Vinsamlegast notið þetta auðkenni þegar þið vitnið til verksins eða tengið í það: http://hdl.handle.net/1946/4393
Discharge rating curve using bayesian statistics
Discharge in rivers is commonly estimated by the use of rating curve constructed from pairs of water level and discharge measurements. Water level measurements are collected continuously from each river while pairs of discharge and water level are only collected couple of times a year due to high cost. The need for accurate estimation of discharge is important for constructions as bridges, hydroelectric power plants as well as for hydrological models. The methodology currently used by the Icelandic Meteorological Office is based on the standard power-law. The power-law is derived from a theoretical basis and serves as an appropriate model in most cases. However, in some natural settings deviations from this form arise. The new methodologies presented in this thesis account for the deviations from the standard power-law by extending it with a smooth B-spline function or by assuming two of the are a function of water level and modeled them with B-splines. These methodologies have shown to perform equally well or better than the current methodology used by the Icelandic Meteorological Office.