BEFORE: It's quite straightforward to use GFS FNL as initial and boudanry conditions for WRF runs (For US regions, one could use NARR, which is higher in resolutions). The most advantage of FNL is that the data is stored in time-sorted formats. For each of those files, it contains alll the variables that needed for WRF runs. But for ERA, things are little different. The ERA data is stored in variable-sorted format. One need to choose variables and the times they need for their runs. Vtable should be modified also to adjust to the data.

1. I tried to use the batch mode to access the data, but failed....do not know why. I used perl script, anyone who is interested to try on this, just go ahead, or refer to the references at the end. Here I just show how to do this manually...sometimes, you want to take a short cut, you get lost. I've been there, trust me.

2. Check all the variables you need for WRF runs. go to http://apps.ecmwf.int/datasets/data/interim_full_daily/

They are: 130/131/132/157 for the pressure level data (choose all available levels);

134/151/165/166/167/168/235/33/34/31/141/139/170/183/236/39/40/41/42 for the surface data.

129/172 for time-invariant data. the codes listed here can be interpreted by looking up in table 128 (http://old.ecmwf.int/publications/manuals/d/gribapi/param/filter%3Dgrib1/order%3DparamId/order_type%3Dasc/p%3D1/table%3D128) 

3. Select the dates, and time (0,6,12,18). Step is '0'. and then just click 'Retrieve GRIB'. (You may need to register before the download). There are multiple choices of spatial resolutions, pick up the most suitable one. the highest is not necessarily the best, sometimes it takes longer to process.

4. About the Vtable. There'are two tables for ERA,  'Vtable.ERA-interim.ml' and 'Vtable.ERA-interim.pl'. Anyway, I think those tables just serve the function of look up by ungrib.exe and metgrid.exe. Different variables in those two tables, which is not convenient for us to use, so here we just combine them into one Vtable, see the attached file.

5. Following things are the same as using FNL data.

6. Two last things :

Add 'missing_value=-1.E30 to "ST" and "SM" variables (e.g., ST000007) in METGRID.TBL.ARW

You need to modify 'num_metgird_levels=27' in the namelist.input. It should be 38 for ERA-interm.

Other useful references:

[1] http://forum.wrfforum.com/viewtopic.php?f=22&t=1001&start=20

[2] http://www.geomar.de/mitarbeiter/fb1/me/jharlass/wrf/erai/ 

[3] http://blog.sina.com.cn/s/blog_d8f6ec6b0102v2pi.html