Portable AMV software

by Mary Forsythe — last modified Feb 01, 2013 12:30 PM

A number of steps were identified during discussions at the 10th International Winds Workshop, which should help to address the needs of the research and operational communities, with the ultimate aim of improving the AMV derivation approach.

CGMS-R39.25. CGMS agencies are invited to:

further test the performance of the NWCSAF AMV software package by testing the products in a NWP data monitoring/assimilation system, and
extend the current software to clear-sky WV AMVs

CGMS-R39.26. Satellite AMV providers are invited to examine the stand-alone AMV software package from the NWCSAF and to report back to CGMS-40.

CGMS-R39.27. EUMETAT NWCSAF to consider providing a tested option to allow running of alternative algorithms to support algorithm intercomparison studies.

IWW10.4. Co-chairs to summarise the proposed approach to address the points raised in the plenary discussion on developing portable software (see CGMS-38 EUM-WP-42). The NWC SAF, producers and co-chairs to undertake work to address the key actions.

Please contact the IWWG co-chairs for further information.

The starting point for the IWW10 debate was the desire to engage more with research groups by providing them with access to a portable AMV software package. The idea was that this package would provide a very good, read-to-go, starting point for researchers interested in getting involved in the problem of deriving atmospheric winds from satellite measurements. Furthermore, it would serve as the focal point for sharing and comparing developmental code among different research groups. It was also noted that such a portable AMV software package could be used by some for operational AMV production (similar to the NWP SAF scatterometer processing software). This could possibly lead to improved consistency between AMVs produced at different centres. As the system would be designed in a generic way it could also simplify the AMV reprocessing efforts.

During the discussion it became apparent that we should consider these two differing requirements separately:

Greater involvement of research groups to help develop and test new approaches for deriving satellite-based AMVs.
Increased collaboration between operational AMV producers to speed up improvements to AMV derivation and move towards more consistent approaches.

As a result of the discussion a proposal was put together to address these two requirements.

Javier Garcia-Pereda has a proceeding which describes a candidate algorithm already in use.

http://www.eumetsat.int/Home/Main/AboutEUMETSAT/Publications/ConferenceandWorkshopProceedings/2012/groups/cps/documents/document/pdf_conf_p60_s8_01_garcaper_v.pdf

The proposed approach to address requirement 1:

Put together a research starter kit via the IWWG web pages:

Provide a simple tutorial to get people started, including a list of up-to-date key references.
Provide links to the NWC SAF AMV portable software, which should be of sufficient skill for research users to use as a test AMV derivation system. In support of this we should make more effort within the IWWG community to evaluate and support development of this system. See also CGMS-38 EUM-WP-28.
Provide guidance on how to access and read geostationary/polar imagery and provide format conversion tools if required.
Provide guidance on how to access and read Numerical Weather Prediction (NWP) forecast model information for use in the height assignment step.
Provide links to visualisation software (McIDAS-V is one example that was proposed, but there may be others).
Provide a training dataset (e.g., inputs and outputs) so that research users can quickly start running and familiarising themselves with the software. One proposal was to provide the next AMV inter-comparison SEVIRI image case in an agreed format (e.g. netcdf). For background and comparison we could also provide access to the AMV inter-comparison producer datasets and analysis of results.
Provide an up-to-date record of the methods used by the operational producing centres (covered by section below).

</blockquote></HTML>

Encourage participation at the International Winds Workshops from research groups

Promote collaborative work within universities through studentships, presentations in university seminars, and conferences such as the annual EUMETSAT Meteorological Satellite Conference or the American Meteorological Society Conference on Satellite Meteorology and Oceanography.

Improve the literature in this field – ideally want current methods described in sufficient detail to allow replication.

The proposed approach to address requirement 2:

Producing centres encouraged to work towards standard interfaces and formats, and more modular code design to enable easier exchange and use of code by other centres.
Develop the IWWG web pages to provide:

<HTML><blockquote> Improved documentation of the AMV derivation system to help centres that are trying to develop and improve their set-up.

Development of a new web page to highlight where we may benefit from further inter-comparisons or development. The idea, proposed by Ken Holmlund, is outlined below:

Summarise the main AMV derivation steps e.g. target selection, tracking etc.

</blockquote></HTML> <HTML><blockquote> <HTML><blockquote> <HTML><blockquote>

For each one identify the different methods that have been tested e.g. for tracking: optical flow, Euclidean distance, cross-correlation – providing details as necessary

</blockquote></HTML></blockquote></HTML></blockquote></HTML> <HTML><blockquote> <HTML><blockquote> <HTML><blockquote>

Summarise whether the different approaches have been well-compared in the past (reference as appropriate) or if they would benefit from further inter-comparison to determine best practise. Also identify whether we would benefit from further research to develop new approaches (e.g. if thought to be an error-prone step).

</blockquote></HTML></blockquote></HTML></blockquote></HTML> <HTML><blockquote>