 |
| Titel |
Diva software, a tool for European regional seas and Ocean climatologies production |
| VerfasserIn |
M. Ouberdous, C. Troupin, A. Barth, Aida Alvera-Azcárate, J.-M. Beckers |
| Konferenz |
EGU General Assembly 2012
|
| Medientyp |
Artikel
|
| Sprache |
Englisch
|
| Digitales Dokument |
PDF |
| Erschienen |
In: GRA - Volume 14 (2012) |
| Datensatznummer |
250069561
|
|
|
|
|
|
| Zusammenfassung |
| Diva (Data-Interpolating Variational Analysis) is a software based on a method designed to
perform data-gridding (or analysis) tasks, with the assets of taking into account the
intrinsic nature of oceanographic data, i.e., the uncertainty on the in situ measurements
and the anisotropy due to advection and irregular coastlines and topography. The
Variational Inverse Method (VIM, Brasseur et al., 1996) implemented in Diva consists in
minimizing a variational principle which accounts for the differences between the
observations and the reconstructed field, the influence of the gradients and variability
of the reconstructed field. The resolution of the numerical problem is based on
finite-element method, which allows a great numerical efficiency and the consideration
of complicated contours. Along with the analysis, Diva provides also error fields
(Brankart and Brasseur, 1998; Rixen et al., 2000) based on the data coverage and
noise.
Diva is used for the production of climatologies in the pan–European network
SeaDataNet. SeaDataNet is connecting the existing marine data centres of more than 30
countries and set up a data management infrastructure consisting of a standardized
distributed system. The consortium has elaborated integrated products, using common
procedures and methods. Among these, it uses the Diva software as reference tool for
climatologies computation for various European regional seas, the Atlantic and the global
ocean.
During the first phase of the SeaDataNet project, a number of additional tools were
developed to make easier the climatologies production for the users. Among these tools: the
advection constraint during the field reconstruction through the specification of a
velocity field on a regular grid, forcing the analysis to align with the velocity vectors;
the Generalized Cross Validation for the determination of analysis parameters
(signal-to-noise ratio);
the creation of contours at selected depths;
the detection of possible outliers;
the hydrostatic constraint for eliminating the potential hydrostatic instabilities
arisen from the combined analysis of temperature and salinity data in several
horizontal planes independently;
the specification of a variable correlation length over the domain, allowing one
to consider different scales of interest according to the location;
the computation of the error field based on the real correlation function of the
considered data;
the generation of semi-normed reference fields.
Collaboration with Diva users (marine data centres) permitted the identification of a
variety of problems that can occur in the Diva analysis due to numerical computations and/or
data types (i.e. negative concentrations for certain data sets of bio-chemical and nutrient
data). To solve these problems, new options were designed and implemented additional for
Diva computation algorithms. Among these new options the user has the possibility to:
avoid negative values performing analyses based on transformed data (i.e.
anamorphosis transformation),
avoid unrealistic and/or negative concentrations due to small number of data
using semi-normed reference field generated with data sets from other layers, to
perform a layer analysis,
filter vertically the background (mean data) or reference fields for vertical field
coherence.
Diva analysis tools and options, as well as Climatologies validation tools will be
presented, with a demonstration of efficiency of the new Diva options using bio-chemical and
physical data, through samples of climatologies for different regions of the Mediterranean
sea. |
|
|
|
|
|
|