Exploring soil radon (Rn) concentrations and their connection to geological and meteorological factors.

Abstract

The relationship between soil radon and meteorological parameters in a region can provide insight into natural processes occurring between the lithosphere and the atmosphere. Understanding this relationship can help models establish more realistic results, rather than depending on theoretical consequences. Radon variation can be complicated to model due to the various physical variables which can affect it, posing a limitation in atmospheric studies. To predict Rn variation from meteorological parameters, a hybrid mod el called multiANN, which is a combination of multi-regression and artificial neural network (ANN) models, is established. The model was trained with 70% of the data and tested on the remaining 30%, and its robustness was tested using the Monte-Carlo method. The regions with low performance are identified and possibly related to seismic events. This model can be a good candidate for predicting Rn concentrations from meteorological parameters and establishing the lower boundary conditions in seismo-ionospheric coupling models.