TEM training Dataset

1. Kashidtrainingdata.matThe synthetic training dataset used in this study, consisting of 30,000 one-dimensional layered-earth Transient Electromagnetic (TEM) models generated for training and validation of the deep learning inversion framework. The dataset includes the simulated TEM responses and corresponding subsurface resistivity models representing a wide range of geological scenarios relevant to submarine groundwater discharge investigations. 2. tem_inversion_model.h5The trained deep learning inversion model saved in HDF5 (.h5) format. This file contains the network architecture, trained weights, and optimizer state required to reproduce the TEM inversion results presented in the manuscript. The model can be loaded directly using compatible versions of the Keras/TensorFlow framework. 3. tem_inversion_model.kerasThe trained deep learning inversion model saved in the native Keras (.keras) format. This file contains the complete trained neural network, including architecture and learned parameters, and can be used for inference and reproduction of the inversion workflow described in the manuscript.