Use default option for config_extrap_airtemp in namelist.init_atmosphere

[ibanos90]

Description

In PR#418, the capability to convert ERA5 model-level analyses to the MPAS initial conditions format was added. However, when using the current linear option in config/mpas/initic/namelist.init_atmosphere, unrealistic values of potential temperature (θ) are produced (min=17K), leading to model forecast failure. In contrast, using lapse-rate (the default in the User’s Guide) or constant both yield physically realistic values. While linear performs well for GFS analyses and ERA5 pressure-level data, it does not properly handle ERA5 model-level data. Therefore, in this PR, the default option is changed to lapse-rate as it works across all external data sources. Additionally, a new scenario is introduced to generate forecasts using ERA5 model-level data.

Issue closed

None

Tests completed

Tier 1:

• 3dvar_OIE120km_ColdStart
• 3dvar_O30kmIE60km_ColdStart
• ForecastFromGFSAnalysesMPT

Tier 2:

• ForecastFromGFSAnalyses
• ForecastFromERAmlAnalyses

[jim-p-w]

Is this going to impact all of the scenarios which run a forecast?

[ibanos90]

That's a good question @jim-p-w! Scenarios that run mpas_init_atmosphere will now be using this option. My test ForecastFromGFSAnalyses confirms it runs with no issue. I will run other tests to confirm any differences.

[ibanos90]

Just confirming the scenarios that run mpas_init_atmosphere complete successfully. As can be expected, some differences are found in theta field at level 0 in the init file due to the change in the method used for the extrapolation of air temperature above/below first-guess levels.

[jim-p-w]

Just confirming the scenarios that run mpas_init_atmosphere complete successfully. As can be expected, some differences are found in theta field at level 0 in the init file due to the change in the method used for the extrapolation of air temperature above/below first-guess levels.

@ibanos90 Thanks for checking this.