Bias is growing with more spin up years

General scientific issues regarding ROMS

Moderators: arango, robertson

Post Reply
Message
Author
Sankar
Posts: 13
Joined: Thu Jan 28, 2021 4:05 am
Location: Indian Institute of Technology Delhi
Contact:

Bias is growing with more spin up years

#1 Unread post by Sankar »

Hello everyone. I am setting up the roms model for the Indian Ocean region using climatology files. I have given initial and boundary files from World Ocean Atlas 18. The atmospheric forcing file is from IMDAA and Scatscat (for wind). When I am spinning up the model, the biases grow with time. For example, the SST bias is 2deg C at the end of the first year, but it becomes 4deg C at the end of the 10th year. I am changing the mixing scheme and viscosity value, but nothing works. Part of the .h file is added here. I am looking forward if someone can help me out here. Please let me know of any further details. The resolution of my model is 0.16deg with 40 vertical levels.


#define MIX_GEO_UV
#define MIX_GEO_TS
#define UV_VIS2
#define VISC_GRID
#define DIFF_GRID
#define UV_QDRAG
#define UV_U3HADVECTION
#define UV_C4VADVECTION
#define NONLINEAR
#define NONLIN_EOS
#define POWER_LAW
#define PROFILE
#undef SPLINES
#define SCORRECTION

#define BULK_FLUXES
#ifdef BULK_FLUXES
# undef LONGWAVE_OUT
# undef LONGWAVE
# define EMINUSP
#endif
#define LMD_MIXING
#ifdef LMD_MIXING
# define LMD_RIMIX
# define LMD_CONVEC
# define LMD_SKPP
# define LMD_BKPP
# define LMD_DDMIX
# define LMD_NONLOCAL
#endif
#undef ASSIMILATION_SST
#define ANA_BSFLUX
#define ANA_BTFLUX

User avatar
wilkin
Posts: 875
Joined: Mon Apr 28, 2003 5:44 pm
Location: Rutgers University
Contact:

Re: Bias is growing with more spin up years

#2 Unread post by wilkin »

I see you are using net longwave radiation forcing.

Does IMDAA have a separate analysis for downward longwave only (i.e. what ROMS expects as lwrad_down)?

If it does, you might find #define LONGWAVE_OUT helps because your growing SST might be reined in by increasing upward outgoing longwave emission that grows as SST to the 4th power. As things stand, your net heat flux is entirely imposed from IMDAA and not mediated by anything going on in ROMS to be self limiting.

Also, many atmospheric analysis products over-estimate downward shortwave radiation, so you might check that.

In the North American Regional Reanalysis the bias in swdown is as much as 40 W/m^2. (see Kennedy, A.D., Dong, X., Xi, B., Xie, S., Zhang, Y. and Chen, J., 2011. A comparison of MERRA and NARR reanalyses with the DOE ARM SGP data. Journal of climate, 24(17), pp.4541-4557.)
John Wilkin: DMCS Rutgers University
71 Dudley Rd, New Brunswick, NJ 08901-8521, USA. ph: 609-630-0559 jwilkin@rutgers.edu

Sankar
Posts: 13
Joined: Thu Jan 28, 2021 4:05 am
Location: Indian Institute of Technology Delhi
Contact:

Re: Bias is growing with more spin up years

#3 Unread post by Sankar »

Thank you for your response, Wilkin. IMDAA has both net Surface Net Long Wave Radiation Flux and Surface Downward Long Wave Radiation Flux. However, as you mentioned, I have used the net longwave radiation flux. In the varinfo.dat, I have seen the variable radlw_in for downwelling longwave radiation flux. So are you suggesting keeping both net longwave as well as downward longwave.

Your last suggestion is indeed very important while giving atmospheric forcings. Due to its high resolution (12 km), I have chosen the IMDAA dataset. But as you have said, I will change the forcing and check.

Besides temperature, biases are also growing in the salinity and at the current speed. For salinity, I used SCORRECTION and changed the mixing schemes as well as the viscosity value, but nothing worked.

For reference here, I am attaching the 1st year and 3rd year run biases (although the model is not stable but for reference, I am attaching the 3rd year run images).
Attachments
Salt_biases_1st_year.png
temp_biases_1st_year.png
resultant_current_biases_1st_year.png
Salt_biases_3rd_year.png
temp_biases_3rd_year.png
resultant_current_biases_3rd_year.png

User avatar
wilkin
Posts: 875
Joined: Mon Apr 28, 2003 5:44 pm
Location: Rutgers University
Contact:

Re: Bias is growing with more spin up years

#4 Unread post by wilkin »

My suggestion was to #define LONGWAVE_OUT and #undef LONGWAVE. Don't set both - that makes no sense. Look in bulk_fluxes.F to see what these options do.

LONGWAVE_OUT will cause ROMS to read lwrad_down (or whatever you renamed it to be in varinfo.dat) but compute outward longwave from model SST.

However, before you test that out I just noticed in your original post that you have not defined SOLAR_SOURCE. You should #define SOLAR_SOURCE. That causes the solar radiation to penetrate into the water column over depth scales set by the water type WTYPE option in roms.in. In the oligotrophic waters of the Indian Ocean you probably want WTYPE = 1.

When SOLAR_SOURCE is not defined, the shortwave radiation is all absorbed in the top model layer. That will overheat SST.
John Wilkin: DMCS Rutgers University
71 Dudley Rd, New Brunswick, NJ 08901-8521, USA. ph: 609-630-0559 jwilkin@rutgers.edu

Sankar
Posts: 13
Joined: Thu Jan 28, 2021 4:05 am
Location: Indian Institute of Technology Delhi
Contact:

Re: Bias is growing with more spin up years

#5 Unread post by Sankar »

Thank you, Prof. Wilkin, for your response. I will definitely try the lwrad_down as you advised. Before that, I would like to mention to you that I have already used the solar source and WTYPE = 1. I have also mailed you the log file. I hope this will help you understand my model setup in detail. Please let me know if I missed out something in setting up the model.

Post Reply