Strålingsdata i tilsigsmodeller for kraftverksdrift

This project's main objective was to improve hydropower producers' inflow forecasts during snowmelt, by evaluating radiation data used as input. Remotely sensed radiation data from EUMETSAT's geostationary MSG satellite are evaluated against in situ measurements, aiming at an operational information service. For short wave radiation (sunlight) and long wave radiation (thermal), the information products are named DSSF (Downwelling Surface Shortwave Flux), and DSLF (Downwelling Surface Longwave Flux), respectively. These are available in good temporal resolution and small delay. In addition, the Norwegian Met Office?s (MET's) forecasts of shortwave radiation are evaluated against ground measurements and the satellite data. Forecasts of longwave radiation have not been evaluated. For shortwave radiation, satellite data and meteorological forecasts are compared to ground measurements at around 50 stations in NIBIO's network, as well as ten sensors located in mountainous areas as a part of the project. The evaluation is focused around clearness indices normalising for solar elevation and clear-sky atmospheric scattering. This is a more critical test than evaluating the raw radiation estimates, which may produce good results merely because all data sources have the same day length and solar elevation. We conclude that the satellite measurements are valuable both as supplement and replacement to ground measurements, particularly in remote areas. For radiation forecasts from MET's Arome model, the investigation shows good correspondence with satellite data as well as ground measurements. Some regional differences are found, in particular, the satellite product show weaknesses for the northernmost part of Norway, near the edge of the satellite field of view. Also along the Western coast there are differences between the data sets, which may have several causes. The good results for radiation forecasts are important for inflow models, which have weather forecasts as their primary input. For longwave radiation, the DSLF product relies on remotely sensed cloud observations, and information on ground temperature and total precipitable water extracted from the ECMWF atmospheric model. To a much greater extent than for shortwave radiation, this background information has a profound influence on the final estimate. With no operational gauge network for longwave radiation, the project has evaluated the DSLF product at six meteorological research stations owned by NIBIO, Sintef and the project participants. The results reveal that DSLF underestimates incoming longwave radiation during winter, and overestimates during summer. The winter underestimation is larger but has less influence on snow melt than the summer overestimation. A literature study showed that models estimating longwave radiation from air temperature and vapour pressure at reference height have similar systematic errors. These models have much in common with how DSLF uses ECMWF information. The results show that a linear correction based on latitude removes most of the bias, both for a larger set of stations reported in literature, and for the few stations in this investigation. It is therefore recommended to use a similar correction also for the DSLF data set. The effect of different radiation data on inflow modelling was investigated by simulating runoff in ten catchments in southern Norway, using four combinations of ground- and satellite-based measurements of short- and longwave radiation. The modelling made use of the Enki framework, which was modified in order to accept longwave radiation as an input variable. Apart from this technical change, the model was not further adapted to the altered data situation. Inflow modelling using forecasted radiation input was not tested. Broadly, the simulated inflow is moderately improved using remotely sensed data for both short- and longwave radiation. The hydrological model was calibrated to each input data set separately. The four input combinations showed considerable differences in the diurnal melt cycle, and in the relative importance of the main energy components. Shortly summarised, the project has shown that: 1. Remotely sensed data for both short- and longwave radiation are available for automated download for operational purposes, with down to 30-min temporal resolution, and less than one hour after acquisition. 2. Inflow simulations using these data are improved compared to ground measured shortwave and simulated longwave radiation, which is the available alternative. 3. For shortwave radiation, there is a good correspondence between the remotely sensed estimates and the forecasted values from the Norwegian Met Office. 4. For longwave radiation, an empirical correction is necessary to improve the remotely sensed data. A correction model is found which needs only latitude as input.

Prosjektet har evaluert nye strålingsdata, og konkluderer med at tilsigsprognosene ved bruk av disse er bedre enn tidligere. Det er gitt anbefalinger for både bruk og korrigering av strålingsinformasjon Flere modeller og prototyper er forbedret og nyskrevet under prosjektet. En tjeneste for foredling av fjernmålingsinformasjon er utviklet og klar til uttesting

Energimarkedet har endret seg betydelig de siste årene og forventes å endre seg enda hyppigere i framtida. Flere typer energiprodukter vil etterspørres og energiprisene forventes å variere hyppigere. Dette åpner nye muligheter for en vannkraft produsent til å tjene penger, men øker også risikoen for å havne i ubalanse og tape penger. Tilsigsprognosens treffsikkerhet vil i mange situasjoner være nøkkelen til om han skal tene eller tape penger. Snømagasinet er den viktigste ressursen til vannkraftprodusenter i Norden. Tilsiget fra dette avgjøres i stor grad av stråling, temperatur og luftfuktighet. Dagens operative modeller tar hensyn til kun den ene av disse faktorene og risikoen for feil prognose er stor der de to andre er dominerende. Dette prosjektet vil muliggjøre bruk av stråling i vannkraft bransjens operative tilsigsmodeller. Dette vil ha stor betydning primært for prognosering av snøsmelting og dermed tilsigsprognosens treffsikkerhet gjennom våren og tidlig sommer, men også for beregning av fordampning og prognosens treffsikkerhet spesielt ved nedbørstilfeller etter lengre tørkeperioder. Tilgjengligheten av informasjon om stråling og fokus på langsiktig vannbalanse fremfor kortsiktig treffsikkerhet har vært sterkt medvirkende til at modellene kun har vært basert på nedbør og temperatur som inngangsdata. Mer detaljerte, fysisk baserte og fordelte tilsigsmodeller samt et økende fokus på korttids treffsikkerhet fremtvinger bruk av stråling for å utnytte modellenes potensiale og oppnå økt treffsikkerhet. Det er flere kilder til informasjon om stråling. Prosjektet vil undersøke hva hver av disse kildene kan bidra med og hvordan de kan kombineres for å gi det best tilgjenglige strålingsgrunnlaget for tilsigsmodellene. Prosjektet vil med støtte av studentprosjekt studere effekten av tettere observasjonsnettverk og strålingsberegnigens bidrag til usikkerheten i prognosene.