Application of artificial neural networks as a tool for moisture prediction in microbially colonized halite in the Atacama Desert

Abstract

The Atacama Desert is the driest and one of the most life-limiting places on Earth. Despite the extreme conditions, microbial endolithic communities have been found inside halite rocks. The presence of these microbial communities is possible due to the hygroscopic properties of evaporitic rocks composed of sodium chloride. It is important to elucidate every possible water source in such a hyperarid environment. Therefore, in the present study, an artificial neural network (ANN) based model has been designed to predict the presence of liquid water on the surface of halite pinnacles. The model predicts the moisture formation using two basic meteorological variables, air temperature, and air relative humidity. ANNs have been successfully employed for the first time as a tool for predicting the appearance of liquid water, a key factor for the endolithic microbial communities living in the driest part of the Atacama Desert. The model developed is able to correctly predict the formation of water on the surface of the halite pinnacles 83% of the cases. We anticipate the future application of this model as an important tool for the prediction of the water availability and therefore potential habitability of lithic substrates in extreme environments on Earth and perhaps elsewhere. Key Point Prediction of moisture in halite from Atacama using an ANN-based model