Running WRF beyond the WRF DART tutorial

Preface

We strongly encourage people to learn the fundamentals before attempting the big problem. Before running the WRF-DART tutorial, and certainly before you attempt to run WRF-DART beyond the WRF-DART tutorial, you should become very familiar with WRF - and with DART - separately.

I would like to know if the following link is the most updated dart-tutorial and install instructions?

Ed note: I removed the link because it was not one people should be using in general.

You are asking three different questions.

1. Installing DART
2. Running the DART tutorial
3. (for people who have completed steps 1,2) Running the WRF-DART tutorial

To become familiar with DART: https://dart.ucar.edu/pages/Getting_Started.html

To run the DART tutorial (HIGHLY RECOMMENDED): https://dart.ucar.edu/pages/Tutorial.html

And then, after you've finished both of those - the tutorial for wrf and dart is models/wrf/tutorial/README.md on the Manhattan and master branches of the public NCAR/DART GitHub repository: https://github.com/NCAR/DART

Running WRF-DART with my own data.

I have a run of 3 nested domains, so I have all the preprocess data. I am reusing the scripts but adapting them to my system and my work. After having the perturbation bank, if they generate the initial conditions, first with the gen_retro_icbc.csh script and then with the init_ensemble_var.csh script.
Now, I have a doubt and I can't understand the need for some files.
Why does REAL run twice and generate two wrfinput files every 6 hours? What is the function of these two files and why each time interval?
Thank you very much for your help.

For the outermost domain, you need to provide lateral boundary conditions. A lateral boundary condition file contains three conditions at a minimum: the current atmospheric state (along the inside edge of your outer model domain), a target state, meaning the forced conditions along the lateral boundaries at the end of an integration period, here at 6 hours (also along the inside edge), and a linear trend that defines the rate of change from the initial to target state. When you use the perturbations generated by the perturbation bank for the integration of the ensemble between analysis steps, they are added to the target state, with unique sets of perturbations added to each ensemble member. This helps maintain spread in the ensemble by providing a source of error growth where flow is coming in from the lateral boundaries. Full wrfinput files are needed at the beginning and the end of the integration window in order to use the tool, pert_wrf_bc, that updates the boundary condition file to 1) match the initial state to what was just updated by DART's filter, 2) change the target state for each ensemble member using a randomly selected perturbation set with the uniform future target state generated in the init_enesmble_var.csh script, and 3) update the tendencies based on the new initial and target state values.