Inconsistent in Log File from WRF-Fire

[zli867]

Hi Dr. Mandel,
I found an inconsistent in WRF-Fire log file. I paste the last time step from log file. As you can see the cumulative burned area is 0.544E+07 m^2 at this time step and cumulative burned short grass is 0.3629E+07kg. The fgi setting for short grass in this simulation is 0.166 kg/m2 (mass loading of surface fuel; I use the default setting as you provided for namelist.fire). Assume all burned area is the short grass, we can estimate the upper bound of the total burned short grass: 0.544E+07 m^2 * 0.166 kg/m2 = 903040kg = 0.9E+06kg which is much less than 0.3629E+07kg in log file. Do you think it is a problem from WRF-Fire or from my analysis? If it is an issue from WRF-Fire, I suspect that the output in log (burned fuel and emissions) are wrongly multiply the coarse grid size not the subgrid size

WRF-SFIRE/phys/module_fr_sfire_atm.F

Line 1089 in 681dc5a

s_chem = s_chem*dx*dy

.

Timing for main: time 2021-04-06_23:59:59 on domain   1:    0.24224 elapsed seconds
SFIRE: domain  1
SFIRE:Time   86399.000 s Average surface wind   0.374E-01 m/s
SFIRE:Time   86399.000 s Maximum surface wind   0.445E+01 m/s
SFIRE:Time   86399.000 s Fire area              0.544E+07 m^2
SFIRE:Time   86399.000 s Heat output            0.150E+10 W
SFIRE:Time   86399.000 s Max heat flux          0.537E+05 W/m^2
SFIRE:Time   86399.000 s Latent heat output     0.143E+09 W
SFIRE:Time   86399.000 s Max latent heat flux   0.513E+04 W/m^2
SFIRE:add_fire_emissions: chem_opt=  0 tracer_opt=  2 num_chem=  0 num_tracer=  9 max tracer=  9
SFIRE:array tracer has dimensions    1: 49   1: 35   1: 36   1:  9
SFIRE:array chem has dimensions    1: 49   1: 35   1: 36   1:  0
SFIRE: tracer   tr17_1   tr17_2   tr17_3   tr17_4   tr17_5
SFIRE:Pointer        2        3        4        5        6
SFIRE: tracer   tr17_6   tr17_7   tr17_8
SFIRE:Pointer        7        8        9
SFIRE:Fire tracers into atmosphere level  1
SFIRE:Inserting chemical species
SFIRE:Computing totals over all processes
SFIRE:Computing totals of tracers over all processes
SFIRE:Total emissions in g
SFIRE:Emissions      tr17_1     tr17_2     tr17_3     tr17_4     tr17_5
SFIRE:Cumulative  0.422E+11  0.515E+10  0.239E+09  0.130E+10  0.175E+10
SFIRE:Rate per s  0.887E+06  0.109E+06  0.420E+04  0.264E+05  0.376E+05
SFIRE:
SFIRE:Emissions      tr17_6     tr17_7     tr17_8
SFIRE:Cumulative  0.178E+09  0.188E+10  0.780E+06
SFIRE:Rate per s  0.363E+04  0.408E+05  0.154E+02
SFIRE:
SFIRE:1: Short grass (1 ft)          burned    0.3629E+07kg, rate     7.484     kg/s
SFIRE:2: Timber (grass and understor burned    0.3490E+08kg, rate     685.7     kg/s
SFIRE:3: Tall grass (2.5 ft)         burned    0.7895E+08kg, rate     1121.     kg/s
SFIRE:4: Chaparral (6 ft)            burned    0.3254E+07kg, rate     41.80     kg/s
SFIRE:5: Brush (2 ft)                burned    0.1227E+08kg, rate     33.27     kg/s
SFIRE:7: Southern rough              burned    0.1358E+08kg, rate     148.1     kg/s
SFIRE:8: Closed timber litter        burned    0.9894E+08kg, rate     2216.     kg/s
SFIRE:9: Hardwood litter             burned    0.1721E+09kg, rate     3700.     kg/s
SFIRE:10: Timber (litter + understor burned    0.5560E+08kg, rate     1581.     kg/s
SFIRE:                    Total fuel burned    0.4732E+09kg, rate     9534.     kg/s
Timing for sfire:    0.09569 elapsed seconds
Timing for main: time 2021-04-07_00:00:00 on domain   1:    0.24292 elapsed seconds
Timing for Writing wrfout_d01_2021-04-07_00:00:00 for domain        1:    2.92382 elapsed seconds
d01 2021-04-07_00:00:00 wrf: SUCCESS COMPLETE WRF

Thanks,
Zongrun

[janmandel]

@Porkking
Hi Zogrun, may I have your namelist.input rsl.error.0000 from the run, please?
I think the discrepancy is because the "Fire area" number is within the patch the rsl file is for, this was meant for quick diagnotiscs only, but the numbers from the emissions are added up over the whole domain by MPI.
I should add some more descriptive prints.
It seems you are running with ifire=1 (WRF-SFIRE) not ifire=2 (WRF-Fire, which we contributed to WRF back in 2012 but then it underwent changes at NCAR we were not involved in)
Jan

[zli867]

Hi Dr. Mandel,
You are right. I am running WRF-SFIRE model. Here is my namelist.input.
namelist.input.txt
I shared the rsl.error.0000 via drive: https://drive.google.com/file/d/1nffAeYQSqzUrzUGyvRi92BXFds6GELVL/view?usp=sharing
I do not think burned area has something wrong. I calculate it based on WRF output (FUEL_FRAC * subgird_mesh_area) and it close to what I have in the log file.
I also estimate emission from WRF-FIRE by the WRF-FIRE output and input. The emission in log file is 100 times of my estimation. My method is exactly the same as your code here except the dx and dy I used. I use the subgrid output for FUEL_FRAC and fgip (fuel load for each fuel category) from namelist.fire to estimate fuel_burnt and then use emission factor from namelist.fire_emission to estimate emis. The difference is that I use subgrid dx and dy at the last step to estimate s_fuel. If you use meshgrid length of the domain rather than the subgrid length, we will have 100 times difference since my subgrid ratio is 10.

WRF-SFIRE/phys/module_fr_sfire_atm.F

Line 724 in 681dc5a

fuel_burnt = fgip(i_f,j_f) * fuel_frac_burnt(i_f,j_f) ! kg/m^2

WRF-SFIRE/phys/module_fr_sfire_atm.F

Line 955 in 681dc5a

emis=dust_3 (cat)*fuel_burnt ! emission from fire cell

WRF-SFIRE/phys/module_fr_sfire_atm.F

Line 1075 in 681dc5a

s_fuel = s_fuel*dx*dy

Thanks,
Zongrun