To reveal the spatiotemporal evolution patterns and influencing factors of backwater effect of runoff in the lower reaches of the Ganjiang River,a hydrological runoff model suitable for plain river network areas was constructed and spatiotemporally coupled with a self-developed two-dimensional hydrodynamic model.The coupled hydrological-hydrodynamic model was adopted to simulate the backwater effect in the lower reaches of the Ganjiang River.Based on the simulation results,a water level difference prediction model was further established through multivariate regression analysis.The results show that the backwater effect in the lower reaches of the Ganjiang River is closely related to the discharge at Waizhou Station.During low-flow periods,the backwater in the north branch elevates the water level at Nanchang Station,while when the discharge increases,the spiral flow in the main branch propagates downstream,generating a backwater effect between Changyi and Nanchang stations.When the discharge exceeds the critical threshold of 513 m3/s,the water level of Poyang Lake is higher than that of the lower reaches of the Ganjiang River,initiating a backwater effect from the lake onto the river.When the discharge further increases to 3 211 m3/s,the backwater effect dissipates,and the flow diversion between the East River and West River tends to stabilize.The rainfall-induced runoff influences the evolutionary process of the generation and recession of the backwater effect,with a retardation effect ranging from 12%to 19%.After incorporating a correction term of rainfall-induced runoff,the precision of the water level difference prediction model is improved.
The main producing areas of Chinese baijiu are mostly distributed in artesian basins,and the groundwater recharge for brewing water exhibits distinct geological characteristics.The contents of minerals and rare earth elements in water body are the primary factors contributing to the regional differences in liquor taste.To investigate the key components and source of liquor-brewing water,multiple geochemical approaches were adopted to conduct hydrochemical analysis,water volume calculation,and hydrogen-oxygen isotopic analysis on surface water and groundwater in the Chishui River Basin.The results show that characteristic mineral components such as strontium,sulfur,and calcium in brewing water originate from water-rock reactions between groundwater and surrounding rock during groundwater migration.More than half of water volume in the Chishui River is supplied by exogenous water,which serves as the main recharge source during dry season;spring recharge area is derived from river seepage water in the Qinghai-Xizang Plateau.The midstream strata of Chishui River host carbonate rock formations containing celestite,barite,and gypsum,which provide sources of strontium,sulfur,calcium,and other substances to water body.It is inferred that seepage water from the Qinghai-Xizang Plateau dissolves minerals such as strontium,calcium,and sulfur from carbonate rocks during transport and converges into Chishui River.The river water is rich in unique minerals and rare earth elements required for brewing high-quality liquors such as Moutai,Xijiu,and Langjiu,endowing Chishui River with its distinctive status as the"river of fine liquors".
Regarding the issue of insufficient attention to artificial intelligence methods in current evaluation of water resources intensive and safe utilization,an evaluation index system for water resources intensive and safe utilization was established.The nutcracker optimization algorithm(NOA)was employed to optimize the parameters of the support vector machine(SVM)model,leading to the construction of an NOA-SVM model for the evaluation of intensive and safe utilization of water resources.By altering the proportions of training set and testing set,the evaluation result of the NOA-SVM model was compared with that of the standard SVM model,and the performance of the NOA-SVM model under different kernel functions was analyzed and compared.Furthermore,an empirical analysis was conducted on the 31 provinces(autonomous regions and municipalities)of China excluding Hongkong,Macao and Taiwan,to verify the effectiveness of the model.The results indicate that the NOA-SVM model achieves a higher evaluation accuracy than the conventional SVM model in the evaluation of the intensive and safe utilization of water resources.In particular,the radial basis function kernel maintains a stable goodness-of-fit under different proportions of the training set,and its corresponding error metrics including MSE,RMSE,and MAE are all desirable and superior to those of the linear kernel and polynomial kernel.The top seven provinces(autonomous regions and municipalities)in the evaluation results of intensive and safe utilization of water resources are Tianjin,Hebei,Beijing,Henan,Shandong,Shanxi,and Inner Mongolia,and their core advantages lie in controlled development intensity,improved water use efficiency,and well-adapted industrial structure.The bottom seven provinces(autonomous regions,municipalities)are Jiangxi,Jiangsu,Hubei,Shanghai,Hainan,Guangxi,and Xizang,and their core issues include excessive resource development,low water use efficiency,and insufficient ecological water use.
To address the difficulty in runoff forecasting in alpine areas,a runoff forecasting method coupling the iRainSnowHydro model and weighted Markov chain was proposed.The iRainSnowHydro model considered the regulation and storage of snowmelt water by constructing a snow reservoir,precisely depicting the process of snowmelt water outflows from wet snow.The method also introduced a weighted Markov chain to capture the transmission and transfer rules of simulation errors,achieving real-time error correction.The forecasting method was applied to the Wunonglong Basin in the upper reaches of the Lancang River.The results show that the determination coefficient and Nash efficiency coefficient in the calibration and validation periods were both over 0.8,and the absolute value of the percentage bias was less than 10%,indicating that the iRainSnowHydro model has a good simulation effect,especially during the snowmelt period(March to May).The weighted Markov chain effectively corrects the systematic underestimation of summer runoff by the model,reducing the average relative error from 15.68%to 7.52%and significantly improving the forecasting accuracy.
To capture the dynamic variation laws of hydrological processes and improve runoff prediction accuracy,by combining three kinds of correlation analysis with covariance analysis,we systematically analyzed the correlation characteristics between different meteorological variables and runoff,dynamically extracted high-contribution features for runoff prediction and optimized their weights.Then,based on the deep temporal convolutional network(DeepTCN),a p-DeepTCN model for multi-feature runoff prediction was constructed to capture both the long-term dependencies and short-term fluctuations in runoff sequence.The measured data of the Atsuma River Basin in Hokkaido from 2015 to 2020 were used to verify the p-DeepTCN model,and the results show that the p-DeepTCN model achieves the optimal prediction performance compared with other deep learning models.Specifically,the Nash-Sutcliffe efficiency coefficient of the p-DeepTCN model is increased by 28.48%,and the root mean square error is reduced by 27.63%.Although its percent bias is 23.34%higher than that of the WPMixer model,it is 17.44%lower than that of the other deep learning models.Meanwhile,although its coefficient of determination is 0.011 lower than that of the original DeepTCN model,it is 10.17%higher than that of other deep learning models.In summary,the p-DeepTCN model significantly improves the accuracy of runoff prediction,and exhibits superior adaptability and robustness.
