A high-resolution comprehensive water quality model based on GPU acceleration techniques

[Display omitted] •A high-resolution comprehensive water quality model based on the GPU acceleration techniques was developed.•Complex mechanisms of water environment and ecology were simulated efficiently and accurately based on physical processes.•The model was suitable for high resolution rivers with large area and complex terrain.•Our model has high simulation accuracy for both processes of pure pollutant reaction and transport and reaction.•On the adopted RTX3070d computer, only 0.12 h was taken to simulate the hydrodynamic and water quality evolution process involving 350.000 cells with 5-h model settings. The study of the temporal and spatial evolution of pollutants in modern water bodies and river water quality management largely depend on river water quality models. These models are usually constructed based on detailed hydrodynamic models but cannot satisfy the high precision and efficiency requirement of model simulations in high-resolution terrain. To solve this problem, we developed a high-resolution comprehensive water quality model based on the graphics processing unit (GPU) acceleration techniques in which hydrodynamics and accompanying contaminants were perfectly coupled. The performance and functionality of the model were improved by using the Compute Unified Device Architecture (CUDA) parallel computing architecture and robust model algorithms. A simulation function of the transport attenuation and interaction among multicomponent pollutants was stablished by solving mass conservation, momentum and chemical concentration equations of the flow process. The simulation results of the pure pollutant reaction process and the transport and reaction process were validated against an analytical solution and the results of a two-dimensional steady water quality model by using the Nash-Sutcliffe efficiency (NSE) coefficient. The coefficients were 0.996 and 0.984, respectively. The water quality of the Xidagou River in Yinchuan was evaluated and analyzed by using the validated high-resolution comprehensive water quality model, and the numerical results of two-dimensional pollutant transport were explored. On the adopted RTX3070d computer, only 0.12 h were needed to simulate the hydrodynamic and water quality evolution process involving 350.000 cells with 5-h model settings. The model achieved a high operation speed and high calculation accuracy and could effectively simulate, predict and evaluate various complex water environment problems. This stu