Seismic phase picking and magnitude estimation are crucial components of real-time earthquake monitoring and early warning. Here, we test the potential of deep learning in real-time earthquake monitoring. We introduce DynaPicker, which leverages dynamic convolutional neural networks for event detection and arrival-time picking, and use the deep-learning model CREIME for magnitude estimation. This workflow is tested on the continuous recording of the Turkey earthquake aftershock sequences.

We present and test an algorithm that integrates geological modelling into deterministic geophysical inversion. This is motivated by the need to model the Earth using all available data and to reconcile the different types of measurements. We introduce the methodology and test our algorithm using two idealised scenarios. Results suggest that the method we propose is effectively capable of improving the models recovered by geophysical inversion and may be applied in real-world scenarios.

Fluids and melts in the mantle are key to the Earth’s evolution. The main driving force for their transport is the compaction of the porous mantle. Numerically, the compaction equations can yield unphysical negative liquid fractions (porosity), and it is necessary to enforce positive porosity. However, the effect of such a treatment on liquid flow and mass conservation has not been quantified. We found that although mass conservation is affected, the liquid pathways are well resolved.

Fourier transform infrared (FTIR) was performed on shale before and after solvent extraction. The extraction yield from shale with THF is higher than other solvents. The organic-C-normalized yield of a mature sample is higher than other samples. The aromaticity of organic matter increases, and the length of organic matter aliphatic chains  does not vary monotonically with increasing maturity. The results will help in the selection of organic solvents for oil-washing experiments of shale.