What To Do When GEOSldas Does Not Run

If GEOSldas successfully completes a simulation, the log file:

[EXP_PATH]/[EXP_ID]/output/[EXP_DOMAIN]/rc_out/Y[BEG_DATE_YEAR]/M[BEG_DATE_MONTH]/[EXP_ID].ldas_log.[BEG_DATE]z.txt

contains line "GEOSldas Run Status: 0". This line is written by GEOSldas.x and does not imply successful completion of the post-processing commands in the GEOSldas job file lenkf.j.

If GEOSldas does not complete successfully:

1. Inspect the log and error files, which are then still in the ./scratch directory ([EXP_PATH]/[EXP_ID]/scratch/). The log file GEOSldas_log_txt provides a rough idea where the run failed, and the error file GEOSldas_err_txt should provide some backtrace information. It is important to distinguish two types of run failures: (a) Clean exit with an "error" message in the log file, which most likely points to a configuration error (see below) or reflects a problem in an input file. (b) Crash (or unexpected failure), which is most likely a source code error (see backtrace information in the error file).

2. The LDAS Nightly Tests provide sample configurations. For example, the "global" test case includes a 24-hour model simulation on the global 36-km EASEv2 grid and a 24-hour assimilation run of SMAP brightness temperature observations with perturbations on the global 9-km EASEv2 grid. While the test cases can be useful to learn about GEOSldas configuration options, they are not meant for science.

3. If the problem is not a source code or GEOSldas configuration error, make sure that nothing in your [t]csh (or bash) environment conflicts with GEOSldas by comparing your ~/.cshrc (or ~/.bashrc) files with those of other GEOSldas users.