The advent of modern supercomputers technology, in conjunction with larger, more comprehensive observation datasets, has led to a paradigm shift in early forecasting of Tsunami modeling. Full geological fault parameters of Tsunami wave generation and scenarios, and historical data of earthquakes are routinely employed as tools to estimate propagation properties of the Tsunami wave with high resolution. The Artificial Intelligent (AI) methods fit simulated Tsunami to real-time observed data by iteratively updating estimates of propagation properties. In this paper, we explore approaches to apply AI by preparing a dataset based on a linear tsunami model (TUNAMI) and conducting an Artificial Neural Network (ANN) training constructed based on Random Forest (RF) regression algorithm. AI Performance is assessed through controlled numerical experiments and resulting Tsunami height (SSH) and Estimated Time of Arrival (ETA) resulted from both approaches.