Tsunami deposits provide information for estimating the magnitude and flow conditions of paleotsunamis, and inverse models have potential for reconstructing hydraulic conditions of tsunamis from their deposits. The majority of the previously proposed models are based on oversimplified assumptions and possess some limitations. We present a new inverse model based on the FITTNUSS model, which incorporates nonuniform and unsteady transport of suspended sediment and turbulent mixing. The present model uses a deep neural network (DNN) for the inversion method. In this method, forward model calculations are repeated for random initial flow conditions (e.g., maximum inundation length, flow velocity, maximum flow depth, and sediment concentration) to produce artificial training data sets of depositional characteristics such as thickness and grain-size distribution. The DNN was then trained to establish a general inverse model based on artificial data sets derived from the forward model. Tests conducted using independent artificial data sets indicated that this trained DNN can reconstruct the original flow conditions from the characteristics of the deposits. Finally, the model was applied to a data set of 2011 Tohoku-oki tsunami deposits. The predicted results of flow conditions were verified by the observational records at Sendai plain. Jackknife resampling was applied to estimate the precision of the result. The estimated results of the flow velocity and maximum flow depth were approximately 5.4 ± 0.1 m/s and 4.1 ± 0.2 m, respectively, after the uncertainty analysis. The DNN shows promise for reconstruction of tsunami characteristics from its deposits, which would help in estimating the hydraulic conditions of paleotsunamis. Plain Language Summary This study presents an inverse model that uses an artificial intelligence technique to estimate the hydraulic conditions of paleotsunamis from deposits. The estimated flow conditions are essential tools for disaster resilience and tsunami hazard mitigation to reduce socioeconomic impact of tsunamis on coastal cities.

Abstract. The 2004 Indian Ocean tsunami caused significant economic losses and a large number of fatalities in the coastal areas. The estimation of tsunami flow conditions using inverse models has become a fundamental aspect of disaster mitigation and management. Here, a case study involving the Phra Thong island, which was affected by the 2004 Indian Ocean tsunami, in Thailand was conducted using inverse modeling that incorporates a deep neural network (DNN). The DNN inverse analysis reconstructed the values of flow conditions such as maximum inundation distance, flow velocity and maximum flow depth, as well as the sediment concentration of five grain-size classes using the thickness and grain-size distribution of the tsunami deposit from the post-tsunami survey around Phra Thong island. The quantification of uncertainty was also reported using the jackknife method. Using other previous models applied to areas in and around Phra Thong island, the predicted flow conditions were compared with the reported observed values and simulated results. The estimated depositional characteristics such as volume per unit area and grain-size distribution were in line with the measured values from the field survey. These qualitative and quantitative comparisons demonstrated that the DNN inverse model is a potential tool for estimating the physical characteristics of modern tsunamis.

Abstract. The 2004 Indian Ocean tsunami caused major topographic changes that resulted in significant economic losses and a large number of fatalities in the coastal areas. The estimation of tsunami flow conditions using inverse models has become a fundamental aspect of disaster mitigation and management. Here, in relation to the 2004 Indian Ocean tsunami, a case study involving the Phra Thong island in Thailand was conducted using inverse modeling that incorporates a deep neural network (DNN). The inverse analysis reconstructed the values of flow conditions such as maximum inundation length, flow velocity and maximum flow depth, sediment concentration from the post-tsunami survey around Phra Thong island. The quantification of uncertainty was also reported using the jackknife method. Using other models applied to areas in and around Phra Thong island, the predicted flow conditions were compared with the reported observed values and simulated results. The estimated depositional characteristics such as volume per unit area and grain-size distribution, were in line with the measured values from the field survey. These qualitative and quantitative comparisons demonstrated that the DNN inverse model is a potential tool for estimating the characteristics of modern tsunamis.

Abstract. The 2011 Tohoku-oki tsunami inundated the Joban coastal area in the Odaka region of Minamisoma City, up to 2,818 m from the shoreline. In this study, the flow characteristics of the tsunami were reconstructed from deposits using the DNN (deep neural network) inverse model, suggesting that the tsunami inundation occurred in the Froude-supercritical condition. The DNN inverse model effectively estimated the tsunami flow parameters in the Odaka region, including the maximum inundation distance, flow velocity, maximum flow depth, and sediment concentration. Despite having a few topographical instabilities that caused the flow height to fluctuate greatly, the reconstructed maximum flow depth and flow velocity were reasonable and close to the values reported in the field observations. The reconstructed data around the Odaka region were characterized by an extremely high velocity (12.1 m/s). This study suggests that the large fluctuation of flow depths at the Joban coast compared with the stable flow depths at the Sendai plain can be explained by the inundation in the supercritical flow condition.

No abstract