A deep learning technique-based data-driven model for accurate and rapid flood prediction

An accurate and rapid urban flood prediction model is essential to support decision-making on flood management, especially under increasing extreme precipitation conditions driven by climate change and urbanization. This study developed a deep learning technique-based data-driven flood prediction model based on an integration of LSTM network and Bayesian optimization. A case study in north China was applied to test the model performance and the results clearly showed that the model can accurately predict flood maps for various hyetograph inputs, meanwhile with substantial improvements in computation time. The model predicted flood maps 19,585 times faster than the physical-based hydrodynamic model and achieved a mean relative error of 9.5 %. For retrieving the spatial patterns of water depths, the degree of similarity of the flood maps was very high. In a best case, the difference between the ground truth and model prediction was only 0.76 % and the spatial distributions of inundated paths and areas were almost identical. The proposed model showed a robust generalizability and high computational efficiency, and can potentially replace and/or complement the conventional hydrodynamic model for urban flood assessment and management, particularly in applications of real time control, optimization and emergency design and plan.

A Novel Objective Function DYNO for Automatic Multi-variable Calibration and Application to Assess Effects of Velocity versus Temperature Data for 3D Lake Models Calibration

This study introduced a novel Dynamically Normalized objective function (DYNO) for multi-variable (i.e., temperature and velocity) model calibration problems. DYNO combines the error metrics of multiple variables into a single objective function by dynamically normalizing each variable's error terms using information available during the search. DYNO is proposed to dynamically adjust the weight of the error of each variable hence balancing the calibration to each variable during optimization search. The DYNO is applied to calibrate a tropical hydrodynamic model where temperature and velocity observation data are used for model calibration simultaneously. We also investigated the efficiency of DYNO by comparing the result of using DYNO to results of calibrating to either temperature or velocity observation only. The result indicates that DYNO can balance the calibration in terms of water temperature and velocity and that calibrating to only one variable (e.g., temperature or velocity) cannot guarantee the goodness-of-fit of another variable (e.g., velocity or temperature). Our study suggested that both temperature and velocity measures should be used for hydrodynamic model calibration in real practice. Our example problems were computed with a parallel optimization method PODS but DYNO can also be easily used in serial applications.

NUMERICAL MODELLING AND ASSESSMENT OF THE UGEN FLOATING WAVE ENERGY CONVERTER

The paper presents a linear hydrodynamic model for the UGEN wave energy converter, an analysis of the dynamics of the system and the predicted ability to extract energy from the waves. The UGEN (floating device with a U tank for GENeration of electricity from waves) consists of an asymmetric floater with a large internal U tank filled with water, where the energy is extracted from the relative motion between the water inside the tank and the rolling of the floater. The floater rolling mode of motion is the main stimulator of the motion of the water in the tank, however the sway and heave motions are also coupled therefore the system has motion.

Incorporating terrain specific beaching within a lagrangian transport plastics model for Lake Erie

Mass estimates of plastic pollution in the Great Lakes based on surface samples differ by orders of magnitude from what is predicted by production and input rates. It has been theorized that a potential location of this missing plastic is on beaches and in nearshore water. We incorporate a terrain dependent beaching model to an existing hydrodynamic model for Lake Erie which includes three dimensional advection, turbulent mixing, density driven sinking, and deposition into the sediment. When examining parameter choices, in all simulations the majority of plastic in the lake is beached, potentially identifying a reservoir holding a large percentage of the lake’s plastic which in previous studies has not been taken into account. The absolute amount of beached plastic is dependent on the parameter choices. We also find beached plastic does not accumulate homogeneously through the lake, with eastern regions of the lake, especially those downstream of population centers, most likely to be impacted. This effort constitutes a step towards identifying sinks of missing plastic in large bodies of water.