Development of new mathematical methods for the optimization of wind and wave
forecasts with emphasis in the local adaptation of the results.
The results of the WP 1 will accurately determine the statistical distributions in which the meteorological data better fit. The main objective of WP 2 is to develop new methods based on techniques of Information Geometry for optimizing the corresponding forecasts. In particular, the elimination of possible systematic biases will be pursued, a problem especially common when focusing on local predictions. This is an essential issue in many applications affecting renewable energy issues.
In the literature, different approaches that could answer this question are
presented with two basic options: The MOS (Model Output Statistics) and dynamic
methods (e.g. Kalman filters). In this context, previous work of our group can
be also categorized (see publications
of Kallos - Galanis et. al.).
Until now, the use
of the above mentioned methods involve the assumption that
the data processed are embedded in an Euclidean space, since the minimization of
possible errors is based on regression techniques using least square methods.
However, this is a simplification of the real problem as the family of
distributions constitutes a generalization of classical Euclidean spaces: it is
a manifold in which the geometrical properties may be quite distant from the
classical ones. For example, in case of a sphere the shortest distance between
two elements is not given by a segment but by the maximum cycle through them. In
general, the distances in such a complex nonlinear environment resulting to the
solution of second order boundary value problems of the form
Such problems, however, hardly admit analytical solutions and in most of the
cases can be only solved using numerical methods.
Methodology
Within the WP 2 the necessary tools ensuring the safe assessment of the
geometric properties that allow the correct estimation of distance and the
minimization of the discrepancies between forecasts and observations will be
identified and studied. . The differential equations describing the curves of
minimum length and hence distances will be formed.
The results of this workpackage can be outlined in the following poster.
On the geometrical entities
and techniques that better fit to the study of wind and wave forecasts for
selected areas in Greece.
Deliverables
·
One paper submitted to a scientific journal and/or a presentation in an
international conference.
·
Report for the geometrical entities and techniques that better fit to the study
of wind and wave forecasts for selected areas in Greece.
Back to the project's homepage.