2.9.3 Auxiliary Data Sets for Level 2 processing
AATSR only uses 2 auxiliary files in its
L2 processing. Both L2 auxiliary
files are maintained by DEFRA.
Clicking on the links provided will take
the reader to the Data Formats
section, which contains a description
of each field within each data set.
AATSR Level 2 Processor
Configuration File (ATS_PC2_AX) 6.5.7.

Use/Contents
 A single data set containing
miscellaneous configuration
parameters including brightness
temperature thresholds for the AST product
and parameters defining the
geometry of the averaged
product.

Source
 Based on requirements of AATSR
processor design.

Update
 To be optimised as part of AATSR L2
algorithm verification.
Unlikely to change beyond this
point.
SST Retrieval Coefficients (ATS_SST_AX) 6.5.8.

Use/Contents
 Used to retrieve full resolution
(see section 2.7.1.2. ) and spatially
averaged (see section 2.7.1.3. ) SST from L1B BTs.
The file contains 3 data
sets:

 Acrosstrack band
mapping LUT
(defines the relationship between
pixel acrosstrack coordinate and
acrosstrack
band number).
 SST retrieval
coefficients (for the
1 km gridded SST retrieval).
 Averaged SST retrieval
coefficients
(for the averaged SST
retrieval).

Source
 The coefficients represent a linear
regression of SST on
brightness temperature. Regression
is based on model brightness
temperatures derived
from an ensemble of atmospheric profiles
using a Radiative
Transfer (RT) model. (A)ATSR
coefficients are based on the model RADGEN
from the Rutherford Appleton
Laboratory. See below for
further details.

Update
 Infrequent. As these coefficients
are derived on the basis of
RT modelling and are not subject
to empirical adjustment, any changes to
this file will be the result
of a new RT calculation. Such an
exercise might be undertaken to
apply new physical
parameters or modelling as the
result of detailed product
validation.
LST Retrieval Coefficients (ATS_LST_AX)

Use/Contents:
 Used to retrieve full resolution
LST from L1b BTs. The file
contains 6 data sets
comprising a short GADS of general
parameters, four LUT GADS of
vegetation class, vegetation fraction,
climatology and topographic
variance data, and one LST
Retrieval Coefficient LUT GADS:
 General Parameters for LST Retrieval
 Index Data for LST Retrieval
 Vegetation Fraction for LST Retrieval
 Climatology Records for LST Retrieval
 Topographic variance flag for LST retrieval
 LST Retrieval Coefficients

Source:
 See Land
Surface Temperature
Measurement from
Space: AATSR Algorithm
Theoretical Basis Document,
Jan 2002 1.3. .

Update:
 Infrequent.
Generation of Retrieval Coefficient Sets
for (A)ATSR
The original ATSR coefficients were based
on the work of Zavody et al, 1995
Ref. [1.18 ]
,
using a scheme with 10 acrosstrack
bands. However, they have
recently been updated to take account of new
developments in coefficient
generation, including an improved stratospheric
aerosol treatment developed by Merchant et al, 1999
Ref. [1.14 ]
,
which uses 38 bands to reduce
discontinuities at the band edges.
Coefficient derivation involves 3 steps:
 Generate a statistically
representative ensemble of
atmospheric profiles;
 Run radiative transfer model
(RADGEN) to derive
brightness temperatures for each
profile;
 Derive regression coefficients of
surface temperature vs.
brightness temperature.
The derivation is therefore dependent on
a) the ensemble of atmospheric
profiles used, b) the version of
RADGEN used, and c) the auxiliary data
used in this process.
The recent update to RADGEN was designed
to address the following
requirements:

Ability to read atmospheric
profiles derived
from ECMWF data
The original ATSR coefficients
were based on a set
of 158 radiosonde profiles
provided by the UK Met Office. However,
Merchant et al
introduced a new set of profiles based
on ECMWF reanalysis data.
Therefore the code
had to be modified to accept these new
profiles.

Enhanced tropospheric aerosol
model
The modifications to the
tropospheric aerosol
subroutine included the
optional
implementation of a 'hybrid'
scattering model, the
addition of asymmetry coefficients (from
Shettle and Fenn,
1979) and the removal of
redundant scaling.
Modifications to the main programme
included the optional inclusion of
'constant number
density' scaling as
alternative to
'constant visibility' scaling,
and the
introduction of exponential aerosol
height distribution,
in place of the original
distribution, which was uniform below 1
km height.

Revised air mass and
emissivity
definitions to accommodate revised
acrosstrack banding scheme
The original acrosstrack
banding scheme used
interpolation between 5
standard air mass
values computed by RADGEN. The new
scheme requires
air mass values at swath centre
and swath edge in each view.

Inclusion of parameterised
wind speed effects
in the forward view.
This scheme modifies surface
reflectivity and
emissivity to account for
roughness effects in
the forward view (Watts et al 1996
Ref. [1.17 ]
),
depending on wind
speed.

Improved portability of the
code.
LST Retrieval Coefficients
The auxiliary file designated ATS_LST_AX
contains the lookup tables and
retrieval coefficients used by the
LST algorithm. It includes the data sets
listed in table.

Table 2.34 Data sets included in the LST auxiliary file ATS_LST_AX

Description of data set 
Contents 
General parameters for
LST retrieval 
3 parameters d,
m, NCLASS

Surface type
classification 
720 x 360 array 
Vegetation fraction 
720 x 360 x 12 array 
Precipitable water 
720 x 360 x 12 array 
Topographic variance flag 
720 x 360 array 
LST retrieval
coefficients 
16 x 2 (vegetation /
soil) x 2
(night/day) coefficient sets 
The auxiliary files define the surface
class and vegetation fraction with
a resolution of 0.5 degrees. Thus
for each cell of 0.5 degrees in latitude
by 0.5 degrees in longitude a
surface class and vegetation
fraction are defined, that are taken to apply to
the whole cell. The arrays are
indexed using the latitude and
longitude indices lat_index and lon_index
defined in Section 2.7.1.2.2.1.1. , and the array
values are ordered so that the
longitude index varies most rapidly.
