Issues with specifying specific datetimes for weather model initialization runs in the CLI

[nwagenbrenner]

A few users have emailed about this issue. There are two options: 1. set the forecast_time or 2. set start_year, end_year, etc. I think 2 works almost accidentally after it was added for the point initialization option. We need to check the functionality of each of these and ensure they are behaving properly for weather model initializations. There is also confusion surrounding forecast_duration and when it needs to be set.

For example, we just got this report from a user:

I am trying to run WindNinja using the CLI (with a configuration file) at a specific forecast time, rather than just when the latest models are initialized, and I am running into some issues. Specifically, I am trying to use the NAM-NEST CONUS 3km model for the 13Z forecast, initialized from the 12Z NAM-NEST run.

When I set the forecast_time variable in the configuration file to 20250426T130000, I get the following error:

Exception caught: Could not find forecast associated with requested time step
Exception badForecastFile caught: Could not find forecast associated with requested time step
There was a problem downloading the forecast file or it had bad data.

I also tried using the start_* and end_* time variables (e.g., start_year, start_month, end_day, etc.), but couldn not get WindNinja to run at the desired time. For example, setting the start time to 2025-05-28 01:00 and the end time to 2025-05-28 14:00 (entered as integers in the config), WindNinja instead ran from 2025-05-28 12:00 to 2025-05-28 19:00.

In both cases, I set forecast_duration to 2 (also changed it a few times), but it didn’t seem to have any effect.

We should update our CLI documentation and add a tutorial example here for how to properly use this functionality.