search

usgs / pestpp

tools for scalable and non-intrusive parameter estimation, uncertainty analysis and sensitivity analysis
127 stars 69 forks source link

Noobsreref option from a PEST control file causes PEST++ to fail. #303

Open rbwinst-usgs opened 1 month ago

rbwinst-usgs commented 1 month ago

PESTPPTest.zip The attached model causes PEST++ to fail. The error message indicated that the noobsreref option on line 5 is the problem. I was surprised that this error message was not saved to one of the PEST++ output files instead of just being written to the screen. The noobsreref option is supported by PEST but evidently not by PEST++. Just removing the noobsreref option by itself does not work because then PEST++ fails because of the trailing comment on the same line. Removing the noobsreref option and the following comment allows PEST to continue until it gets to the observation groups section where it stops because of the presence the names of the covariance files. It would be nice if PEST++ could either handle the noobsreref option with a following comment or handle the case where noobsreref is absent but a trailing comment is still present. Obviously, being able to use the covariance files could also be an improvement but I understand that would be a considerably larger task.

jtwhite79 commented 1 month ago

The control data section issue will be fixed in the next release (the error should also now be recorded in the rec file). Not sure we will ever move to supporting covariance based regularization - if it would help you, we can move that from an error to a warning...

In general tho, the control data section is becoming increasingly hard to support - the logic rules about what options are expected where and when is getting complex! This was a major motivation to move the version 2 control file...not sure if you want to support that in model muse tho...

rbwinst-usgs commented 1 month ago

Thanks. I don't see any need for you to support covariance files just for me. I would like to more fully support PEST++ at some point. There is a new guy who is planning on working on ModelMuse soon. Once that happens, support for PEST++ may move up in the priority queue.