surface pressures in TL mode

[Jepaung Sin]

I am upgrading my code from RTTOV v13 to v14 and have noted the significant changes in how pressure levels are handled between the two versions.
Specifically, I observed that the surface pressure variable has been removed from the rttov_near_surface structure in v14 (it was defined within rttov_skin in v13). My primary confusion stems from its handling in the Tangent-Linear (TL) mode.
In my existing code using the RTTOV v13, the surface pressure was explicitly assigned to profiles_tl%skin%p for TL mode calculations. Since this variable has been eliminated in v14, I need to know:
If I omit the assignment of surface pressure within profiles_tl when using RTTOV v14, will the TL mode results differ from those obtained using v13?
I have also compared the provided example tests: rttov_test_make_inc_mod.F90 (v14) and rttov_make_profile_inc.F90 (v13). The TL surface pressure variable is indeed absent in the v14 test. Therefore, are the outputs of these two version-specific example tests different?
Any clarification on the v14 design philosophy regarding the implicit handling of surface pressure would be greatly appreciated.

Thanks
Jepaung

[James Hocking]

Hi Jepaung,

Can I refer you to the RTTOV v14 documentation: this document (which is included in the RTTOV tarball in the docs/ dirtectory) goes into some detail about the differences between RTTOV v13 and v14.

I would also refer you to the RTTOV v14 user guide and to the RTTOV v14 Science and Validation Report, both also in the docs/ directory, and available on the RTTOV documentation page.

To summarise: the input profile now requires pressure-half levels to be defined which bound the layers for which all other input variables (temperature, gases, hydrometeors/aerosols) are provided. The bottom pressure half-level is the surface pressure in RTTOV v14, surface pressure is no longer independent of the pressure levels. If you require the surface pressure to be an active variable in the RTTOV v14 TL/AD/K then you can set the opts%interpolation%pressure_gradients option to true to make the p_half profile variable active. Then the surface pressure TL input is provided in profiles_tl(:)%p_half(nlevels), and the surface pressure Jacobian is output in profiles_k(:)%p_half(nlevels).

Best wishes,
James

• This reply was modified 3 days, 6 hours ago by James Hocking.

[Author]

Posts