How to use ROMS For Assimilation

[m_amar1982]

I want to learn how to use ROMS for assimilation purposes. I am a beginner in this field so please help me in which ever way you can.

Yours Sincerely

Amaresh
Center for Atmospheric and Oceanic Sciences
Indian Institute of Science
Bangalore
India

[arango]

I noticed that you just joined ROMS a few days ago. The first rule of ocean modeling and data assimilation is to know your application well before you even think what data is needed to assimilate. From the scientific point of view, data assimalation cannot be treated as a black box because it requires extensive knowledge. There are a lot of books and publications on the subject. I assume that you can find these types of publications at your institution's library. Some of these books are very advanced and require knowledge in linear algebra, statistics, variational calculus, partial differential equations, and control theory.

The current released version of ROMS has an option for nudging (the simplest melding between model and observations), and optimal interpolation (OI). The OI is based on Dombrowsky and De May (1992) JGR paper. Perhaps, this paper will help you to understand what is involved.

Future versions of ROMS/TOMS will include variational data assimilation via adjoint techniques. These include strong- and weak constraint 4D Variational (4DVAR) data assimilation schemes.

I highly recommend you that you start learning the model first. ROMS is a very complex and modern code and the learning curve is high. Then, you need to set-up your application and understand very well its initialization and forcing. You need to have a knowledge of the predictability aspects in your application and the data availability before you start thinking about data assimilation.

Good luck

[slvester]

I noticed that you just joined ROMS a few days ago. The first rule of ocean modeling and data assimilation is to know your application well before you even think what data is needed to assimilate. From the scientific point of view, data assimalation cannot be treated as a black box because it requires extensive knowledge. There are a lot of books and publications on the subject. I assume that you can find these types of publications at your institution's library. Some of these books are very advanced and require knowledge in linear algebra, statistics, variational calculus, partial differential equations, and control theory.

The current released version of ROMS has an option for nudging (the simplest melding between model and observations), and optimal interpolation (OI). The OI is based on Dombrowsky and De May (1992) JGR paper. Perhaps, this paper will help you to understand what is involved.

Future versions of ROMS/TOMS will include variational data assimilation via adjoint techniques. These include strong- and weak constraint 4D Variational (4DVAR) data assimilation schemes.

I highly recommend you that you start learning the model first. ROMS is a very complex and modern code and the learning curve is high. Then, you need to set-up your application and understand very well its initialization and forcing. You need to have a knowledge of the predictability aspects in your application and the data availability before you start thinking about data assimilation.

Good luck

It is really a good advice!
I will learn use the ROMS to do assimilation and will use the Argo data.Now the ROMS include strong- and weak constraint 4D Variational (4DVAR) data assimilation schemes,I think it is difficult and maybe
OI is easyer then 4dvar.If I want to use 4dvar,what i will prepare to do?and OI?Is OI still in ROMS?I can not find it,Maybe it is a good choice to do assimilation using more easy method.Any suggestion are welcome!
Thanks!


Second institution of oceanography,state oceanic administration
HangZhou
China