Operational sea level forecasting in Estonia/Operatiivne meretaseme prognoosisusteem Eestis

The operational sea level forecasting system in Estonia, based on the HIROMB model forecasts and on 11 online sea level observation stations, is described and validated. The system is operational since 08.08.2005. Statistical analysis for the period 2006-2008 data is performed to investigate the properties of low-frequency sea level error and error estimation of high sea level events. A 7-day backwards moving average filter is the most appropriate for Estonian waters to correct the raw sea level forecast. For the time period 2006-2008 the forecast error of critically high sea level events is ±25 cm within ± 3 h between the observed and forecasted maxima. Taylor skill assessment procedures are applied to the data, covering the period 2009-2011. The data is divided into three forecast sub-ranges (low, medium and high sea level), to investigate the possibilities of the online error estimation. Smaller errors are present for medium and low sub-range and larger errors for the high sub-range. The necessities for further development are outlined. Key words: sea level, water level, modelling, HIROMB, Baltic Sea, Estonia, operational forecasting. On kirjeldatud ja valideeritud operatiivset meretaseme prognoosisusteemi Eestis. Prognoosisusteem pohineb HIROMB-mudeli prognoosil ja 11-l reaalajas toimival meretaseme mootejaamal. Susteem tootab alates 08.08.2005. Toos on analuusitud prognoosisusteemi efektiivsust aastail 2006-2008 ja 2009-2011. On uuritud madalsagedusliku veetaseme prognoosivea omadusi ja hinnatud prognoosiviga ekstreemsete veeseisude esinemise korral, mille tapne ennustamine on prognoosisusteemi pohirakenduseks. Uuring naitab, et 7-paevane tagasiulatuva liikuva keskmisega filter on Eesti rannikul sobivaimaks meetodiks, et prognoosimudeli valjundandmetest madalsageduslikku viga valja filtreerida. Ekstreemsete (ule kriitilise taseme) veetasemete prognoosi tapsus on ± 25 cm ja ajaline maksimaalse meretaseme saabumise tapsus ± 3 tundi. Varskemate, 2009.-2011. aasta meretaseme prognooside hindamiseks kasutatakse Taylori tapsushindamise meetodit. Reaalajas prognoosivea arvutamise voimalikkuse uurimiseks jagatakse meretaseme prognoosid kolme alamruhma (madal, keskmine ja korge). On leitud, et madalamate ja keskmiste veetasemete prognoosid on tapsemad kui korgemate veetasemete omad. On vaadeldud prognoosisusteemi edasiarendamise voimalusi ja pohjendatud vajadust parendada reaalajas toimivat mooteandmete kvaliteedikontrolli.