Monitoring and Water and Sediment based on Linear Regression and ARIMA Models

Affected by climate change and human activities, the water and sediment conditions of most rivers in the world have changed significantly. This study takes a hydrological station as the object to explore the monitoring and prediction of water and sediment under the influence of climate change and human activities. Through the multiple linear regression model, the relationship between sediment concentration and time, water level and flow is constructed, and the least square method is used to predict the sediment concentration in the past 6 years. With the help of sliding window and seasonal decomposition, the variation law of water and sediment flux is analyzed, and the ARIMA model combined with genetic algorithm is used to optimize the sampling scheme and predict the trend of water and sediment flux in the next two years. The assessment shows that after the implementation of the " water and sediment regulation " measures from June to July, the water and sediment flux decreased significantly, effectively maintaining the sediment transport and navigation capacity of the river. It is found that the sediment concentration is positively correlated with water level and flow rate, and the prediction error of ARIMA model is less than 5 %. This study provides method support for water and sediment monitoring, confirms the effect of water and sediment regulation on reducing sediment deposition, and suggests that deep learning and remote sensing technology should be combined to improve prediction accuracy in the future.