Case Studies/Beach Water Temperature
English version

Modelling of beach water temperatures - 2002 and 2003



The beach water temperature predictions of the DMI-BSHcmod ocean model have been compared to the measurements of beach water temperatures made by TV2-vejret.

Observations are made from the end of June to the end of August, and all temperatures are measured once a day at a depth of one meter.

The DMI-BSHcmod ocean model divides the ocean into many small boxes in a grid. The model temperature is taken from the gridpoint closest to the point of observation. This point is up to 10 km away, though most points are closer. Click here for more information. A point within one hour of the observation time is marked with a circle on the graphs and used in the statistics.

Click on the station map or choose a station from the list to view graphs of the observed and modelled temperatures, or click here to see an overview of the bias and standard deviation for all stations.


Stations
Assens
Balka
Bandholm
Bogense
Båring
Charlottenlund
Faaborg
Fanø
Flyvesandet
Grenaa
Henne strand
Hestehovedet
Hornbæk
Hvide Sande
Juelsminde
Kalundborg
Karrebæksminde
Kerteminde
Klitmøller
Kobæk
Lohals
Læsø
Løkken
Marielyst
Marstal
Nivå
Nyborg
Nyrup
Risskov
Rømø
Saltum
Sandvig
Skagen
Smørmosen
Solrød
Sæby
Thyborøn
Ulvshale
Veddelev
Vemmingbund
Østersøbadet


The bias and standard deviation (std) has been calculated for each station for the summer seasons of 2002 and 2003. The mean values are:
Mean bias Mean std
2003 0.62 1.11
2002 0.04 1.26
The values for each station are:
Statistics for 2003
Statistics for 2002
Corrections by comparison to observations

The above model data contains a bias. To minimize this, the model for the upcoming days can be corrected with the present day difference between model and observation. Click here to se a table of bias, standard deviation, covariance and correlation for the uncorrected and corrected models for 2003.

If our forecasts are to be used, they should be better than the simple assumption that the temperature in the upcoming days will be the same as the most resent observed temperature. The figures below show bias, standard deviation and correlation for the corrected model as described above and for the assumption that the temperatures are the same as observed one or two days ago. It is generally seen that the corrected model gives a smaller bias, and that the standard devialtion is higher and correlation is lower for the two days old observations.

2003

Statistics for 2003
Statistics for 2003
Statistics for 2003

2002

Statistics for 2002
Statistics for 2002
Statistics for 2002

Kristine Madsen - March 12th 2004