| Summary: | In this project, a statistical model for the surface mass balance of Vatnajökull glacier is proposed. The goal was to build a model that can be used to predict the surface mass balance of Vatnajökull glacier with a small number of explanatory variables. A model was built for both winter mass balance and summer mass balance. Only three explanatory variables where used in each model. Surface mass balance measurements on Vatnajökull glacier have been carried out since 1992. There exists a simulation of Vatnajökull's surface mass balance for the period 1981-2016, which is extracted from a regional climate model. The simulations for the years for which mass balance measurements are available, 1992 to 2016, fits the measured data well. Therefore, the simulations for the years 1981 to 1991 are added to the data available from the surface mass balance measurements. It is taken into account that the simulated surface mass balance values for the years 1981 to 1991 are not measurements, but outputs from a model, by assuming a higher standard deviation for these years within the statistical model. This is done by giving the years where direct measurements were made more weight through a weight matrix. Thus, the statistical model for Vatnajökull glacier’s mass balance is a weighted linear regression model. To evaluate the prediction performance of the statistical model, a long term estimation of Vatnajökull glacier´s mass balance, which exists for the years 1890 to 1980, is compared with the prediction from the weighted linear regression model. This comparison shows that the proposed statistical model is in line with the long term estimation of Vatnajökull glacier´s mass balance. In addition, it gives annual predictions of mass balance over the years from 1890 to 1980 as opposed to predictions that are averages over decades. Í þessu verkefni var búið til tölfræðilíkan fyrir afkomu Vatnajökuls. Markmiðið var að gera líkan til að spá fyrir um afkomu Vatnajökuls með fáum skýribreytum. Gert var líkan fyrir vetrarafkomu og ...
|
|---|