – 1m MT ST FireFlux picture from Clements et al. 2008 5 5

FireFlux Experiment 6 6

WRF-Sfire LES simulation of the FireFlux experiment (wind speed and water vapor shown) 12 [m/s] 12 [g/kg] 4 [m/s] 6 [g/kg] Visualization by Bedrich Sousedik 7

in- plume concentration ~3000μg /m 3 (3mg/m 3 ) particulate emission (PM 10) Idealized FireFlux simulation 8

Idealized FireFlux simulation - updraft structure w (m/s) 0 3200m SHORT TOWER MAIN TOWER 9

Timing of the fire front passage through the towers (5m and 4.5m air temperature) 10

Thermal structure of the fire plume (2m and 10m above the ground) 11

Thermal structure of the fire plume (28m and 43m above the ground) 12

Thermal structure of the fire plume 13

Fire-atmosphere interaction wind speed 14

Upward velocity at 2m and 10m AGL - short tower (WRF vs. observations) Main tower Short tower Downdrafts ahead of the fire front 15

FireFux Simulation look from the top z-vorticity (rotation) Horizontal Wind Speed Main Tower Short Tower Horizontal divergence Vertical Wind Speed 16

FireFux simulation look from a side Horizontal Wind Speed Vertical Wind Speed Main Tower Short Tower 17

Impact of the fire-atmosphere feedback on the local wind 18

Smoke modeling in a coupled fire-atmosphere framework 19

An integrated system for smoke forecasting based on WRF-Sfire WRF-SFIRE METEO OUTPUT METEO INPUT DATA Standard weather forecast WRF framework (atmosphere) ‣ wind speed and direction ‣ ARW atmospheric core ‣ air temperature HEAT AND MOISTURE ‣ Chemistry (WRF-Chem) ‣ air humidity FLUXES OF TRACER OR ‣ WPS preprocessing system CHEMICAL SPECIES ‣ precipitation LOCAL WINDS Fire Spread Model: ‣ cloudiness etc... RELATIVE HUMIDITY AIR TEMPERATURE ‣ Rothermel fire spread model PRECIPITATION ‣ Fire front tracking based on the level set method FIRE OUTPUT MOISTURE Fire Emission Model: High-resolution fire forecast: FUEL FIRE INPUT DATA ‣ smoke concentration Emission of a passive scalar or ‣ plume height chemical fluxes ‣ fire area Fuel Moisture Model ‣ fire heat flux ‣ drying and wetting due to ‣ fire intensity changes in T and RH ‣ fire rate of spread ‣ wetting due to rain ‣ fuel moisture 20 20

An integrated system for smoke forecasting based on WRF-Sfire Integrating WRF-Fire with WRF-Chem allows for a representation of interesting fire-atmosphere interactions (aerosols and radiation) 21 21

Simplified estimation of fire emissions (passive tracer) Albini Fuel Categories (13) Simplified approach – no chemistry MODIS Land Cover Types: 96h simulation done in 12h 52min • Mixed Forest • Shrublands on 640 CPUs, with the first 24h • Grasslands forecast ready in 3h 13min No chemistry Fuel consumption rates CONCENTRATION OF PASSIVE TRACERS: tracer1 user-define emission tracer2 factors for a tracer tracer3 tracer4 tracer5 tracer6 Emission of tracers tracer7 tracer8 Simplified approach – no chemistry fast

Example #1 Simulation of Barker Canyon Fire (smoke as a passive tracer) in- plume concentration ~3000μg /m 3 (3mg/m 3 ) Simulated fire perimeter Observed fire perimeter Simplified approach – no chemistry 96h simulation done in 12h 52min on 640 CPUs, with the first 24h forecast ready in 3h 13min 23

Example #1 Simulation of Barker Canyon Fire (smoke as a passive tracer) in- plume concentration ~3000μg /m 3 (3mg/m 3 ) Fuel Moisture 24

Simulated fire area and fuel moisture for Barker Canyon fire 2012 Simulated fire area and fuel moisture in- plume concentration ~3000μg /m 3 (3mg/m 3 ) 50000 22.0% Simulated fire area 45000 20.0% Observed fire area 40000 18.0% Integrated fuel moisture simulated by the fuel moisture model 35000 16.0% Fuel moisture Fire area (ha) 30000 14.0% 25000 12.0% 20000 10.0% 15000 8.0% 10000 6.0% 5000 4.0% 0 2.0% -12 0 12 24 36 48 60 72 84 96 Time since 09.09.2012 00:00 local (h) 25

Simulation of maximum plume height from 2012 Barker Canyon Fire (WA) in- plume concentration ~3000μg /m 3 (3mg/m 3 ) Braker Canyon fire (WA): diurnal variations in weather conditions translate into highly variable plume height and smoke dispersion 26

Maximum plume height simulated by WRF-Sfire vs. satellite observations (MISR) 4500� 4500� in- plume concentration ~3000μg /m 3 (3mg/m 3 ) MISR� plume� height� WRF-SFIRE� plume� height� 4000� 4000� Eleva on� 3500� 3500� (m)� 3000� 3000� ASL� (m)� Eleva on� height� 2500� 2500� 2000� 2000� Plume� 1500� 1500� 1000� 1000� 500� 500� 0� 0� 0� 10� 20� 30� 40� 50� 60� Distance� from� origin� 27

Example #2 Santa Ana fire simulation with full atmospheric chemistry Domain setup: D01 151x127x37 D02 184x142x37 D03 406x283x37 D04 712x364x37 D05 196x193x37 Time step: 120s, 40s, 13.3s 4.44s 1.48s

Estimation of fire emissions with full chemistry Albini Fuel Categories (13) 48h WSFC simulation with MOZART chemistry took 29h 56min on 324 CPUs MODIS Land Cover Types: • Mixed Forest First 24h forecast ready in 15h • Shrublands (3 times longer than passive • Grasslands racer) RADM2 MOZART bigalk NMOC: NMOC: ald bigene Fuel consumption rates csl c10h16 eth co co c2h4 hc3 no c3h5oh ch4 hc5 no2 c2h6 FINN emission factors h2 hcho so2 c3h6 no iso nh3 c3h8 no2 ket pm25i so2 ch3cooh mgly Emission of chemical pm25j ch3oh nh3 ol2 oc1 cres p25 species olt oc2 glyald oc1 oli bc1 hyac oc2 ora2 bc2 isop bc1 tol Conversion from macr bc2 xyl mek MOZART to RADM2 mvk tol 29

Simulated progression of the 2007 Santa Ana fires simulated vs. observed fire progression 10.22.2007 02:45 local 10.22.2007 05:00 local time time Observed fire area WRF-fire area � � 10.22.2007 20:00 local 10.23.2007 15:00 local time time � � 30

Simulation of smoke emissions from 2007 Santa Ana fires (Witch and Guejito) d04 (500m) in- plume concentration ~3000μg /m 3 (3mg/m 3 ) 31

Simulation of smoke emissions from 2007 Santa Ana fires (Witch and Guejito) 2km in- plume concentration ~3000μg /m 3 (3mg/m 3 ) 32

Simulated smoke emission from 2007 Santa Ana fires – WRF-Sfire vs. MODIS MODIS WRF-Sfire 2km max wind speed 32 m/s max wind speed 32 m/s 33 33

Simulation of maximum plume height from 2008 Santa Ana Fires (Witch and Guejito) in- plume concentration ~3000μg /m 3 (3mg/m 3 ) Very dry and and windy conditions during 2007 Santa Ana fires lead to almost no diurnal variability in the plume height and smoke dispersion 34

Simulated fire area for 2007 Santa Ana fires (Witch and Guejito) in- plume concentration ~3000μg /m 3 (3mg/m 3 ) 35

Simulation of PM2.5 emissions from 2007 Santa Ana fires (Witch and Guejito) 500m Simulated vs. observed PM2.5 for Escondido in- plume concentration ~3000μg /m 3 (3mg/m 3 ) 700� Observa ons� (Escondido)� 600� WRF-SFIRE-CHEM� WRF� hourly� average� 500� ( u g/m 3 ) � 400� PM 2.5� 300� 200� 100� 0� 0� 6� 12� 18� 24� 30� 36� 42� 48� 54� Time� (hr)� since� 10.21.200� 12:00� UTC� (05:00� local)� 36

Simulation of ozone from 2007 Santa Ana fires (Witch and Guejito) 2km in- plume concentration ~3000μg /m 3 (3mg/m 3 ) 75� 35� Observa ons� (Escondido)� WRF-SFIRE-CHEM� 70� 30� WRF� hourly� average� (ppb)� (ppb)� 65� 25� O 3 � O 3 � 60� 20� Simulated� Observed� 55� 15� 50� 10� 45� 5� 40� 0� 0� 6� 12� 18� 24� 30� 36� 42� 48� Time� (hr)� since� 10.21.2007� 12:00� UTC� (05:00� local)� 37