Scheduled Downtime
On Friday 21 April 2023 @ 5pm MT, this website will be down for maintenance and expected to return online the morning of 24 April 2023 at the latest

WRF Output Discrepency

U

utkarsh-7

New member
Hi,
I was actually trying to run the WRF model using 2 configurations -
1) 48 hr duration (2022-07-18_12:00:00 to 2022-07-20_12:00:00)
2) 48 hr duration but in 3hr chunks (Simulation_1: "2022-07-18_12:00:00 to 2022-07-18_15:00:00", ... ,Simulation_16: "2022-07-20_09:00:00 to 2022-07-20_12:00:00")
even after Aggregating (curr_chunk_rain = prev_chunk_rain + curr_chunk_rain)
same met data(2022-07-18_12:00:00 to 2022-07-20_12:00:00) and namelist files(except the time ranges)
So, when I observed the output of both these simulations for the first 3 hrs, I observed that for RAINNC variable there was a huge difference in the output of both these simulations though they were actually running for same region and same duration(first 3 hrs).

Here is the Output of both these for your reference -
1) 48 hr duration (2022-07-18_12:00:00 to 2022-07-20_12:00:00) -
_c3 Column = "RAINNC" in mm
2023-09-19 06:00:18,336 - aggregator - INFO - {'_c0': '36.44705', '_c1': '-83.300964', '_c2': '2022-07-18T12:00:00.000', '_c3': '0.0'}
2023-09-19 06:00:18,340 - aggregator - INFO - {'_c0': '36.44705', '_c1': '-83.300964', '_c2': '2022-07-18T13:00:00.000', '_c3': '0.0'}
2023-09-19 06:00:18,343 - aggregator - INFO - {'_c0': '36.44705', '_c1': '-83.300964', '_c2': '2022-07-18T14:00:00.000', '_c3': '0.0'}
2023-09-19 06:00:18,346 - aggregator - INFO - {'_c0': '36.44705', '_c1': '-83.300964', '_c2': '2022-07-18T15:00:00.000', '_c3': '0.0'}
2023-09-19 06:00:18,347 - aggregator - INFO - {'_c0': '36.44705', '_c1': '-83.300964', '_c2': '2022-07-18T16:00:00.000', '_c3': '0.0'}
2023-09-19 06:00:18,349 - aggregator - INFO - {'_c0': '36.44705', '_c1': '-83.300964', '_c2': '2022-07-18T17:00:00.000', '_c3': '0.0'}
2023-09-19 06:00:18,351 - aggregator - INFO - {'_c0': '36.44705', '_c1': '-83.300964', '_c2': '2022-07-18T18:00:00.000', '_c3': '0.0'}
2023-09-19 06:00:18,353 - aggregator - INFO - {'_c0': '36.44705', '_c1': '-83.300964', '_c2': '2022-07-18T19:00:00.000', '_c3': '0.0026964406'}
2023-09-19 06:00:18,355 - aggregator - INFO - {'_c0': '36.44705', '_c1': '-83.300964', '_c2': '2022-07-18T20:00:00.000', '_c3': '0.009049861'}
2023-09-19 06:00:18,356 - aggregator - INFO - {'_c0': '36.44705', '_c1': '-83.300964', '_c2': '2022-07-18T21:00:00.000', '_c3': '25.35142'}
2023-09-19 06:00:18,357 - aggregator - INFO - {'_c0': '36.44705', '_c1': '-83.300964', '_c2': '2022-07-18T22:00:00.000', '_c3': '32.02499'}
2023-09-19 06:00:18,359 - aggregator - INFO - {'_c0': '36.44705', '_c1': '-83.300964', '_c2': '2022-07-18T23:00:00.000', '_c3': '32.032722'}
2023-09-19 06:00:18,361 - aggregator - INFO - {'_c0': '36.44705', '_c1': '-83.300964', '_c2': '2022-07-19T00:00:00.000', '_c3': '32.369686'}
Click to expand...
2) 48 hr duration but in 3hr chunks (Simulation_1: "2022-07-18_12:00:00 to 2022-07-18_15:00:00", ... ,Simulation_16: "2022-07-20_09:00:00 to 2022-07-20_12:00:00") -
even after Aggregating (curr_chunk_rain = prev_chunk_rain + curr_chunk_rain)
_c3 Column = "RAINNC" in mm
2023-09-19 06:00:15,350 - aggregator - INFO - {'_c0': '36.44705', '_c1': '-83.300964', '_c2': '2022-07-18T12:00:00.000', '_c3': '0.0', '_c4': '1.6581456E9'}
2023-09-19 06:00:15,352 - aggregator - INFO - {'_c0': '36.44705', '_c1': '-83.300964', '_c2': '2022-07-18T13:00:00.000', '_c3': '0.0', '_c4': '1.6581492E9'}
2023-09-19 06:00:15,353 - aggregator - INFO - {'_c0': '36.44705', '_c1': '-83.300964', '_c2': '2022-07-18T14:00:00.000', '_c3': '0.0', '_c4': '1.6581528E9'}
2023-09-19 06:00:15,354 - aggregator - INFO - {'_c0': '36.44705', '_c1': '-83.300964', '_c2': '2022-07-18T15:00:00.000', '_c3': '0.0', '_c4': '1.6581564E9'}
2023-09-19 06:00:15,356 - aggregator - INFO - {'_c0': '36.44705', '_c1': '-83.300964', '_c2': '2022-07-18T16:00:00.000', '_c3': '0.0', '_c4': '1.65816E9'}
2023-09-19 06:00:15,359 - aggregator - INFO - {'_c0': '36.44705', '_c1': '-83.300964', '_c2': '2022-07-18T17:00:00.000', '_c3': '0.0', '_c4': '1.6581636E9'}
2023-09-19 06:00:15,360 - aggregator - INFO - {'_c0': '36.44705', '_c1': '-83.300964', '_c2': '2022-07-18T18:00:00.000', '_c3': '0.0', '_c4': '1.6581672E9'}
2023-09-19 06:00:15,362 - aggregator - INFO - {'_c0': '36.44705', '_c1': '-83.300964', '_c2': '2022-07-18T19:00:00.000', '_c3': '0.0', '_c4': '1.6581708E9'}
2023-09-19 06:00:15,363 - aggregator - INFO - {'_c0': '36.44705', '_c1': '-83.300964', '_c2': '2022-07-18T20:00:00.000', '_c3': '1.8566412E-5', '_c4': '1.6581744E9'}
2023-09-19 06:00:15,365 - aggregator - INFO - {'_c0': '36.44705', '_c1': '-83.300964', '_c2': '2022-07-18T21:00:00.000', '_c3': '8.5701403E-4', '_c4': '1.658178E9'}
2023-09-19 06:00:15,366 - aggregator - INFO - {'_c0': '36.44705', '_c1': '-83.300964', '_c2': '2022-07-18T22:00:00.000', '_c3': '0.03614223375916481', '_c4': '1.6581816E9'}
2023-09-19 06:00:15,368 - aggregator - INFO - {'_c0': '36.44705', '_c1': '-83.300964', '_c2': '2022-07-18T23:00:00.000', '_c3': '0.142098069190979', '_c4': '1.6581852E9'}
2023-09-19 06:00:15,370 - aggregator - INFO - {'_c0': '36.44705', '_c1': '-83.300964', '_c2': '2022-07-19T00:00:00.000', '_c3': '0.14225231111049652', '_c4': '1.6581888E9'}
Click to expand...
For example in the below files if you will observe then at TIMESTAMP="2022-07-18T21:00:00.000" the RAIN value for appoach1 is 25.35142mm and for approach2 it is 8.5701403E-4 nowhere near the value for single-core. And these values keep diverging as we go further.

Can you please explain why is this happening? And also is there a way to minimize the difference b/w approach1 and approach2?
 
Hi,
Just to be sure I'm understanding correctly, are you saying that if you have 2 namelist setups, that are otherwise identical except that

namelist 1) 2022-07-18_12:00:00 to 2022-07-20_12:00:00
namelist 2) 2022-07-18_12:00:00 to 2022-07-18_15:00:00

If you compare RAINNC values at the time stamp of 2022-07-18_15:00:00 between the 2 runs, the values are different?

Can you attach the two different namelists you used for these 2 methods? Thanks!
 
Yes I have two setups
Approach 1 - Single Namelist file with dates - [2022-07-18_12:00:00 to 2022-07-20_12:00:00]
Approach 2 - Each Namelist file having 3 hour chunk of the above 48 hr duration -
Namelist1 - 2022-07-18_12:00:00 to 2022-07-18_15:00:00
Namelist2 - 2022-07-18_15:00:00 to 2022-07-18_18:00:00
....
Namelistn- 2022-07-20_09:00:00 to 2022-07-20_12:00:00
When I compare the data b/w approach1 and approach2(combining each 3 hour chunk to make 48 hour, also handling the RAINNC accumulation - adding previous 3hr accumulation to curr so on) There is huge difference b/w the RAINNC values.
Namelist using - Just changing the date values in the namelist for each chunk/approach
&time_control
start_year = 2018, 2018
start_month = 7, 7
start_day = 15, 15
start_hour = 15, 15
end_year = 2018, 2018
end_month = 7, 7
end_day = 15, 15
end_hour = 18, 18
interval_seconds = 10800
input_from_file = .true., .true.
history_interval = 60, 60
frames_per_outfile = 100, 100
restart = .false.
restart_interval = 7200
io_form_history = 2
io_form_restart = 2
io_form_input = 2
io_form_boundary = 2
start_minute = 0, 0
start_second = 0, 0
end_minute = 0, 0
end_second = 0, 0
nocolons = .true.
/

&domains
time_step = 40
time_step_fract_num = 0
time_step_fract_den = 1
max_dom = 2
e_we = 31, 31
e_sn = 31, 31
e_vert = 33, 33
p_top_requested = 5000
num_metgrid_levels = 32
num_metgrid_soil_levels = 4
dx = 9000.0, 3000.0
dy = 9000.0, 3000.0
grid_id = 1, 2
parent_id = 1, 1
i_parent_start = 1, 11
j_parent_start = 1, 11
parent_grid_ratio = 1, 3
parent_time_step_ratio = 1, 3
feedback = 1
smooth_option = 0
/

&physics
physics_suite = 'CONUS'
mp_physics = 0, 0
cu_physics = 0, 0
radt = 9, 3
bldt = 0, 0
cudt = 0, 0
icloud = 0
num_land_cat = 21
sf_urban_physics = 1, 1
/

&dynamics
hybrid_opt = 2
w_damping = 0
diff_opt = 1, 1
km_opt = 4, 4
diff_6th_opt = 0, 0, 0
diff_6th_factor = 0.12, 0.12
base_temp = 290.0
damp_opt = 3
zdamp = 5000.0, 5000.0
dampcoef = 0.2, 0.2
khdif = 0, 0
kvdif = 0, 0
/

&bdy_control
spec_bdy_width = 5
specified = .true.
/

&namelist_quilt
/
Click to expand...
Also attaching the RAINNC values for both approaches -
TIMESTAMPApproach1_RAINNCApproach2_RAINNC
18-07-2022 12:000.00000000000.0000000000
18-07-2022 13:000.00000000000.0000000000
18-07-2022 14:000.00000000000.0000000000
18-07-2022 15:000.00000000000.0000000000
18-07-2022 16:000.00000000000.0000000000
18-07-2022 17:000.00000000000.0000000000
18-07-2022 18:000.00000000000.0000000000
18-07-2022 19:000.00269644060.0000000000
18-07-2022 20:000.00904986100.0000185664
18-07-2022 21:0025.35142000000.0008570140
18-07-2022 22:0032.02499000000.0361422340
18-07-2022 23:0032.03272200000.1420980740
19-07-2022 00:0032.36968600000.1422523140
19-07-2022 01:0078.41131000000.1422523140
19-07-2022 02:00134.12024000000.1422523140
19-07-2022 03:00162.14316000000.1422523140
19-07-2022 04:00178.10588000000.1422523140
19-07-2022 05:00180.21828000000.1422523140
19-07-2022 06:00180.21931000000.1422683311
19-07-2022 07:00180.31068000000.1422683311
19-07-2022 08:00180.36360000000.1422861090
19-07-2022 09:00180.36360000000.1422683311
19-07-2022 10:00180.36450000000.1422683311
19-07-2022 11:00180.36464000000.1423843801
19-07-2022 12:00180.36641000000.1426712104
19-07-2022 13:00180.36707000000.1426712104
19-07-2022 14:00181.36707000000.1426712104
19-07-2022 15:00182.36707000000.1426712104
19-07-2022 16:00183.36707000000.1426712104
19-07-2022 17:00180.36719000000.1426712104
19-07-2022 18:00180.36737000000.1426712104
19-07-2022 19:00180.41635000000.1426712104
19-07-2022 20:00180.42662000000.1426712104
19-07-2022 21:00180.42670000000.1426712104
19-07-2022 22:00180.42682000000.1426712104
19-07-2022 23:00181.42682000000.1426712104
20-07-2022 00:00182.42682000000.1426712104
20-07-2022 01:00183.42682000000.1426712104
20-07-2022 02:00184.42682000000.1426712104
20-07-2022 03:00185.42682000000.1426712104
20-07-2022 04:00186.42682000000.1426712104
20-07-2022 05:00180.43942000000.1426712104
20-07-2022 06:00180.44067000000.1426712104
20-07-2022 07:00181.44067000000.1426712104
20-07-2022 08:00182.44067000000.1426712104
20-07-2022 09:00180.44113000000.1426712104
20-07-2022 10:00180.44159000000.1426712104
20-07-2022 11:00180.44853000000.1426712104
20-07-2022 12:00180.45956000000.1571877114
Click to expand...
 
Ah okay. The reason for this is because when you run for the full 48 hours, you have initial conditions (wrfinput_d0*) that come from the first guess model (your input data to the ungrib program) at the beginning of that simulation. But when you run for 3 hour intervals, you are using new initial conditions at the beginning of each 3 hour interval. So these simulations aren't actually comparable. Over time, this can cause fairly significant differences.

What is your reason for wanting to run these two different ways?
 
You must log in or register to reply here.
Top