Replenishment of magma chambers: comparison of fluid-mechanic experiments with field relations

 Magma chambers are commonly replenished at low Reynold’s number by liquids denser than that already resident in the chamber, and, nearly always, the injected liquid is less viscous than the resident liquid in the magma chamber. We report fluid-mechanic experiments that are designed to investigate this case, specifically treating the injection process itself, before a multilayered, convecting system develops. The injected fluid is emplaced as a viscous gravity current with a nose slightly elevated above the floor, trapping a thin layer of ambient fluid beneath it. Further, these injections develop a flow-front instability that takes the form of fingers that extend in the direction of the flow. Using scaling arguments, we model the height and the length of the current as functions of time, and we use dimensional analysis and our experimental data to model the wavelength of the fingers. The fingers and other structures observed in the experiments correlate well with similar features found in silicic intrusions that have been replenished with basic liquid. We argue that these and associated mafic pillows are formed by this finger instability. Our scaling can be combined with finger widths measured in the field to estimate magma reinjection rates. We suggest that structures observed in similar environments–composite lava flows, sills, and dikes–also formed by a flow-front instability. When applied to basic replenishments of basic magma chambers, the scaling arguments constrain the time required for emplacement. The emplacement time scale is short relative to the cooling time scale, indicating that, even when traveling tens of kilometers, such injections cool little during emplacement. Received: 27 October 1994/Accepted: 13 June 1995