**This is an old revision of the document!**

# RTTOV

RTTOV stands for **R**adiative **T**ransfer for **TOV**S (top operational vertical sounding) and is a very fast radiative transfer model for nadir looking/scanning passive infrared and microwave satellite radiometers, spectrometers and interferometers. It is a FORTRAN-90 code for simulating satellite radiances, designed to be incorporated within users’ applications. First steps werde done in the early 90's at ECMWF (J.R. Eyre) for calculating fast radiative transfer with the TOVS (Tiros/top Operational Vertical Sounder) instruments HIRS, MSU and SSU. Subsequent development followed (R.W. Saunders, M. Matricardi). RTTOV is now an extensive well-validated model delivered with readily calculated coefficient data sets for the channels / frequency regions of all important earth observing satellite instruments/sensors and will be consequently extended for new ones.

The RTTOV approach is used operationally at several NWP centers for variational assimilation purposes. The actual version is RTTOV 9_1, released in March 2008. Computational speed is reported as, e.g., ~1 ms for 40 channel ATOVS on a desktop PC.

### Basics

For all the satellite sensors supported (see listing below), given an atmospheric profile of temperature, water vapour and optionally other atmospheric constituents (see below) together with satellite zenith angle and surface temperature, pressure and optionally surface emissivity, RTTOV will compute the top of atmosphere radiances in each of the channels of the sensor being simulated. Users can choose the channels to be simulated. Mathematically, in vector notation, given a state vector, **x**, which describes the atmospheric/surface state as a profile and surface variables and a radiance vector, **y**, for all the channels required to be simulated then:

**y** = *H*(**x**)

where *H* is the radiative transfer model, i.e. RTTOV (also referred to as the observation operator in data assimilation parlance). This is known as the ‘direct’ or ‘forward’ model.

In addition RTTOV also computes the Jacobian matrix **H** which gives the change in radiance δ**y** for a change in any element of the state vector δ**x** assuming a linear relationship about a given atmospheric state **x0**:

δ**y** = **H**(**x0**)δ**x**

The elements of **H** contain the partial derivatives **δyi**/**δxj** (**dyi**/**dxj**) where the subscript **i** refers to channel number and **j** to position in state vector. The Jacobian gives the top of atmosphere radiance change for each channel from each level in the profile given a unit perturbation at any level of the profile vectors or in any of the surface/cloud parameters. It shows clearly, for a given profile, which levels in the atmosphere are most sensitive to changes in temperature and variable gas concentrations for each channel. RTTOV_K (and its associated subroutines ending in K) compute the **H(x0)** matrix for each input profile.

It is not always necessary to store and access the full Jacobian matrix **H** and so the RTTOV package also has routines to only output the *tangent linear* values δ**y**, i.e. the change in top of atmosphere radiances, for a given change in atmospheric profile, δ**x**, about an initial atmospheric state **x0**. The tangent linear routines all have *TL* as an ending. Conversely the adjoint routines (ending in *AD*) compute the change in the gradient of any scalar quantity with respect to the atmospheric state, **x0**, given a change in the gradient of that quantity with respect to the radiances, **y**. These routines are normally used as part of the variational assimilation of radiances.

For users only interested in the direct or forward model for radiance simulations the TL/AD/K routines are not required.

**Atmospheric profiles and other essential items respected with RTTOV**

Temperature (mandatory), variable gas concentrations, cloud and surface properties (all referred to as the state vector, see above). The only mandatory variable gas is water vapour. Optionally ozone, carbon dioxide, nitrous oxide, methane and carbon monoxide can be variable with all other constituents assumed to be constant.

**Satellite platforms presently supported**

TIROS-N + NOAA.., DMSP, Meteosat, GOES, GMS, FY-2, TRMM, ERS, EOS, METOP, ENVISAT, MSG, FY-1, ADEOS, MTSAT, CORIOLIS

**Instruments presently supported**

HIRS, MSU, SSU, AMSU-A, AMSU-B, AVHRR, SSMI, VTPR, TMI, SSMIS, AIRS, HSB, MODIS, ATSR, AMSR, MVIRI, SEVIRI, GOES-Imager, GOES-Sounder, GMS/MTSAT imager, FY2-VISSR, FY1-MVISR, WINDSAT, SSM/T-2, IASI, (CriS, VIIRS)

*Documents partially used here (as at 6 May 2008):*

- users_guide_91_v1.4.pdf

### RTTOV fast RT model approach

RTTOV contains a fast model of the transmittances of the atmospheric gases that is generated from accurate line-by-line transmittances (GENLN2 / kCarta / LBLRTM, or Liebe-89 MPM, resp.) for a set of diverse atmospheric profiles (43L TIGR profile dataset / t101L 52 profile ERA-40 dataset) over the wave-number range of specific channels/instruments in question. The monochromatic transmittances are convolved with the appropriate spectral response functions and are used to compute channel-specific regression coefficients by use of a selected set of predictors. These regression coefficients can then be used by a fast transmittance model to compute transmittances given any other input profile. This parameterization of the transmittances makes the model computationally efficient and in principle should not add significantly to the errors generated by uncertainties in the spectroscopic data used by the line-by-line model. Assessing accuracy by means of comparing computations, e.g., for the instrument ATOVS see www.metoffice.gov.uk/research/nwp/satellite/rtm/rttov_accuracy.html (as at 6 May 2008). The plot shows the accuracy of the RTTOV-6 and earlier RTTOV-5 model calculations for NOAA-15 ATOVS by comparing them with calculations by an accurate line-by-line model for a diverse set of different atmospheric profiles. The accuracy of simulating very broad channels (e.g. SEVIRI channel 4 at 3.9 microns) is addressed to be poor, here significant bias can occur.

So, it is important to state that the primary goal of RTTOV is very fast computing RT for pre-calculated fixed channels of instruments, but with atmospheric profiles and additionals (surface and cloud items, zenith angles etc.) relatively free in choice.

— *R. Buell 2008/05/06 13:21*