| An estuary is the transition between a river and a sea, where saltwater intrusion occurs frequently. The region near the estuary is usually densely populated, highly developed and convenient transportation. Drinking water, industry water and agricultural irrigation water must up to the standard. According to the "Standards for drinking water sources quality of China" (CJ3020-93), the chloride content in the drinking water must be less than 250 mg/L. The population of Hangzhou is about 8.7 million, and the drinking water of the city is almost all pumped from the Qiantang River. However, almost all the waterworks of Hangzhou will be subjected to saltwater intrusion during dry season and spring tide every year, which will have a great influence on the lives of local residents.This paper adopted the prototype observation data analysis method to analysis the change of tidal saltwater. By the long time series statistical analysis of discharge and tide impact on saltwater intrusion, we obtained the characters of the saltwater intrusion vary in day, months, seasons, and years, and the way of salinity diminishes with the discharge increases. Besides, the statistical data of salinity content in water above standards and the discharge was used to analyze saltwater intrusion impacted by the discharge and the terrain changes. Moreover, the continuous wavelet transform was used to analyze the annual and interannual time scale variation of saltwater intrusion and the factors. And cross wavelet transform and wavelet coherence were used to exam the periodic change and correlations between salinity and discharge, as well as tidal range. It showed that saltwater intrusion was affected by the factors on different time scales. Through prototype observation data analysis method, we obtained the basic law of saltwater intrusion in Qiantang estuary and the quantitative and qualitave correlation between the factors which impact the saltwater intrusion. These are the cornerstone for further saltwater intrusion research.Currently, the general method used to model the saltwater intrusion are two dimensional or three dimensional numerical models. However, they both requires a lot of calculation and more terrain boundary parameters and hydrological data. Unfortunately, the terrain is constantly changing and the hydrology data is not often enough. In this paper, in order to easily and quickly determine variation of the saltwater intrusion, and then to evaluate its impact on the water intake in time, the high water slack mode of one dimension analysis was adopted to compute the salinity in the Qiantang river and its intrusion length. Saltwater intrusion was computed under different discharge when the tide was fixed, and before and after the river bed deepened. The computed results showed the deepened bed has a great impact on the saltwater intrusion, the influence of saltwater intrusion will increase with the river bed deepening; While salinity will rapidly decrease with discharge increasing, when the discharge incteases to a certain extent, the salinity no longer apparently decrease. This method can quickly simulate saltwater intrusion in future, additionally, the parameters required in this mode is less and more simple calculation. These advantages suggest that saltwater intrusion can be predicted in time which will promote a more rational scheme to pump water in advance.The daily maximum salinity time series at Zhakou, Qibao and Cangqian hydrological station along the Qiantang River were modeled and predicted 5 days ahead using the wavelet multi-resolution analysis and the NARX (Nonliner autoregressive model with exogenous inputs) model, which according to the relationship between the factors and the salinity, in particular to these salinity forecast for the observation locations. For the non-linear, non-stationary and seasonal pattern of the salinity and factors time series, it cannot be recognized simply by ANN networks. So the MODWT (Maximal Overlap Discrete Wavelet Transform) based on the multi-resolution was used to decompose the hydrology data, then decomposed J0+1 sub-signals were modeled using NARX neural networks, the predictions for the different sub-signals were summed up to obtain the final prediction. Results from the comparison of different models showed that the best performance in terms of the performance criteria was obtained by the WT-mNARX (Wavelet method coupled with multi-nonliner autoregressive model with exogenous inputs) model. Further more, the WT-mNARX model performance outside the domain illustrated it has good generalization ability. The salinity was estimated much more accurately by the multivariable WT-mNARX model, which is a useful tool for nonlinear and non-stationary time series forecasting. It can achieved a rapid alert on the Qiantang River estuary saltwater intrusion which has an important practical significance for guarantee the water source quality security in Qiantang River estuary. |
