Right now dgldas2 always does the full pipeline: login to EarthData, search CMR, download granules, verify sizes, then concatenate, fix time, convert accumulated depth to flux units, and delete the temp files.
We want to split that in two so the transform half is reusable on its own. Concretely, add a -skd / --skip-download flag (following the existing 3-letter short / descriptive long convention) that tells the tool the granules are already on disk — skip the login/search/download/verify steps, find the expected files in the output directory, run the same concat + time + unit-conversion logic, and don't delete the source files at the end since those now belong to the user.
This unblocks two workflows from the same binary: download-and-transform (current default) and transform-only (for users who already have the GLDAS files staged, e.g. on MAAP or in a shared archive).
Right now
dgldas2always does the full pipeline: login to EarthData, search CMR, download granules, verify sizes, then concatenate, fix time, convert accumulated depth to flux units, and delete the temp files.We want to split that in two so the transform half is reusable on its own. Concretely, add a
-skd / --skip-downloadflag (following the existing 3-letter short / descriptive long convention) that tells the tool the granules are already on disk — skip the login/search/download/verify steps, find the expected files in the output directory, run the same concat + time + unit-conversion logic, and don't delete the source files at the end since those now belong to the user.This unblocks two workflows from the same binary: download-and-transform (current default) and transform-only (for users who already have the GLDAS files staged, e.g. on MAAP or in a shared archive).