S F SAF/NWC/INM/SCI/SUM/08 Software Users Manual

advertisement
EUMETSAT Satellite Application
Facility to NoWCasting & Very
Short Range Forecasting
Software Users Manual
of the SAFNWC/MSG:
Scientific part for the PGE08
SAF/NWC/INM/SCI/SUM/08
Issue: 1.0 Date: 17 January 2002
Page: i/i
Software Users Manual
of the SAFNWC / MSG:
Scientific part for the PGE08
SAF/NWC/INM/SCI/SUM/08, Issue 1, Rev. 0
17 January 2002
Prepared by INM
EUMETSAT Satellite Application
Facility to NoWCasting & Very
Short Range Forecasting
Software Users Manual
of the SAFNWC/MSG:
Scientific part for the PGE08
SAF/NWC/INM/SCI/SUM/08
Issue: 1.0 Date: 17 January 2002
Page: i/iii
REPORT SIGNATURE TABLE
Function
Prepared by
Reviewed by
Name
Signature
Date
M. Velazquez (INM)
17 January 2002
M.A. Martinez (INM)
M. Manso (INM)
17 January 2002
L.F. Lopez-Cotin
Authorised by
SAFNWC/MSG Project
Manager
17 January 2002
Software Users Manual
of the SAFNWC/MSG:
Scientific part for the PGE08
EUMETSAT Satellite Application
Facility to NoWCasting & Very
Short Range Forecasting
SAF/NWC/INM/SCI/SUM/08
Issue: 1.0 Date: 17 January 2002
Page: ii/iii
DOCUMENT CHANGE RECORD
Version
1.0
Date
17 January 2002
Pages
13
CHANGE(S)
First published version
SAF/NWC/INM/SCI/SUM/08 :
EUMETSAT Satellite Application
Facility to NoWCasting & Very
Short Range Forecasting
Software Users Manual
of the SAFNWC/MSG:
Scientific part for the PGE08
SAF/NWC/INM/SCI/SUM/08
Issue: 1.0 Date: 17 January 2002
Page: iii/iii
Table of contents
1.
1.1
1.2
1.3
2.
INTRODUCTION ...................................................................................................................1
SCOPE OF THE DOCUMENT .......................................................................................................1
SCOPE OF OTHER DOCUMENTS.................................................................................................1
SOFTWARE VERSION IDENTIFICATION .....................................................................................1
DESCRIPTION OF SAI PRODUCT ....................................................................................2
2.1
GOAL OF SAI PRODUCT ...........................................................................................................2
2.2
SAI ALGORITHM DETAILED DESCRIPTION ...............................................................................2
2.2.1.
Algorithm outline...........................................................................................................2
Spatial smoothing of SEVIRI radiances ........................................................................................3
Normalitation. ...............................................................................................................................4
Satellite zenith angle correction....................................................................................................4
SAI Neural Network.......................................................................................................................4
Quality control ..............................................................................................................................5
De-normalitation. ..........................................................................................................................5
LI Spatial Smoothing .....................................................................................................................6
2.3
LIST OF INPUT FOR SAI ............................................................................................................6
2.4
DESCRIPTION OF SAI OUTPUT .................................................................................................6
Description of the conversion from Celsius degrees to brightness values ....................................6
Description of SAI output format ..................................................................................................7
2.5
EXAMPLE OF SAI VISUALISATION ...........................................................................................8
2.6
IMPLEMENTATION OF SAI.......................................................................................................9
2.6.1
The SAI installation step: ............................................................................................10
2.6.2
The SAI preparation step: ...........................................................................................10
2.6.3
The SAI execution step: ...............................................................................................10
ANNEX 1: ANCILLARY DATA.................................................................................................10
ANNEX 2. REFERENCES ...........................................................................................................12
ANNEX 3. ACRONYMS ..............................................................................................................13
EUMETSAT Satellite Application
Facility to NoWCasting & Very
Short Range Forecasting
Software Users Manual
of the SAFNWC/MSG:
Scientific part for the PGE08
SAF/NWC/INM/SCI/SUM/08
Issue: 1.0 Date: 17 January 2002
Page: 1/13
1. INTRODUCTION
1.1
SCOPE OF THE DOCUMENT
This document contains a scientific description of the algorithms used for PGE08 implementation,
including the expected input data, the output product and the method adopted for a better
understanding of the software.
Due to the lack of real SEVIRI data availability, a most detailed optimal estimation and realistic
testing of the resulting models constitute the baseline to the Initial Operations Phase (IOP).
1.2
SCOPE OF OTHER DOCUMENTS
The software part of the Software User Manual (SUM) describes the individual processes, which
make up the system with a description of the software and operational environment associated
Details of input and output data format of this product are described in the Interface Control
Documents ICD1 (for the External and Internal Interfaces of the SAFNWC/MSG) and ICD3
(MSG Output Product Format Definition).
The general architecture of the software is described in the Architectural Design Document
(ADD) (interface with the SAFNWC software, architecture of the PGE). The document is denoted
as SAF/NWC/INM/SW/AD/4, 1.01.
The product generator elements are described in the Software Version Description (SVD). The
document is denoted as SAF/NWC/INM/SW/SVD/3, 1.01.
The evolution of the scientific developments have been presented in the SAFNWC Scientific
Report from MTR on PGE08 Stability Analysis Imagery, together with two verbal presentations
in the EUMETSAT Meteorological Satellite Data User's Conferences (Martınez et al, 1999 and
Martınez & Velazquez, 2001).
1.3
SOFTWARE VERSION IDENTIFICATION
This document describes the algorithms implemented in the release v.1.0.1 of SAFNWC/MSG for
PGE08
EUMETSAT Satellite Application
Facility to NoWCasting & Very
Short Range Forecasting
Software Users Manual
of the SAFNWC/MSG:
Scientific part for the PGE08
SAF/NWC/INM/SCI/SUM/08
Issue: 1.0 Date: 17 January 2002
Page: 2/ 13
2. DESCRIPTION OF SAI PRODUCT
2.1
GOAL OF SAI PRODUCT
The software to extract Stability Analysis Imagery (SAI) from MSG SEVIRI Imagery over MSG
N has been designed within the EUMETSAT SAF for support to Nowcasting and Very Short
Range Forecasting (hereafter referred as SAFNWC).
The Stability Analysis Imagery (SAI), developed within the SAF NWC context, aims to support
nowcasting applications. The SAI allows obtain one measure of the stability in cloud free areas.
The central aim of the SAI is therefore to provide estimations of the atmospheric instability in
cloud-free areas. Among all potential indexes the Lifted Index (LI) has been implemented and
codified. The product output provides quantitative information on the LI. A conversion from
Celsius degrees to grey levels or digital count is necessary for writing LI output in only 7 bits.
The SAI is designed to derive estimations of the Lifted Index using seven MSG IR multispectral
radiances (6.2, 7.3, 8.7, 9.7, 12.0 & 13.4µm). The PGE01 (Clod Mask-CMa) product identifies the
first cloud free for which the SAI is later obtained.
SAI's main algorithm has been devised as a neural network with a resulting topology. It is possible
that during the tuning phase, other topology with different number of neurones in the hidden layer
could present better performances when the real data will be available. Therefore, the software has
been designed to change the topology and weights only changing the name of the new neural
networks in the configuration file. When the PGE08 is started, all the names of the files with the
neural networks are read from the configuration file and the topology and weights are allocated
and loaded.
The software has been structured in order to introduce easily other stability indexes. Other
stability index can be obtained, just by changing the sea and land neural network files, the Look
Up Table (LUT) using in the denormalization process and the thresholds used in the conversion to
grey levels or counts.
2.2
SAI ALGORITHM DETAILED DESCRIPTION
2.2.1. Algorithm outline
The implementation and validation of SAI algorithms are based mainly on synthetic cases and
sensitivity studies, as real satellite data availability (NOAA and GOES) did not present any
channels comparable to MSG's 8.7µm during the development phase.
The detailed description of these developments has been shown in different scientific documents:
Scientific Report from MTR on PGE08 Stability Analysis Imagery, together with two verbal
presentations in the EUMETSAT Meteorological Satellite Data User's Conferences (Martınez et
al, 1999 and Martınez & Velazquez, 2001).
The CMa output is used to identify clear air pixels. Only if the pixel is labelled as air clear and the
satellite zenith angle of this pixel lies within the configurable threshold the pre-processing and
processing is calculated in this pixel.
The whole process includes:
1. Pre-processing:
EUMETSAT Satellite Application
Facility to NoWCasting & Very
Short Range Forecasting
Software Users Manual
of the SAFNWC/MSG:
Scientific part for the PGE08
SAF/NWC/INM/SCI/SUM/08
Issue: 1.0 Date: 17 January 2002
Page: 3/ 13
•
Smoothing of SEVIRI radiances (optional) performing a median o mean/average
(configurable in the model configuration file) over a window centred at the processing
pixel. Boundaries of the region are not considered.
•
Normalisation of SEVIRI radiances.
•
Correction of SEVIRI radiances to 45¼MSG zenith angle. This step will use a
dedicated neural network, whose inputs are the SEVIRI radiances and the satellite
zenith angle.
2. Processing:
The calculus of the lifted index will use dedicated Neural Networks (NN). The NNs will
be fed with normalized corrected IR SEVIRI radiance values (WV6.2, WV6.3, IR8.7,
IR9.7, IR10.8, IR12.0 & IR13.4µm), and will provide as output the normalized lifted
index.
For land pixels the normalised altitude is also needed as input for the neural network.
3. Quality control:
To check the product quality, different flags have been added to the output products.
•
Clear air flag: Derived from the cloud mask product PGE01 CMa.
•
Out of radiance flag: Indicates how many bands are out of range for product
confidence purposes.
•
Number of cloudy pixels in the processing window: indicating the confidence of the
smooth pre-processing step.
4. Post-processing:
The post-processing step will include:
•
Denormalisation of the lifted index to Celsius degrees. A LUT is used in this step.
•
Smoothing of the LI when it is required in the model configuration file, given the
proper parameters such as method to use (mean/average or median), length of the
processing box and minimum number of cloud free pixels in the window to perform
the smoothing. The process consists of making a mean/average or a median of the
parameter over a window centred at the pixel.
•
Conversion of LI form Celsius degrees to grey levels using coefficients specified in
the model configuration file.
•
Over cloudy pixel, a configurable IR channel is shown.
Spatial smoothing of SEVIRI radiances
The seven IR spectral bands observed by the MSG used as input in this SAF product, will be
smoothed regardless of the configuration files having or not a line with the prefixed spatial
smoothing.
For each channel it is possible to introduce an independent line in the configuration file. This line
specifies the width of the window to use in the smoothing and the applied method.
Two different smoothing methods have been codified: the first one is the average and the second
one the median (See ADD).
Only the pixels in the predefined window labelled as clear air and with satellite zenith angles
within the configurable threshold will be taken in account during the smoothing process. The
number of the cloudy pixels in the biggest window will be used as a quality control flag.
EUMETSAT Satellite Application
Facility to NoWCasting & Very
Short Range Forecasting
Software Users Manual
of the SAFNWC/MSG:
Scientific part for the PGE08
SAF/NWC/INM/SCI/SUM/08
Issue: 1.0 Date: 17 January 2002
Page: 4/ 13
Normalitation.
The NN used for obtaining corrected satellite zenith radiances needs normalised inputs. A
minimum-maximum method for normalization of the radiances has been used and the
correspondent coefficients appear in the configuration file.
Satellite zenith angle correction
Dependence of the atmospheric attenuation on the satellite zenith angle has been taken into
account in practically all algorithms that have been implemented to obtain derived products from
satellite data.
In deriving the general algorithm for SAI and also for LPW, the construction of the algorithm for
a constant satellite zenith angle (45¼
) was found the most convenient.
The dependence of the atmospheric attenuation is corrected using a neural network, the name of
the file with the topology and weights appear in the configuration file. This neural network has 8
inputs (13.4, 12.0, 10.8, 9.7, 8.7, 7.3, 6.2 µm normalized radiances and the normalized satellite
zenith angle) and 7 outputs (13.4, 12.0, 10.8, 9.7, 8.7, 7.3, 6.2 µm normalized corrected to 45¼
radiances) and only has one hidden layer. The name of the file with this neural network is in the
configuration file.
The angular dependence has been determined training a Multilayer Perceptron (MLP) neural
network with radiative transfer simulations for 17 different satellite angles versus the radiative
transfer simulation at 45¼
. The values of MSG IR radiances were determined from the TIGR
database (with 2311 profiles) (Chedin et al, 1985). The SYNSATRAD radiative transfer model
(Tjemkes & Schmetz, 1998) was used for each profile and each satellite zenith angle to
determinate the MSG infrared radiance dataset. For further details see Martınez and Velazquez
(2001).
SAI Neural Network.
Neural network-based algorithms
The relationship between the simulated MSG radiance values and the calculated SAI from the
profiles has been parameterised under the form of a Multilayer Preceptor (MLP).
Many statistical methods have been used in similar works to infer links from output data
(response) to input data (predictor). The most widespread technique is linear regression, but the
linear hypothesis on which these techniques are based limit the accuracy of the models. While,
other algorithms rely on nonlinear statistical methods, for instance MLP. In the last years, MLP
has been increasingly used for meteorological-related problems (e.g. Chevallier et al. 1998)
After the zenith satellite angle correction has been applied the software take into account two
types of surfaces: land and sea, and two different MLPs have been implemented for each type.
Sea pixels
The sea neural network has 7 inputs (13.4, 12.0, 10.8, 9.7, 8.7, 7.3, 6.2 µm normalized radiance)
and 1 output (Normalized LI). In the current version only one hidden layer have been introduced.
EUMETSAT Satellite Application
Facility to NoWCasting & Very
Short Range Forecasting
Software Users Manual
of the SAFNWC/MSG:
Scientific part for the PGE08
SAF/NWC/INM/SCI/SUM/08
Issue: 1.0 Date: 17 January 2002
Page: 5/ 13
One or more "hidden" layers of the neurones can be introduced between the input layer and the
output layer, however significant improvements have not been found using more than one
"hidden" layer.
The name of the file with this neural network is in the configuration file.
Land pixels
The difference with the sea neural network is another input to be added. The land neural network
has 8 inputs (13.4, 12.0, 10.8, 9.7, 8.7, 7.3, 6.2 µm normalized radiance and normalized altitude)
and 1 output (Normalized lifted index). In the current version only one hidden layer have been
introduced.
The land neural network has been trained using normalized pressure values as input. So, a
conversion from pressure to altitude is necessary. The implemented conversion in this first version
is the following:
•
Standard atmosphere is assumed.
•
1013 hPa in pressure is equivalent to 0 metres.
•
725 hPa in pressure is equivalent to 2700 metres.
During the IOP, different conversions will be tested and applied depending of the type of
atmosphere.
The name of the file with this neural network is in the configuration file.
Quality control
Quality flags are attached to SAI.
The first flag allows the identification of cloud-free and cloudy pixels.
The number of cloudy pixels in the biggest window used to perform the spatial smoothing
preprocess is used as a level of confidence.
The number of bands whose radiances are out of a range specified in the model
configuration file is a quality flag that should allow to identify areas where the algorithm
may be not accurate enough.
De-normalitation.
The main aim is to separate between stable and unstable cases. As the number of stable cases is
greater in the databases, and the range of values is also wider in the stable cases, the usual
minimum-maximum normalization is not adequate. Among all the possibilities to avoid this effect
we decide to introduce a hyperbolic tangent to normalize the LI values in the pattern files used in
the training. The two parameters of the hyperbolic tangent were adjusted to get the best resolution
in the transition from stable to unstable and the stable and unstable cases have the same range (0,5
for each one).
Because the neural networks used to obtain SAI have been trained using a hyperbolic tangent to
normalize the pattern (LI), the outputs of these NNs are values of LI hyperbolic tangent. To invert
this hyperbolic tangent and to obtain LI values in Celsius degrees a LUT is used.
EUMETSAT Satellite Application
Facility to NoWCasting & Very
Short Range Forecasting
SAF/NWC/INM/SCI/SUM/08
Software Users Manual
of the SAFNWC/MSG:
Scientific part for the PGE08
Issue: 1.0 Date: 17 January 2002
Page: 6/ 13
LI Spatial Smoothing
As it is not possible to decide which smoothing is most adequate (preprocess, postprocess or both)
until real data are available, another configurable spatial smoothing process (similar to the
smoothing radiance) has been implemented on LI values. It is necessary only to introduce a line in
the configuration file to activate this smoothing process of LI values. This line specifies the
window width to make the smoothing and the method to apply. The methods are described in the
ADD.
2.3
•
LIST OF INPUT FOR SAI
MSG radiance values:
The following SEVIRI radiance values are needed at full IR spatial resolution:
R13.4 µm
R12.0 µm
R10.8µm
R9.7µm
R8.7µm
R7.3µm
R6.2µm
Mandatory
Mandatory
Mandatory
Mandatory
Mandatory
Mandatory
Mandatory
The SAI system processing checks the availability of SEVIRI channels for each pixel; no results
are produced for pixels where at least one channel is missing.
The SEVIRI channels are input by the user in HRIT format, and extracted on the processed region
by SAFNWC software package.
•
CMa
The CMa is mandatory. After the CMa process is executed, the SAI system processing is
computed.
•
Satellite zenith angles associated to selected region
This information is mandatory. It is computed by the SAI software itself, using a function
available in the NWCLIB.
•
Ancillary data sets:
The following ancillary data, remapped onto satellite images, are mandatory:
o Land/sea mask
o Elevation mask
These ancillary data are available in the SAFNWC software package on MSG full disk in the
default satellite projection at full IR resolution. They are extracted on the processed region by the
SAI software itself.
Files: Region Configuration Files, Neuronal Network Configuration Files, Denormalisation LUT
Files, PGE08 Model Configuration Files, are available in the SAFNWC software package, and is
needed by the SAI software.
2.4
DESCRIPTION OF SAI OUTPUT
Description of the conversion from Celsius degrees to brightness values
The main aim of SAI (LI) is to provide information on the stability of the troposphere. The unit
of LI is Celsius degrees so to allow resolution greater than 1¼
C in the LI values it is necessary to
SAF/NWC/INM/SCI/SUM/08
Software Users Manual
of the SAFNWC/MSG:
Scientific part for the PGE08
EUMETSAT Satellite Application
Facility to NoWCasting & Very
Short Range Forecasting
Issue: 1.0 Date: 17 January 2002
Page: 7/ 13
use float values or to convert from Celsius degrees to counts. The conversion to count is more
efficient because the final output size is smaller.
The LI output is written in 7bits (1:128), and in order to increase resolution in the unstable range
two different linear transformations have been used.
Find hereunder the graph that shows the conversion of SAI.
The coefficient minimun/maximum values if SAI and the offsets are all configurable in the model
configuration file.
The allowing coefficients in the model configuration files correspond to the graph:
SAI_MIN
-20
SAI_CUT
10
SAI_MAX
40
BRIT_SAI_MIN
8
BRIT_SAI_RANGE1
90
BRIT_SAI_RANGE2
30
128
B
R
I
G
H
T
N
E
S
S
98
68
38
8
40
30
20
10
0
-10
-20
Lifted Index ( C)
Figure 1: Celsius degree Lifted Index versus brightness (or counts) Lifted Index
Each linear transform is calculated as follows:
• First linear transform: (with higher resolution)
SAI ∈ [SAI_min , SAI_cut], in the figure between [-20¼
C,10¼
C]
SAI (brightness )1 = BRIT _ SAI _ min + BRIT _ SAI _ range 2 + BRIT _ SAI _ range1
( SAI − SAI _ cut )
( SAI _ min − SAI _ cut )
• Second linear transform: (with lower resolution)
SAI ∈ (SAI_cut , SAI_max], in the figure between (10¼
C,40¼
C]
SAI (brightness) 2 = BRIT _ SAI _ min + BRIT _ SAI _ range2
(SAI − SAI _ max)
( SAI _ cut − SAI _ max)
Description of SAI output format
The format will be SAFNWC, the content of the SAI is the following in modified HRIT format:
•
7 bits to show the product
EUMETSAT Satellite Application
Facility to NoWCasting & Very
Short Range Forecasting
Software Users Manual
of the SAFNWC/MSG:
Scientific part for the PGE08
Lifted index
air clear pixels.
Configurable
IR channel
cloudy pixels
•
0
1
•
SAF/NWC/INM/SCI/SUM/08
Issue: 1.0 Date: 17 January 2002
Page: 8/ 13
1 to describe clear sky flag:
Air clear pixel. SAI is calculated
Cloudy pixel. IR value is given
2 bits for Number of out of range radiances
Number of bands whose radiances are out of a range specified in the configuration file. The
ranges of the radiances for the seven different channels are specified in the configuration file.
Possible values are:
0
All bands within the range
1
1 band out of range
2
[2, 4] bands out of range
3
[5, 7] bands out of range
•
2 bits to label number of cloudy pixels in the biggest window
This label presents information about the neighbours of the processed pixel depending on the
cloudy pixels in the biggest window used to perform the spatial pre-processing of bands. The
window width is a configurable parameter (normally: none, 3x3 or 5x5 pixels).
Possible values are:
0
1
2
3
[0, 1/8] cloudy pixels
Window 3x3:
[0, 1] pixel
Window 5x5:
[0, 3] pixels
(1/8, é] cloudy pixels
Window 3x3:
[2] pixels
Window 5x5:
[4, 6] pixels
(1/4, 1/2] cloudy pixels
Window 3x3
[3, 4] pixels
Window 5x5
[7, 12] pixels
(1/2, 1] cloudy pixels
2.5
Window 3x3
[5, 8] pixels
Window 5x5
[13, 24] pixels
EXAMPLE OF SAI VISUALISATION
The LI has been trained with TIGR and tested with Numerical Weather Prediction (NWP) output
products, as no image similar to MSG's spatial resolution was available during the development
phase.
A similar visualisation model as the one used by CIMSS with LI from GOES will be possible.
(http://cimss.ssec.wisc.edu/goes/realtime/grtmain.html#usli).
EUMETSAT Satellite Application
Facility to NoWCasting & Very
Short Range Forecasting
Software Users Manual
of the SAFNWC/MSG:
Scientific part for the PGE08
SAF/NWC/INM/SCI/SUM/08
Issue: 1.0 Date: 17 January 2002
Page: 9/ 13
A new working environment at INM is beginning to use MODIS data as a reference. During
January 2002, SSEC has allowed free access through ADDE McIDAS servers for MODIS data to
interested users. This visualisation has been the very first example derived in such an
environment, for data received on January 9th. at 15:58Z. Quality of the final product is expected
to improve as the new environment evolves.
Figure2: This LI image has been derived applying the NNs obtained for MSG using as input similar MODIS
channels at full resolution data, and remapped to Polar Stereographic at 3km, centred at (30텒N, 85텒W).
2.6
IMPLEMENTATION OF SAI
Three main steps are identified. The user manually interacts with the SAI software during the
installation step, and the SAI preparation and execution steps are automatically monitored by the
Task Manager (if real-time environment is selected).
EUMETSAT Satellite Application
Facility to NoWCasting & Very
Short Range Forecasting
Software Users Manual
of the SAFNWC/MSG:
Scientific part for the PGE08
SAF/NWC/INM/SCI/SUM/08
Issue: 1.0 Date: 17 January 2002
Page: 10/ 13
2.6.1 The SAI installation step:
Previous condition and licences
The NWCLIB v.1.4 for the SAFNWC/MSG must be correctly installed in the same directory
where the PGEs are going to be delivered (see SVD).
Installation and building of the executable
The software installation procedure does not require special resources and, it is restricted to
decompress the distribution file (a gz-compressed tar file) and to successfully build the executable
PGE08 file to be stored into the $SAFNWC/bin directory. The installation steps that must be done
for PGE08 follow the SVD.
After the steps shown in the SVD, PGE08 of the SAFNWC/MSG is installed and configured in
the system. The operational use of SAFNWC PGE08 SAI requires the definition of some
configuration files in order to select the regions to be processed and some needed configurable
parameters of the SAI model configuration file. Also it is necessary to specify the name of the
configuration file, define the neural network configuration files and the ASCII LUT file for
denormalization.
The definition of the region follows the steps in the SAFNWCLIB.
•
The satellite angles function is provided by the SAFNWCLIB.
•
The topography is obtained from GTOPO30. The ReadTOPO module manages and
retrieves the topography for the region to be processed. The land/sea mask is thus derived
from this data.
The functions stored in the NN.c module manage the neural network processing.
The PGE08 model configuration file is provided with the software package. The configuration file
is one ASCII file, further modification can be easily performed with a test editor. The
configuration file manages all processes in the executable file.
The automatic set of pre-defined time scheduling (of the preparation step) relies on Programmed
Task Definition Files.
2.6.2 The SAI preparation step:
The preparation step (region configuration) is performed when the PGE01(CMa) run. This PGE is
mandatory for PGE08.
2.6.3 The SAI execution step:
The execution step is the real-time processing of SEVIRI images over the region. This process
consists in the launch of the command: PGE08 along with the required parameter: slot, region file
name and model configuration file by the Task manager after PGE01 has finished. The LI is then
performed following the configuration file.
ANNEX 1: ANCILLARY DATA
Atlas and sea/land mask dataset covering the whole MSG disk in the default satellite projection at
full SEVIRI IR horizontal resolution are available within the SAFNWC software package and are
used in the elaboration of SAI. The source is GTOPO30 (available on internet
http://edcwww.cr.usgs.gov/landaac/gtopo30/gtopo30.html) which is a global digital elevation
EUMETSAT Satellite Application
Facility to NoWCasting & Very
Short Range Forecasting
Software Users Manual
of the SAFNWC/MSG:
Scientific part for the PGE08
SAF/NWC/INM/SCI/SUM/08
Issue: 1.0 Date: 17 January 2002
Page: 11/ 13
model. Elevations in GTOPO30 are regularly spaced at 30-arc seconds (approximately 1
kilometre). The horizontal coordinate system is decimal degrees of latitude and longitude
referenced to WGS84. The vertical units represent elevation in meters above mean sea level.
The elevation values range from -407 to 8,752 meters. In the DEM, ocean areas have been
masked as "no data" and have been assigned a value of -9999. Lowland coastal areas have an
elevation of at least 1 meter, so in the event that a user reassigns the ocean value from -9999
to 0 the land boundary portrayal will be maintained. Due to the nature of the raster structure
of the DEM, small islands in the ocean less than approximately 1 square kilometre will not be
represented. For more detail information see the GTOPO30 documentation web page
http://edcdaac.usgs.gov/gtopo30/README.html#h1.
The altitude of each SEVIRI IR pixel over land (and lakes) is obtained by averaging GTOPO30
values located inside this pixel, whereas oceanic pixel are given a zero value.
EUMETSAT Satellite Application
Facility to NoWCasting & Very
Short Range Forecasting
Software Users Manual
of the SAFNWC/MSG:
Scientific part for the PGE08
SAF/NWC/INM/SCI/SUM/08
Issue: 1.0 Date: 17 January 2002
Page: 12/ 13
ANNEX 2. REFERENCES
Bishop C.M. 1995: Neural networks for pattern recognition. Oxford: Clarendon Press, 482pp.
Chedin A., Scott N.A., Wahiche C. & Moulinier P., 1985: The improved initialization inversion
method: A high resolution physical method for temperature retrievals from satellites of the
TIROS-N series. Journal of Climatology and Applied Meteorology, 24, pp 124-143.
Chevallier, Che ruy F., Scott N.A. & Chedin A., 1998: A neural network approach for a fast and
accurate computation of longwave radiation. J. Appl. Meteor., 37, pp 1385-1397.
Emery, W.J., Y. Yunyue, G.A. Wick, Pschluessel, and R.W. Reynolds,1994: Correcting infrared
satellite estimates of sea surface temperature for atmospheric water vapor attenuation. J.
Geophys. Res., 99, 11 586-11 601.
Eyre J.R., 1991: A fast radiative transfer model for satellite sounding systems, ECMWF technical
memorandum, 176. ECMWF,Reading, United Kingdom.
Martınez M.A., et al. 1999: MSG clear air products development with synthetic data. Proc. The
1999 EUMETSAT Meteorological Satellite Data User's Conference, Copenhagen, Denmark.
Martınez M.A., and Velazquez M., 2000. SAFNWC Scientific Report for MTR on PGE07 and
PGE08.
Martınez M.A., and Velazquez M., 2001: A new method to correct dependence of MSG IR
Radiances on satellite zenith angle, using a neural network. Proc. The 2001 EUMETSAT
Meteorological Satellite Data User's Conference, Antalya, Turkey.
Pankiewicz G.S., Martınez, M.A., Fernandez J.M. & Velazquez M., 1999: An evaluation of the
neural networks being used to obtain MSG clear-sky products: TPW, LPW & SAI.
Nowcasting SAF visiting scientist activity, Madrid, Spain.
Tjemkes S.A. & Schmetz J., 1998: Radiative transfer simulations for the thermal channels of
Meteosat Second Generation, EUMETSAT technical memorandum.
Zufiria P., Berzal, A., Martinez, M.A., & Fernandez J.M., 1999: Neural network processing for
the support of nowcasting and very short range forecasting. Nowcasting SAF visiting scientist
activity, Madrid, Spain.
EUMETSAT Satellite Application
Facility to NoWCasting & Very
Short Range Forecasting
Software Users Manual
of the SAFNWC/MSG:
Scientific part for the PGE08
SAF/NWC/INM/SCI/SUM/08
Issue: 1.0 Date: 17 January 2002
Page: 13/ 13
ANNEX 3. ACRONYMS
ADD
Architectural Design Document
CMa
Cloud Mask (also PGE01)
EUMETSAT European Meteorological Satellite Agency
GOES
Geostationary Operational Environmental Satellite
HRIT
High Rate Information Transmission
ICD
Interface Control Document
INM
Instituto Nacional de Meteorologıa
IR
Infrared
LI
Lifted Index
LPW
Layer Precipitable Water (also PGE07)
MLP
Multilayer Perceptron
MNN
Multilayer Neural Network
MSG
Meteosat Second Generation
NN
Neural Network
NOAA
National Oceanic and Atmospheric Administration
NWP
Numerical Weather Prediction
PGE
Product Generation Element
RTM
Radiative Transfer Model
RTTOV
Rapid Transmissions for TOVs
SAI
Stability Analysis Imagery (also PGE08)
SAF
Satellite Application Facility
SAF NWC
SAF to support NoWCasting and VSRF
SEVIRI
Spinning Enhanced Visible & Infrared Imager
SUM
Software User Manual
SVD
Software Version Description
SYNSATRAD Synthetic Satellite Radiance
TIGR
TOVS Initial Guess Retrieval
Download