I am currently running a regional nesting suite driven by ERA5 data and would like to change the initial conditions to formulate an ensemble, but it’s not in the traditional way (hence I’m not using the regional ensemble suite). Currently I’m working on u-da773, with a view to copy and change it to implement other conditions.
What I’d like to do is replace the initial conditions (but not the LBCs) with data from ERA5 files at different times, but have the LBCs and model time stay as it is.
I think one way to do this would be using the transplant data option, but I have no idea how to get started with that and I’m not sure it differentiates between initial conditions and lateral boundary conditions. Alternatively I suppose I could hard-code changes to AINITIAL or some other variables? But I’m not quite sure what to do and would appreciate any guidance or suggestions!
Just popping another message in here before the thread gets auto-closed after a month since posting: I have not managed to make any more progress on this. My current situation is as follows…
I understand that there is a method to use data transplanting. It gets to the point where you get a memory error. I have no idea how to make it work and I can’t find any guides for it anywhere?
There is maybe something else I could do with the reconfiguration step as outlined in UMDP 302 part 2: https://code.metoffice.gov.uk/doc/um/vn13.2/papers/umdp_302.pdf. Perhaps if I could create an external dump from alternative reconfiguration of ECMWF data from a different time, it would be possible to select option 7 to use that dump as initial conditions for some variables - but that would require creating entries for every prognostic variable in um > namelist > Reconfiguration and Ancillary Control > Configure ancils and initialise dump fields… which feels a little inefficient?
I have also now tried changing AINITIAL to try to get the initial conditions to pick up a different time while boundary conditions get the same time (see suite u-db026). But there’s some sort of sensible checking as the model runs which throws an error when there’s disagreement between the initial condition times and the lateral boundary condition times (an INBOUNDA “Boundary data starts before the start of the current boundary data interval” error) - which of course makes a lot of sense and is a very sensible error to throw. But I don’t know how to stop that error from happening.
I’m hoping to reach out to some colleagues at the Met Office about this but thought I’d put this here for anyone reading this who might be interested in future… and in case anyone sees it and has any suggestions! Thanks!