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\pm0.140 m/s and 4.11\pm0.152 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.

Despite the importance of turbidity currents in environmental and resource geology, their flow conditions and mechanism are not well understood. To resolve this issue, a new method for the inverse analysis of turbidity current using deep learning neural network (DNN) was proposed. This research aims at verifying and calibrating this method using artificial and flume experiment datasets. The forward model based on the shallow water equation was employed in this study to produce artificial datasets of turbidite deposits. DNN was applied to two hundred artificial datasets and two sets of experiment data. As a result of inversion by DNN, spatial distributions of grain size and thickness of experimental turbidites were reconstructed accurately. With regard to hydraulic conditions, the flow heights were reasonably estimated, and sediment concentrations reconstructed were reasonable except for regions where the values measured in experiments were extremely low. Flow duration also showed reasonable reconstructed values. In contrast to the other values, there was a large discrepancy between the measured and reconstructed values of flow velocity. The reason for this discrepancy in velocity may be attributed to inaccuracy in closure functions employed in the forward model. Future application to actual data of natural scale turbidite is anticipated.

This study aims to establish an inverse model for reconstructing properties of turbidity currents from their deposits. Although several attempts of in-situ measurements have been succeeded, flow properties of large-scale turbidity currents that build submarine fans remain unclear. Recurrence interval of turbidites in natural levees or lobes of submarine fans is generally 100s years, while in-situ observation in modern submarine canyons reported yearly to monthly occurrence of turbidity currents. That is, turbidity currents that substantially form submarine fans may be extremely low frequency and large-scale than currently observed. To understand the behavior of fan-building flows occurring with sub-millennial scale recurrence interval, the quantitative reconstruction of flow conditions from geologic records is required. However, inversion of flow properties in natural scale has been impractical because the heavy computational load of 2D or 3D models of turbidity currents has prevented to perform inversion requiring iteration of calculations. To this end, we propose a method using 2D shallow-water model with the deep-learning neural network (DNN). In this method, a horizontal 2D shallow-water model of turbidity currents was employed as the forward model. Numerical simulation was repeated to obtain horizontal distribution of thickness of turbidites under various initial conditions, and then this synthetic data set was used for supervised training of DNN. After the training phase finished, DNN properly estimated initial conditions of turbidity currents (e.g. initial flow height, velocity, etc.) from artificial test data set that was also produced from the forward model. In previous methods, the computational cost of 2D model was too high to be employed as the forward model for turbidite inversion, while our methodology can reconstruct initial conditions of turbidity currents instantaneously. Although production of the training data set requires 1000s times repetition of calculation, all calculation of forward model can be perfectly parallelized, so that the problem of calculation efficiency can be easily solved by using PC cluster. Future application of our method to actual deposits will contribute to understand the development processes of submarine channels or lobes in the long term.

This study proposes a new definition of plane bed regimes. The upper plane bed is defined as a flat bedform that appears under conditions of active suspended load, whereas the lower plane bed forms when sediment particles are moved only by bed-load. Previous studies have recognized that two types of plane bed could be distinguished by shear stress, particle size, or flow regime. However, the compilation of a large amount of existing open-channel flow data indicates that the plane bed regimes cannot be differentiated by existing definitions. Newly produced phase diagrams indicate that plane bed data plot in two separate regions, and the gap between the two regions corresponds to the threshold condition of suspension. Further investigation is needed to understand the physical mechanisms generating lower plane beds and characterizations of parallel laminae from lower and upper plane beds.

Despite the importance of turbidity currents in environmental and resource geology, their flow conditions and mechanism are not well understood. To resolve this issue, a novel method for the inverse analysis of turbidity current using deep learning neural network (DNN) was proposed. This study aims to verify this method using artificial and flume experiment datasets. Development of inverse model by DNN involves two steps. First, artificial datasets of turbidites are produced using a forward model based on shallow water equation. To develop a inverse model, DNN then explores the functional relationship between initial flow conditions and characteristics of the turbidite deposit through the processing of artificial datasets. The developed inverse model was applied to 200 sets of artificial test data and four sets of experiment data. Results of inverse analysis of artificial test data indicated that the flow conditions can be precisely reconstructed from depositional characteristics of turbidites. For experimental turbidites, spatial distributions of grain size and thickness were accurately reconstructed. With regard to hydraulic conditions, reconstructed values of flow heights, sediment concentrations, and flow durations were close to the measured values. In contrast to the other values, there was a larger discrepancy between the measured and reconstructed values of flow velocity, which may be attributed to inaccuracies in sediment entrainment functions employed in the forward model.

Despite the importance of turbidity currents in environmental and resource geology, their flow conditions and mechanisms are not well understood. This study proposes a novel method for the inverse analysis of turbidity currents using a deep learning neural network (DNN) to better explore the properties of turbidity currents. The aim of this study is to verify the DNN inverse method using numerical and flume experiment datasets. Numerical datasets of turbidites were generated with a forward model. Then, the DNN model was trained to find the functional relationship between flow conditions and turbidites by processing the numerical datasets. The performance of the trained DNN model was evaluated with 2000 numerical test datasets and 5 experiment datasets. Inverse analysis results on numerical test datasets indicated that flow conditions can be reconstructed from depositional characteristics of turbidites. For experimental turbidites, spatial distributions of grain size and thickness were consistent with the sample values. Concerning hydraulic conditions, flow depth H, layer-averaged velocity U, and flow duration Td were reconstructed with a certain level of deviation. Greater discrepancies between the measured and reconstructed values of flow concentration were observed relative to the former three parameters (H, U, Td), which may be attributed to difficulties in measuring the flow concentration during experiments. The precision of the reconstructions for experimental datasets was estimated using Jackknife resampling. Although the DNN model did not provide perfect reconstruction, it proved to be a significant advance for the inverse analysis of turbidity currents.