Steady state Kalman filter for operational storm surge forecasting

Transcription

Steady state Kalman filter for operational storm surge forecasting
Steady state Kalman filter for operational storm surge
forecasting system based on the
Dutch Continental Shelf Model v6
JULIUS SUMIHAR 1 , FIRMIJN ZIJL1 , MARTIN VERLAAN1,2
1
Deltares, P.O. Box 177, 2600 MH Delft, The Netherlands
TU Delft, P.O. Box 5, 2600 AA, Delft, The Netherlands
2
In the Netherlands, accurate sea level prediction is important. For example, during storm surge conditions accurate sea level prediction is needed to support the decision
to close the storm surge barriers in the Eastern Scheldt and the Rotterdam Waterway, send
out the dike watch or even activate an evacuation scenario. Sea level prediction is also important for computing tidal reduction for hydrographic survey. Since recently, the newly
developed Dutch Continental Shelf Model v6 has been running operationally to produce the
sea level prediction (Zijl et al, 2013). A steady state Kalman filter has been implemented
for DCSMv6 to increase its short term forecast accuracy (Zijl et al, 2015). In this talk, the
Kalman filter setup will be presented, including the choice of the noise model, the use of covariance localization, and the technique for computing the steady state Kalman gain as well
as the selection of the assimilation stations, aided by modified ensemble-based observation
impact analysis (Liu and Kalnay, 2008; Verlaan and Sumihar, 2015). The Kalman filter for
DCSMv6 has been running operationally since 2013. Along the Dutch coast, it successfully
improves on average the hindcast accuracy by around 50% and the forecast accuracy up to
lead times of 9 – 15 hours after the last assimilated water level measurements. Crucially,
the Kalman filter especially improves the forecast accuracy during storm events.
REFERENCES
Liu, J. and Kalnay, E., 2008. Estimating observation impact without adjoint model
in an ensemble Kalman filter. Quart. J. of the Royal Meteorological Society, 134, 13271335.
Verlaan, M. and Sumihar, J., 2015. Observation sensitivity based on ensemble. in
preparation.
Zijl, F., Verlaan, M. and Gerritsen, H., 2013. Improved water-level forecasting for
the Northwest European Shelf and North Sea through direct modelling of tide, surge and
non-linear interaction. Ocean Dynamics, 63, 823-847.
Zijl, F., Sumihar, J. and Verlaan, M., 2015. Application of data assimilation for
improved operational water-level forecasting on the Northwest European Shelf and North
Sea. Ocean Dynamics, submitted.