- This topic has 7 replies, 2 voices, and was last updated 1 month, 1 week ago by
.
-
Posts
-
Hi,
While updating profiles from RTTOV v12/13 to v14, I would like to confirm a change in the input vertical discretization that does not seem to be stated explicitly in the v14 changelog.
In v14, it appears that:
– Atmospheric pressure is defined on half-levels (interfaces),
– Temperature, humidity, and gases are defined on full levels (layer centres),
– Hence, temperature/humidity arrays are one shorter than pressure,
– 2m p (hPa) has been removed.
Could you please confirm whether this understanding is correct?
Additionally, since datasets such as ERA5 and WRF provide state variables on full model levels, could you briefly advise on the recommended approach for constructing the required pressure half-levels for RTTOV v14 profiles?
Best regards,
AmirHi Amir,
The general approach when moving from RTTOV v13 to v14:
– the pressure levels in RTTOV v13 become the full-levels in v14
– temperature, water vapour, trace gas profiles therefore remain unchanged
– pressure half levels interleave the full levels, with the bottom half-level being set to the surface pressure.
For ERA5 (or any ECMWF or IFS-based NWP model), for example: the v14 profile structure maps directly onto the ERA5 vertical structure. ERA5/IFS defines pressure half levels based on the surface pressure and a set of published linear coefficients (for example see here for the current 137L). Pressure full-levels are computed as the mean of adjacent pressure half-levels (which is how RTTOV v14 computes them internally if full-levels are not specified). All other variables (T, q, trace gases) are specified on the pressure full levels.
Therefore, it is simple to pass ERA5 data into RTTOV v14. Note that you must use ERA5 data on the native IFS model levels, rather than ERA5 data on the fixed pressure levels.
I also recommend the “RTTOV v14 user interface changes” document if you have not seen it, included in the v14 tarball, which goes into some detail about the changes in v14 compared to v13.
Best wishes,
JamesHi James,
Thank you for your explanation, just to confirm one remaining point for my toolkit design:
For ERA5, am I correct that:
RTTOV v12/v13 should be driven by ERA5 pressure-level data, whereas
RTTOV v14 should be driven by ERA5 model-level data (hybrid half-levels + full-level state variables),
Best regards,
AmirHi Amir,
I would strongly recommend using ERA5 data on model levels for all radiative transfer calculations (including all versions of RTTOV). The fixed pressure levels are quite coarse and do not really go up high enough.
In general it is better to use NWP data on the native vertical grid of the atmospheric model. The only exception might be when simulating window channels and where speed is a critical factor, in which case using the pressure levels might be OK for some applications. But the changes in RTTOV v14 mean that you can only use the model levels because the surface always lies on the bottom pressure half-level.
Best wishes,
JamesHi James,
Thank you for your recommendation.
Just to confirm my understanding: ERA5 pressure half-levels (N+1) cannot be passed directly to RTTOV v12/v13, since those versions require all profile arrays to have the same length, whereas the half-level pressures have one more level than the state variables (T, q, gases). Therefore, when using ERA5 model-level data with RTTOV-12, the pressure levels (N+1) must first be collapsed to full levels (N) only.
Is this correct?
Best regards,
AmirHi Amir,
For RTTOV v13 and earlier with ERA5 data, you would pass in the temperature, water vapour, and trace gas arrays unmodified, and the RTTOV pressure level values are given by the ERA5 full-level pressures.
The only difference in RTTOV v14 for clear-sky simulations is that you pass in the ERA5 pressure half-levels instead. You can optionally choose to specify the ERA5 full-level pressures too, but it is not necessary.
Best wishes,
JamesHi James,
Following your recommendation, I am now using ERA5 model-level data (full 137 levels) instead of pressure-level data for my RTTOV application.
From ERA5 I retrieve:
– Temperature (t) on 137 model levels
– Specific humidity (q) on 137 model levels
– Surface pressure (sp) from levtype = sfcI would like to clarify whether:
1) RTTOV internally supports ERA5 L137 coordinates and contains the corresponding coefficients, so that providing T, Q, and sp is sufficient
or
2) I should explicitly compute the pressure profile using the ERA5 coefficients before passing the profiles to RTTOV.
Best regards,
Amir
- You must be logged in to reply to this topic.