Interpreting magma dynamics through a statistically refined thermometer. Implications for clinopyroxene Fe–Mg diffusion modeling and sector zoning at Stromboli

We present a more accurate clinopyroxene-only thermometer for shoshonite melts calibrated through the statistical analysis of both experimental and thermodynamically derived compositions. The calibration data set refers to the compositions of magmas at Stromboli, one of the most studied volcanoes in Italy and of broad significance as the type locality for Strombolian activity. The predictive power of the clinopyroxene-only thermometer is potentially very high, with an uncertainty of only ±6 °C for crystallization conditions of 10–900 MPa, 1,000– 1,200 °C, 1–5 wt.% H2O, and −1.66 to 1.27 NNO. For compositions extremely different from those of Stromboli magmas (i.e., trachytic and phonolitic melts), the error of estimate may increase up to ±30 °C but is still a relatively low uncertainty with respect to other thermometers from the literature. The strong linear correlation between T and Mg#melt allows to test for equilibrium between natural clinopyroxene compositions and their host magmas, supporting the use of the clinopyroxene-only thermometer with alkaline eruptive products to minimize the error of temperature. We illustrate the potential of this integrated approach by modeling temperature changes for clinopyroxene compositional zoning caused by recharge episodes of mafic magmas at Stromboli. With respect to previous thermometric equations, the percentage error on diffusion timescales decreases from ±510% to ±21%. Owing to its high accuracy, the thermometer is also suitable for estimating the crystallization temperature of hourglass sector-zoned clinopyroxenes, where kinetic effects induce differential elemental partitioning into {100} and {−111} sectors.