What is feeding the eruption of Geldingadalir is not known, but lava flows steadily to the surface. As it is not forbidden to think about it, this model here is used to figure out a possibility. Convection rolls under the tectonic plates are ultimately providing inflow of magma. The location of the upwelling can be connected with the Reykjanes Ridge, because it is obviously responsible for the creation of the northern part of the Atlantic Ocean, and must therefore have convection rolls at each side of it. Magma will therefore rise in between those convection rolls, and eventually some of that material can enter the tectonic plates.
The upwelling takes place along a line that can be traced along the Reykjanes Ridge. As the line turns abrubtly close to the shore of Iceland, the original, regularly shaped line can be prolonged further, as the convecting, half-molten magma can not make sharp turns, and can therefore be traced logically over long distances according to large scale features like the Reykjanes Ridge. Seeing the most likely spot for origin of ascending magma, we can keep on tracing the flow within the tectonic plate. The tectonic plate, in turn, is combined of two main layers, ductile and brittle. It is found that the molten lava emerging on the surface has been melted at the depth of 14-16 km. The relevant research is described here: https://www.visindavefur.is/svar.php?id=81494 It happens to be around the Moho division. Accordingly, we have a clue about how much material is available for the eruption. It is a dike reservoire, 23 km long and 2 km high, or close to 50 square km large ‘wall’. If we knew the width, we could guess for how long this eruption will be going on.
Magic Magma 