 |
| Titel |
Impacts of high resolution model downscaling in coastal regions |
| VerfasserIn |
Lucy Bricheno, Judith Wolf |
| Konferenz |
EGU General Assembly 2013
|
| Medientyp |
Artikel
|
| Sprache |
Englisch
|
| Digitales Dokument |
PDF |
| Erschienen |
In: GRA - Volume 15 (2013) |
| Datensatznummer |
250076857
|
|
|
|
|
|
| Zusammenfassung |
| With model development and cheaper computational resources ocean forecasts are becoming
readily available, high resolution coastal forecasting is now a reality. This can only
be achieved, however, by downscaling global or basin-scale products such as the
MyOcean reanalyses and forecasts. These model products have resolution ranging
from 1/16th - 1/4 degree, which are often insufficient for coastal scales, but can
provide initialisation and boundary data. We present applications of downscaling
the MyOcean products for use in shelf-seas and the nearshore. We will address
the question ’Do coastal predictions improve with higher resolution modelling?’
with a few focused examples, while also discussing what is meant by an improved
result.
Increasing resolution appears to be an obvious route for getting more accurate forecasts
in operational coastal models. However, when models resolve finer scales, this
may lead to the introduction of high-frequency variability which is not necessarily
deterministic. Thus a flow may appear more realistic by generating eddies but the
simple statistics like rms error and correlation may become less good because the
model variability is not exactly in phase with the observations (Hoffman et al.,
1995).
By deciding on a specific process to simulate (rather than concentrating on reducing rms
error) we can better assess the improvements gained by downscaling. In this work we will
select two processes which are dominant in our case-study site: Liverpool Bay. Firstly we
consider the magnitude and timing of a peak in tide-surge elevations, by separating out
the event into timing (or displacement) and intensity (or amplitude) errors. The
model can thus be evaluated on how well it predicts the timing and magnitude of
the surge. The second important characteristic of Liverpool Bay is the position
of the freshwater front. To evaluate model performance in this case, the location,
sharpness, and temperature difference across the front will be considered. We will
show that by using intelligent metrics designed with a physical process in mind,
we can learn more about model performance than by considering ’bulk’ statistics
alone.
R. M. Hoffman and Z. Liu and J-F. Louic and C. Grassotti (1995) ’Distortion
Representation of Forecast Errors’ Monthly Weather Review 123: 2758–2770 |
|
|
|
|
|
|