The computation of radiances is initiated and controlled by the HOP routine. Thicknesses, PWC and TPW are also computed in HOP and SCAT data too are processed in HOP. The general structure of HOP has been detailed in Subsection 2.5.2.

6.4.1 Radiances

The routine HOP interpolates the model profiles of temperature, humidity and ozone (

and

) to the 43 RT levels (

and

) and calls the interface RADTR to the RT code RTTOV. The standard routines PPT (Section 5.6 of Chapter 5 `Conventional observational constraints' ) and PPQ (Section 5.5) are used to carry out the vertical interpolation, and they are called through the PPOBSA interface, as usual. Various radiance preparations have been gathered in the new routine HRADP. In HRADP The model's pressure at the surface height of the observation location (given in the report) is calculated, using PPPMER. For the purpose of radiance calculations

and

. These quantities represent a very shallow layer of air near the surface and contribute little to the calculated radiances-it was not considered necessary to use PPT2M and PPRH2M (Section 5.7) in this context. In order to make the radiance cost function continuous in

it was necessary to ensure that

and

approach

and

as the pressure on any of the RT levels approches

. This is done in a section of HRADP. More details on the radiative transfer code RTTOV can be found in Eyre (1991), updated by Saunders and Matricardi (1998), (available on-line ps-file).

Some of the radiance channels are highly sensitive to the surface skin temperature, which is also not part of the variational control variable when RTOVS data are used. It was found that the best results were obtained by replacing the model's

with those retrieved by 1D-Var. The 1D-Var retrieval is carried out in a call to ADVAR from HRETR, called from TASKOB in the screening configuration only.

In the case of 1C, or `raw' radiance data, as used since May 1999 (McNally et al. 1999) 1D-Var is no longer required. The radiance processing in HOP is similar for both 1C and RTOVS radiances, with the exception that surface skin temperature is retrieved by 4D-Var at each 1C-field of view, if the switch LTOVSCV is on (default is on).

In HOP the observation array is searched for radiance data. The compressed ODB (after screening) contains only those data to be used by the analysis. A list of existing channel numbers for each report is constructed. Model radiances for exactly those channels are then requested from the RT-code, via the interface RADTR. The routine RADTR checks that the input model profile is within the valid range of the transmittance regression. It packets the profiles into chunks of work of the appropriate maximum size for the RT-code (currently 65). The RT packet size has been communicated to IFS in the call to RTSETUP. The output is radiances for the channels requested.

The tangent linear HOPTL and the adjoint HOPAD follow the same pattern as HOP. In both the TL and the adjoint

and

have to be recomputed before the actual tangent linear and adjoint computations can start. The pointers to the radiance data in observation array are obtained just as it was done in the direct code. The input gradient to the adjoint is obtained as explained in Subsection 2.5.2.

6.4.2 Thicknesses

The pressures of layer bounds (top T, and bottom B) are found (in HOP) by scanning the observation array for thickness data. The geopotential for the top and the bottom of the layer are computed, using PPGEOP (Section 5.3), and the thickness is given by the difference

6.4.3 Precipitable water from SATEM and SSM/I

As for thicknesses, the pressures of layer bounds are found by scanning the observation array for TOVS PWC data. For SSMI TPW, the top pressure is set to the top of the model and the lower pressure bound is

. The PWC for the top and the bottom of the layer are computed, using PPPWC (Section 5.5), and the layer PWC is given by the difference

6.4.4 Scatterometer winds

In HOP, the observation array is scanned for SCAT data. Normally two ambiguous pairs of

-component and

-component observations are found at each SCAT location-with directions approximately 180 degrees apart. In 3D/4D-Var both winds are used and the ambiguity removal takes place implicitly through the special SCAT cost-function, Eq. (2.8), in HJO (Stoffelen and Anderson, 1997 ; Gaffard et al. 1997). If however LQSCATT=.true. (namjo), the normal quadratic

will b e used. In this case only the SCAT wind nearest the high resolution background will be used (which is determined in a section of HOP).

As PPUV10M (Section 5.7) is used also for SCAT data (since cy18r6), the observation operator is exactly the same as for SYNOP. SHIP and DRIBU winds. The

(surface roughness) comes from the coupled wave model. The simpler logarithmic wind law can be used optionally under the switch LSCASUR=.F. in NAMOBS (true by default).

In the adjoint (SURFACAD) there is a separate section of HOP for the calculation of the

Next Section
Previous Section

20.03.2002

DATA ASSIMILATION

IFS documentation Front Page

Table of contents

6.4 Observation operators

6.4.1 Radiances

6.4.2 Thicknesses

6.4.3 Precipitable water from SATEM and SSM/I

6.4.4 Scatterometer winds