Study On Driving Factors And Simulation Prediction Of Sediment Transport In Typical Watershed In Ningxia

Soil erosion and loss of water and soil are among the most pressing environmental concerns in China.The degradation of water and soil resources has worsened erosion and sediment transport in river basins,particularly in recent decades.The increasing frequency of extreme weather events,as a result of global climate chan ge,has also impacted sediment transport.Meanwhile,human activities such as river regulation and ecological construction measures like returning farmland to forests have further affected the underlying surface conditions of the river basin.This study selected the typical tributaries of the Yellow River in Ningxia,the Kushui River Basin and the Qingshui River Basin,as research objects.It analyzed the interannual variations of meteorological factors,runoff sediment transport,and ecological construction measures in the two basins,and identified the driving factors of sediment transport changes.A machine learning model was established to simulate and predict sediment transport in the basins.Additionally,a coupling contribution rate decomposition method based on machine learning algorithms was developed to analyze the contribution of water and soil conservation measures to sediment transport changes.Finally,sediment transport in the Kushui River Basin and the Qingshui River Basin from 2020 to 2050 was predicted and analyzed under four emission scenarios,namely,SSP1-2.6,SSP2-4.5,SSP3-7.0,and SSP5-8.5,using the BCC-CMS2-MR future climate model in CMIP6.The main research achievements are as follows:(1)The Mann-Kendall trend test,Pettitt change point test,and Pearson correlation analysis methods were used to analyze the inter-annual variation trends and correlations of underlying surface and hydro-meteorological factors in the Kushui River Basin and the Qingshui River Basin.The annual precipitation in the Kushui River Basin showed a non-significant increasing trend,while all extreme precipitation indices,except for P25 and R25,showed a non-significant increasing trend.The change points of meteorological factors were not significant.In the Qingshui River Basin,meteorological factors showed no significant increasing trend,and the change points were not significant.The inter-annual variation of meteorological factors in the basin was relatively large.As for the annual runoff,both the Kushui and Qingshui River Basins showed a significant increasing trend,with significant change points in 1992 and 1993,respectively.The annual sediment transport showed a non-significant decreasing trend,with significant change points in 2002 and 2007.After the change points,the sediment transport sharply decreased,indicating that soil and water conservation measures in the basin had achieved initial results,effectively reducing sediment transport and preventing major soil erosion in the basin.The correlation analysis showed that all soil and water conservation measures were negatively correlated with sediment transport in both basins,indicating that these measures had a reducing effect on sediment transport.The correlation coefficients between annual runoff and sediment transport in the two basins were 0.695 and 0.56,respectively,showing a significant positive correlation,indicating that the change in sediment transport in the basin was mainly affected by annual runoff.(2)Four machine learning models were used to construct sediment transport simulation and prediction models in both the Kushuihe River Basin and the Qingshui River Basin.The Shapley value method was used to extract and analyze the interpretability of feature factors in the model,and a machine learning-based sediment reduction contribution rate decomposition method for soil and water conservation measures was constructed to extract the contribution rate of various soil and water conservation measures to sediment transport change.In the simulation results of machine learning models in each basin,the simulated values can maintain consistency with the trend of actual measured data in each basin.Through evaluation and comparison using NSE,MRE,and RMSE,it was found that the SVR models in the Kushuihe River Basin and Qingshui River Basin can achieve better results under extreme climate index input scenarios(F2 feature scenario).The Shapley value method showed that annual runoff and three extreme climate indices in both basins have a positive impact on sediment transport,while all soil and water conservation measures have a negative impact.By calculating the contribution rate based on the machine learning contribution rate decomposition method,the contribution rates of terraced fields,forests,grasslands,closure management,and dam area to sediment transport change in the Kushuihe River Basin were 16.52%,2.08%,15.80%,0.54%,and 3.88%,respectively,and the coupling contribution rate of soil and water conservation measures to sediment transport change was 14.99%.In the Qingshui River Basin,the contribution rates of terraced fields,forests,grasslands,closure management,and dam area to sediment transport change were 34.96%,20.30%,17.35%,11.76%,and 8.34%,respectively,and the coupling contribution rate of soil and water conservation measures to sediment transport change was 4.82%.The coupling contribution rate of soil and water conservation measures to sediment transport in both basins was greater than 0,indicating that the comprehensive effect of soil and water conservation measures can play a more effective role in reducing sediment transport in the basin than implementing them separately.(3)Based on the constructed machine learning models,the sediment transport of the Kushui River basin and the Qingshui River basin from 2020 to 2050 was simulated and predicted using CMIP6 climate data.The simulation and prediction results of sediment transport in the Kushui River basin and the Qingshui River basin from 2020 to 2050 show that the comprehensive deployment of measures can significantly reduce the sediment transport of the basin under changing environmental conditions.In the Kushui River basin,the predicted order of sediment reduction for each measure is:grassland>forest land>terraced fields>check dam>closure management.In the Qingshui River basin,the predicted order of sediment reduction for each soil and water conservation measure is:closure management>grassland>terraced fields>forest land>check dam.