We compare one year of observed and predicted high waters. This may be a calendar year, or the past 12 months. A total of 18 fixed Danish coastal stations are used, as indicated on the map below.

Q is the absolute peak error in percent, averaged for the top 3 high waters at each station.

Calculation method:
HW = High Water, HW1=highest, HW2=second highest, HW3=third highest
Relative Absolute Peak Error PE = 100 * | HW_obs - HW_pred | / HW_obs [%]
Mean for a station MPE = (PE1+PE2+PE3) / 3
Mean for all stations Q = (MPE_st1+...+MPE_stN) / N [N=number of stations]

Q is calculated for forecast ranges 0-6, 6-12, ... 54-60 hours (for models run on a 6 hour schedule), or ranges 0-12, 12-24, ... 48-60 hours (for models run on a 12 hour schedule.

In Ministry reports, we use the one day forecast (18-24 or 12-24 hours).
For model intercomparison, we use the shortest forecast (0-6 or 0-12 hours).

• BSHcmod Operational 3-dimensional hydrodynamical model. Storm surge model since ultimo 2007 (for Limfjorden, since medio 2008).
• DKSS2010. New version of BSHcmod. Operationalised medio 2011.

12-month running mean of Q, valid for the model that is operational at any given time.

• Until primo 2008: Mike21 (table)
• Since primo 2008: BSHcmod (table)
• Since primo 2008: BSHcmod (plot)

• Operational (table)

• BSHcmod 2008-2011 (plot)
• BSHcmod 2008-2011 (table)

• 12 month running opr/test model intercomparison 2007-2011
• Opr/test model intercomparison 2007-2011
• Opr/test model intercomparison 2007-2011 - Limfjord stations omitted

• BSHcmod, 2011
• BSHcmod, several years: click on the map to pick a station