How to understand the strength of nudging?

[aurorrmok]

Dear Sir or Madam,

I am using grid nudging in my tropical cyclone case study with three nested domains (12km, 4km, 1.3km). I would like to know (1) if grid nudging needs to be enabled for each domain? (2) Also, how should I understand the strength of nudging? I know that the timescale for nudging can be controlled individually for winds, temperature and moisture. Typically the namelist value of 0.0003 s⁻¹ is used, corresponding to a timescale of about 1 hour. However, I am using ERA5 reanalysis with a 6-hour as my WRF input data. (3) So if the value set in the namelist is 0.003 s⁻¹, how should it correspond to the nudging with the six-hour interval reanalysis data? (4) Below is my namelist. If I only set the timescale for uv nudging and do not set it for t and q, what impact will that have? Thank you in advance.

99 grid_fdda = 1, 0, 0,
100 gfdda_inname = "wrffdda_d<domain>",
101 gfdda_interval_m = 360, 360, 360,
102 gfdda_end_h = 99999, 99999, 99999,
103 if_zfac_uv = 0, 0, 0,
104 k_zfac_uv = 10, 10, 10,
105 if_zfac_t = 0, 0, 0,
106 k_zfac_t = 10, 10, 10,
107 if_zfac_ph = 0, 0, 0,
108 k_zfac_ph = 10, 10, 10,
109 if_no_pbl_nudging_uv = 1, 1, 1,
110 if_no_pbl_nudging_t = 1, 1, 1,
111 if_no_pbl_nudging_q = 1, 1, 1,
112 guv = 0.0003, 0.0000, 0.0000,
113 gt = 0.0000, 0.0000, 0.0000,
114 gq = 0.0000, 0.0000, 0.0000,
115 if_ramping = 0,
116 dtramp_min = 60.0,
117 io_form_gfdda = 2,

 

[Ming Chen]

Please see my answers below:

Dear Sir or Madam,

I am using grid nudging in my tropical cyclone case study with three nested domains (12km, 4km, 1.3km). I would like to know (1) if grid nudging needs to be enabled for each domain?

Personally I would like to do nudging over the 12km domain. This is because nudging over fine grids may be potentially harmful to mesoscale features you intend to simulate.

(2) Also, how should I understand the strength of nudging? I know that the timescale for nudging can be controlled individually for winds, temperature and moisture. Typically the namelist value of 0.0003 s⁻¹ is used, corresponding to a timescale of about 1 hour. However, I am using ERA5 reanalysis with a 6-hour as my WRF input data.

The namelist value of 0.0003 s⁻¹, which is actually the nudging factor that determines the magnitude of nudging, is independent of analysis data used for nudging. I guess this value is more or less empirical, and you can run with diffrent values to check its impact.

(3) So if the value set in the namelist is 0.003 s⁻¹, how should it correspond to the nudging with the six-hour interval reanalysis data? (4) Below is my namelist. If I only set the timescale for uv nudging and do not set it for t and q, what impact will that have? Thank you in advance.

I am not sure of this. Please refer to the literature for more information.

99 grid_fdda = 1, 0, 0,
100 gfdda_inname = "wrffdda_d<domain>",
101 gfdda_interval_m = 360, 360, 360,
102 gfdda_end_h = 99999, 99999, 99999,
103 if_zfac_uv = 0, 0, 0,
104 k_zfac_uv = 10, 10, 10,
105 if_zfac_t = 0, 0, 0,
106 k_zfac_t = 10, 10, 10,
107 if_zfac_ph = 0, 0, 0,
108 k_zfac_ph = 10, 10, 10,
109 if_no_pbl_nudging_uv = 1, 1, 1,
110 if_no_pbl_nudging_t = 1, 1, 1,
111 if_no_pbl_nudging_q = 1, 1, 1,
112 guv = 0.0003, 0.0000, 0.0000,
113 gt = 0.0000, 0.0000, 0.0000,
114 gq = 0.0000, 0.0000, 0.0000,
115 if_ramping = 0,
116 dtramp_min = 60.0,
117 io_form_gfdda = 2,

 

[aurorrmok]

Hi, Ming. Thank you for your prompt reply. I am still a bit confused about one point. Although nudging factor is independent of analysis data. if I set the strength of nudging to be shorter than the time interval of my initial field, how can nudging be performed without corresponding time data? For example, if my met_em file only has data at 00:00, 06:00, 12:00, and 18:00, but I set the nudging values to 0.0003 s⁻¹, which initial field should be referenced for nudging at times like 01:00, 02:00, etc.? Thank you.

 

[aurorrmok]

I am also confused about the two parameters "gfdda_interval_m" and "guv/gt/gq", how to understand the difference between them? Do the have the relationship? Thank you.

 

[aurorrmok]

Just want to make sure that if I set guv=0.0003, gt=0.0000, and gq=0.0000, does that mean I am only nudging the uv wind components, without nudging temperature and moisture?

 

[aurorrmok]

Personally I would like to do nudging over the 12km domain. This is because nudging over fine grids may be potentially harmful to mesoscale features you intend to simulate.

If I only apply nudging to the 12 km domain, does that mean the 4 m and 1.3 km domains are not affected by nudging? When I present the results, can I only use the 12 km domain and not the 4 km and 1.3 km domains?

 

[Ming Chen]

Please take a look at the paper :

Stauffer D. R., and N. L. Seaman, 1994: Multiscale four-dimensional data assimilation. J. Appl. Meteor., 33, 416–434.
doi:10.1175/1520-0450(1994)033<0416:MFDDA>2.0.CO;2

This paper describes in detail how nudging works in WRF.

 

[Ming Chen]

If I only apply nudging to the 12 km domain, does that mean the 4 m and 1.3 km domains are not affected by nudging? When I present the results, can I only use the 12 km domain and not the 4 km and 1.3 km domains?

In this case nudging is conducted only over D01. However, since the model results over D01 are changed due to nudging, they will affect results over D02 and D03 by feedback between parent and child domains.
A nested case is run because the fine-resolution domain is more concerned. If you are only concerned of 12km domain, then there is no need to run nested case.