An operational ensemble numerical prediction method and system for estuary saltwater intrusion

Abstract

The present invention discloses a method for operational ensemble numerical prediction of estuarine saltwater intrusion, comprising: constructing a estuarine saltwater intrusion numerical model and setting relevant parameters of the estuarine saltwater intrusion numerical model; calculating and processing the bottom friction parameters of the estuarine area, and calibrating the estuarine saltwater intrusion numerical model based on the calculation results; using optimal interpolation data assimilation technology to perform data assimilation processing on the estuarine saltwater intrusion numerical model; constructing a runoff ensemble sample based on the measured runoff in the estuarine area; and inputting the runoff ensemble sample into the estuarine saltwater intrusion numerical model for calculation and processing to obtain a numerical prediction result. The present invention can predict the intensity and duration of saltwater intrusion with high prediction accuracy, can guide reservoir operation, ensure normal water supply during saltwater intrusion, and guarantee the safety of freshwater resource extraction.

Description

Technical Field

[0001] The present invention relates to the technical field of estuary data calculation methods, and in particular to an operational ensemble numerical prediction method for estuary saltwater intrusion and a system thereof. Background Art

[0002] Saltwater intrusion in estuaries is a common phenomenon in estuaries. It is affected by the combined effects of tides, runoff, wind stress, mixing, topography and extraestuarine circulation. The dynamic process is complex and is closely related to estuary circulation, sediment flocculation, maximum turbidity zone and ecological environment.

[0003] For example, the Yangtze River estuary is a large, multi-branched estuary. Saltwater intrusion is driven by a variety of dynamic factors, resulting in the unique phenomenon of saltwater backflow from the northern branch. The Yangtze River estuary region boasts a developed economy and dense population, creating a high demand for freshwater resources. Before 2010, Shanghai's water supply primarily came from the Huangpu River, accounting for approximately 80% of its total water use. However, this water quality was poor and the water supply was insufficient. After 2010, the Qingcaosha Reservoir was completed, enabling a strategic shift in Shanghai's water source. Currently, the Qingcaosha, Chenhang, and Dongfengxisha Reservoirs in the Yangtze River estuary supply over 80% of Shanghai's total water supply. Water withdrawal from the Yangtze River estuary presents the risk of saltwater intrusion, making numerical forecasting of saltwater intrusion crucial for the operation of reservoirs in these water sources and ensuring water security.

[0004] Methods for studying saltwater intrusion in estuaries primarily include physical models, field observations, theoretical analysis, and numerical simulations. International research on saltwater intrusion began with field observations at the U.S. Waterways Experiment Station (WES) in 1929. Simultaneously, the Delft Hydraulics Institute in the Netherlands conducted extensive work analyzing field data and conducting experimental studies. Systematic research on saltwater intrusion began in the 1950s, focusing primarily on the fundamental theory of saltwater intrusion in estuaries. Pritchard (1952), analyzing observational saltwater intrusion data from Chesapeak Bay and the James River estuary, comprehensively expounded on the dynamic mechanisms of longitudinal estuarine circulation and derived the classic equilibrium relationship between estuarine circulation and salinity. Schijf and Schonfeld (1953) derived an analytical solution for the length of saltwater intrusion in estuaries based on saltwater wedge theory. Bowden (1966) used field data to study the longitudinal velocity distribution characteristics and salinity mixing processes in estuaries, and also analyzed the longitudinal circulation in the Mersey estuary. Saltwater intrusion in estuaries is closely related to vertical mixing, which in turn is related to estuary circulation and type. Hansen and Rattray (1966) derived an empirical formula for the vertical mixing coefficient based on measured data and used dimensionless parameters to describe estuaries, resulting in a classification of stratified-circulation estuaries. Subsequently, various scholars applied different criteria to categorize saltwater intrusion types. Pritchard (1967) classified estuaries into three types: highly stratified, partially mixed, and fully mixed, based on the vertical mixing characteristics of freshwater and saltwater in estuaries. Simmons and Brown (1969) proposed the concept of a mixing index, classifying estuaries into highly stratified, partially mixed, and vertically uniform types. Officer (1975) classified estuaries into weakly mixed, partially mixed, and strongly mixed types. MacCready (2004) derived the equilibrium relationship between salinity advection and diffusion in an ideal estuary, noting that vertical circulation, where bottom flow enters and surface flow exits, can increase vertical differences in salinity, while vertical mixing tends to reduce these differences. Prandle (2004) conducted a detailed study of saltwater upwelling and tidal intrusion peaks in some mixed estuaries. Increased vertical mixing leads to a weakening of vertical circulation, which in turn weakens stratification. Simultaneously, this weakening of vertical circulation reduces the salt flux into the estuary, thereby shortening the length of the saltwater intrusion (Park and Kuo, 1996; MacCready, 2004). In multi-branched estuaries, lateral exchange between channels can affect the longitudinal saltwater intrusion (Conroy et al., 2019). Seasonal variations in winds and the passage of storms influence saltwater intrusion in estuaries through a combination of barotropic and barotropic processes.

[0005] With the development of computer technology and numerical calculation methods, models related to saltwater intrusion began to develop in the 1980s, and a large number of scholars have used numerical simulation methods to study the problem of saltwater intrusion. Currently, the most widely used ocean and estuarine numerical models in the world include POM, ECOM, FVCOM, ROMS and UNTRIM, as well as commercial software such as Delft-3D and MIKE model, which have played an important role in the study of estuarine hydrodynamics and material transport and diffusion. With the improvement of observation technology and the widespread application of numerical models, the study of saltwater intrusion in estuaries has achieved many results in understanding the dynamic processes and mechanisms.

[0006] Research on saltwater intrusion at estuaries in my country began in the 1980s, primarily focusing on the Yangtze and Pearl River estuaries. The Yangtze and Pearl Rivers are my country's first and second largest rivers in terms of runoff. Their estuaries host megacities with dense populations, resulting in enormous water demands for both domestic and industrial use, and both are home to large reservoirs. Previous research has shown that tidal strength and runoff volume are the two primary drivers of saltwater intrusion at the Yangtze River estuary, and numerous findings have been published. The most prominent characteristic of saltwater intrusion at the Yangtze River estuary is the backflow of saltwater from the North Branch into the South Branch. The primary driving mechanism for this intrusion is the unique topography of the North Branch's bell-shaped mouth, its small diversion ratio, the upwelling of tidal levels at Qinglonggang in the upper section, and tidal residual flow. This backflowing saltwater is transported downstream by runoff, gradually impacting the Dongfengxisha, Chenxingshui, and Qingcaosha Reservoirs downstream of the South Branch. Wu et al. (2006) used numerical models to quantitatively investigate the relationship between the intensity of water backflow in the North Branch and upstream runoff and the tidal range at Qinglonggang. They also derived relationships between the daily backflow volume and intensity, as well as the bimonthly backflow volume and runoff. Zhu Jianrong et al. (2011) calculated the salt flux of saltwater backflow in the North Branch under different runoff volumes. Lyu and Zhu (2018) proposed that the bottom friction drag coefficient, which uses the quadratic law of velocity in the bottom boundary layer, is not suitable for the ultra-shallow North Branch channel. They proposed a bottom friction drag coefficient based on the Xie Cai-Manning coefficient, significantly improving the accuracy of saltwater backflow calculations in the North Branch.

[0007] Saltwater intrusion at the Yangtze River Estuary is also affected by wind stress. In winter, northerly winds generate onshore Ekman transport, exacerbating saltwater intrusion. Climate change is causing changes in basin precipitation and sea level rise. Sea level rise has generally increased the depth of the estuary, leading to increased saltwater intrusion at the Yangtze River Estuary. The evolution of estuary flow, including long-term natural evolution and short-term local engineering projects, has affected hydrodynamics and diversion ratios, which in turn have an impact on saltwater intrusion. Human activities such as water conservancy projects in the basin, deep-water channel projects at the estuary, and mudflat enclosure projects have affected the hydrodynamics and material transport at the estuary. Basin engineering projects affect saltwater intrusion and freshwater resources at the Yangtze River Estuary by changing runoff, while estuary engineering projects mainly affect local topography.

[0008] With the continuous development of numerical formats, the computational accuracy of numerical models has been continuously improved, playing an important role in the research and numerical prediction of saltwater intrusion in estuaries. Wu and Zhu (2010) used the third-order accurate HSIMT numerical calculation method to improve the numerical format of the advection term in the salinity transport and diffusion equation, eliminating the computational dispersion and significantly reducing the numerical dissipation.

[0009] The applicant has long conducted on-site observations of estuaries, improving and applying three-dimensional numerical models and theoretical analysis of estuaries, and has achieved numerous research results in the study of estuary hydrodynamics and saltwater intrusion. These achievements have played an important role in reservoir construction (determining the maximum time when water withdrawal is not suitable, and designing storage capacity) and operation (numerical prediction of saltwater intrusion at estuaries). Therefore, given the problem of saltwater intrusion encountered during water withdrawal from estuaries, it is urgent to develop numerical predictions of saltwater intrusion, which has important application significance for the operation of water source reservoirs and ensuring water safety.

[0010] Therefore, the applicant has found a solution to the above-mentioned problem through beneficial exploration and research. The technical solution to be introduced below is produced in this context. Summary of the Invention

[0011] The technical problem to be solved by the present invention is to provide an operational ensemble numerical prediction method for estuary saltwater intrusion based on data assimilation to address the shortcomings of the existing technology, so as to guide reservoir operation, ensure normal water supply during saltwater intrusion, and protect the safety of freshwater resource extraction.

[0012] The second technical problem to be solved by the present invention is to provide a system for realizing the above-mentioned operational ensemble numerical forecasting method for saltwater intrusion in estuaries.

[0013] As a first aspect of the present invention, a method for operational ensemble numerical prediction of estuary saltwater intrusion comprises:

[0014] Constructing a numerical model of saltwater intrusion at the estuary and setting relevant parameters of the numerical model;

[0015] Calculating and processing the bottom friction parameters of the estuary area, and calibrating the estuary saltwater intrusion numerical model according to the calculation results;

[0016] The optimal interpolation data assimilation technology is used to perform data assimilation processing on the estuary saltwater intrusion numerical model;

[0017] Constructing a runoff aggregate sample based on measured runoff in the estuary area; and

[0018] The runoff aggregate samples are input into the estuary saltwater intrusion numerical model for calculation and processing to obtain numerical prediction results.

[0019] In a preferred embodiment of the present invention, the steps of constructing a numerical model of saltwater intrusion at the estuary and setting relevant parameters of the numerical model of saltwater intrusion at the estuary include:

[0020] A numerical model of saltwater intrusion in estuaries was constructed, taking into account dynamic factors such as runoff, tides, wind stress, and bottom friction. A non-orthogonal curved grid was used horizontally and σ coordinates were used vertically.

[0021] Set the calculation area and perform local encryption on the grid in key areas;

[0022] Set vertical stratification, time step and critical water depth;

[0023] Set initial field values ​​such as water level field, flow field and salinity field; and

[0024] Set boundary conditions such as tide level, residual water level, runoff and wind field.

[0025] In a preferred embodiment of the present invention, the calculating and processing of the bottom friction parameters of the estuary area and the calibration processing of the estuary saltwater intrusion numerical model according to the calculation results include:

[0026] The bottom friction parameters of the estuary area include the Manning roughness coefficient and the bottom friction drag coefficient. The Manning roughness coefficient of the estuary area is assigned according to Table 1:

[0027] Table 1 Values ​​of Manning's roughness coefficient

[0028] Water depth D(m) Manning's roughness D<2 0.030 2≤D<3 0.024 3≤D<10 0.014 D≥10 0.010

[0029] Calculate the Xiecai coefficient C according to formula (1):

[0030]

[0031] Where H is the water depth, n is the Manning roughness coefficient;

[0032] The bottom friction drag coefficient C in the estuary area is calculated according to formula (2): d :

[0033]

[0034] Where C is the Xie Cai coefficient, g is the acceleration due to gravity;

[0035] The Manning roughness coefficient and bottom friction drag coefficient of the estuary area are used to calibrate the estuary saltwater intrusion numerical model.

[0036] In a preferred embodiment of the present invention, the optimal interpolation data assimilation technology is used to perform data assimilation processing on the estuary saltwater intrusion numerical model, including:

[0037] Select the actual measurement sites for data assimilation. If the sites in some areas are too sparse, set up virtual sites.

[0038] Conduct quality control on the measured data of selected sites;

[0039] The optimal interpolation method shown in formula (3) is used to assimilate data and produce the initial field of the model;

[0040]

[0041] Among them, x a and x b are the analysis field and background field, respectively, K is the optimal weight, y is the observation data, H is the linear operator for interpolating the background field value to the observation position, the superscript T represents the transpose of the matrix, B is the background field error covariance matrix, and O is the observation field error covariance matrix.

[0042] In a preferred embodiment of the present invention, the quality control includes format inspection, relevance inspection, missing inspection, range inspection, continuity inspection, visual graphics drawing inspection and comprehensive analysis inspection.

[0043] In a preferred embodiment of the present invention, constructing a runoff sample set based on the measured runoff in the estuary area includes:

[0044] To construct a short-term runoff prediction sample, the measured runoff volume in the estuary area is subtracted from the drainage volume along the way. The specific value is determined according to formula (4) and Table 2:

[0045] y=ax+b (4)

[0046] Among them, y is the runoff input of the model, x is the measured flow in the estuary area, and a and b are adjustment coefficients, which mainly reflect the drainage along the way;

[0047] Table 2 Model runoff input adjustment coefficients

[0048] Flow value a b <![CDATA[>=20000m 3 / s]]> 1.0 0 <![CDATA[>12000m 3 / s,<20000m 3 / s]]> 1.0 -700 <=12000m3 / s 1.0 -2000

[0049] The runoff prediction samples are constructed based on the forecast information of the authoritative agencies. If there is no forecast information, the runoff prediction samples are constructed based on experience. The prediction value is constructed based on the last measured value or the average value of the most recent period. The specific value is determined according to formula (5) and Table 3:

[0050] y=ax+b (5)

[0051] Where y is the runoff input of the model, x is the latest observed value of the measured flow in the estuary area, and a and b are adjustment coefficients, which mainly reflect the change of continuity;

[0052] Table 3 Model runoff input adjustment coefficients

[0053] Flow value a b >=20000m3 / s 1.0 200 >12000m3 / s,<20000m3 / s 1.0 -700 <=12000m3 / s 1.0 -2200

[0054] If there is no measured data for empirical construction, the climatological state value is used to construct the runoff sample value;

[0055] Construct a runoff aggregate sample, and generate multiple runoff prediction samples according to the above by modifying the measured runoff value to form a runoff aggregate sample.

[0056] As a second aspect of the present invention, a system for implementing the above-mentioned operational ensemble numerical prediction method for estuary saltwater intrusion comprises:

[0057] A model construction module, wherein the model construction module is used to construct a numerical model of estuarine saltwater intrusion and set relevant parameters of the numerical model of estuarine saltwater intrusion;

[0058] A parameter calibration module, wherein the parameter calibration module is used to calculate the bottom friction parameters of the estuary area and calibrate the estuary saltwater intrusion numerical model according to the calculation results;

[0059] A data assimilation module, wherein the data assimilation module is used to perform data assimilation processing on the estuary saltwater intrusion numerical model by using an optimal interpolation data assimilation technology;

[0060] a sample construction module, the sample construction module being used to construct a runoff aggregate sample based on the measured runoff in the estuary area; and

[0061] The numerical prediction module is used to input the runoff set samples into the estuary saltwater intrusion numerical model for calculation and processing to obtain numerical prediction results.

[0062] Due to the adoption of the above technical solution, the beneficial effect of the present invention is that: the present invention generates data forecast results by constructing a set of runoff samples and inputting them into a numerical model of estuary saltwater intrusion based on data assimilation and with bottom friction coefficient determined, and predicts the intensity and duration of saltwater intrusion with high prediction accuracy, which can guide reservoir operation, ensure normal water supply during saltwater intrusion, and protect the safety of freshwater resource extraction. BRIEF DESCRIPTION OF THE DRAWINGS

[0063] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0064] Figure 1 The present invention is a flow chart of the operational ensemble numerical prediction method for estuary saltwater intrusion.

[0065] Figure 2 It is a schematic structural diagram of the system of the present invention.

[0066] Figure 3 It is a schematic diagram of the model calculation area, where a is the model calculation area and grid, b is the enlarged grid of the north-south branch bifurcation area, and c is the grid of the north-south trough bifurcation area.

[0067] Figure 4 This is a schematic diagram of the measured water depth distribution of the Yangtze River estuary in 2021 (Yellow Sea 85 datum).

[0068] Figure 5 This is a schematic diagram comparing the salinity simulation results of Chongxi and Nanmen hydrological stations from March 2 to 20, 2017.

[0069] Figure 6 This is a schematic diagram of the location of Wusong Water Plant Station.

[0070] Figure 7 This is a location diagram of Pudong Airport Station.

[0071] Figure 8 It is the salinity data quality control process.

[0072] Figure 9 is a schematic diagram of the runoff collection sample construction. DETAILED DESCRIPTION

[0073] In order to make the technical means, creative features, objectives and effects achieved by the present invention easier to understand, the present invention is further described below with reference to specific illustrations.

[0074] See also Figure 1 The figure shows an operational ensemble numerical prediction method for saltwater intrusion in estuaries, which includes the following steps:

[0075] Step S10: constructing a numerical model of estuary saltwater intrusion and setting relevant parameters of the numerical model of estuary saltwater intrusion.

[0076] Step S20 , calculating and processing the bottom friction parameters of the estuary area, and calibrating the estuary saltwater intrusion numerical model based on the calculation results.

[0077] Step S30: using the optimal interpolation data assimilation technology to perform data assimilation processing on the estuary saltwater intrusion numerical model.

[0078] Step S40: constructing a runoff aggregate sample based on the measured runoff in the estuary area.

[0079] Step S50: Input the runoff sample set into the estuary saltwater intrusion numerical model for calculation and processing to obtain a numerical prediction result.

[0080] In step S10, a numerical model of estuarine saltwater intrusion is constructed, and relevant parameters of the numerical model of estuarine saltwater intrusion are set, including the following sub-steps:

[0081] Step S11, constructing a numerical model of saltwater intrusion in the estuary, taking into account dynamic factors such as runoff, tide, wind stress and bottom friction, using a non-orthogonal curved grid horizontally and a σ coordinate vertically;

[0082] Step S12: setting the calculation area and performing local encryption processing on the grid of the key area;

[0083] Step S13, setting vertical stratification, time step and critical water depth;

[0084] Step S14, setting initial field values ​​such as water level field, flow field and salinity field;

[0085] Step S15, setting boundary conditions such as tide level, residual water level, runoff and wind field.

[0086] In step S20, the bottom friction parameters of the estuary area are calculated and the estuary saltwater intrusion numerical model is calibrated according to the calculation results, which includes the following sub-steps:

[0087] In step S21, the bottom friction parameters of the estuary area include the Manning roughness coefficient and the bottom friction drag coefficient. The Manning roughness coefficient of the estuary area is assigned according to Table 1:

[0088] Table 1 Values ​​of Manning's roughness coefficient

[0089] Water depth D(m) Manning's roughness D<2 0.030 2≤D<3 0.024 3≤D<10 0.014 D≥10 0.010

[0090] Step S22, calculate the coefficient C according to formula (1):

[0091]

[0092] Where H is the water depth, n is the Manning roughness coefficient;

[0093] Step S23, calculate the bottom friction drag coefficient C of the estuary area according to formula (2) d :

[0094]

[0095] Where C is the Xie Cai coefficient, g is the acceleration due to gravity;

[0096] Step S24 , calibrating the estuary saltwater intrusion numerical model using the Manning roughness coefficient and bottom friction drag coefficient of the estuary area.

[0097] In step S30, the optimal interpolation data assimilation technology is used to perform data assimilation processing on the estuary saltwater intrusion numerical model, including the following sub-steps:

[0098] Step S31: Select the measured sites for data assimilation. If the sites in some areas are too sparse, set up virtual sites.

[0099] Step S32: Perform quality control on the measured data of the selected site. The quality control includes format inspection, correlation inspection, missing data inspection, range inspection, continuity inspection, visualization graph drawing inspection, and comprehensive analysis inspection.

[0100] Step S33, using the optimal interpolation method shown in formula (3) to perform data assimilation and produce the model initial field;

[0101]

[0102] Among them, x a and x b are the analysis field and background field, respectively, K is the optimal weight, y is the observation data, H is the linear operator for interpolating the background field value to the observation position, the superscript T represents the transpose of the matrix, B is the background field error covariance matrix, and O is the observation field error covariance matrix.

[0103] In step S40, based on the measured runoff in the estuary area, a runoff aggregate sample is constructed, which includes the following sub-steps:

[0104] Step S41: construct a short-term runoff prediction sample by subtracting the drainage volume along the river from the measured runoff volume in the estuary area. The specific value is determined according to formula (4) and Table 2:

[0105] y=ax+b (4)

[0106] Among them, y is the runoff input of the model, x is the measured flow in the estuary area, and a and b are adjustment coefficients, which mainly reflect the drainage along the way;

[0107] Table 2 Model runoff input adjustment coefficients

[0108] Flow value a b <![CDATA[>=20000m 3 / s]]> 1.0 0 <![CDATA[>12000m 3 / s,<20000m 3 / s]]> 1.0 -700 <=12000m3 / s 1.0 -2000

[0109] Step S42: construct the runoff prediction sample. The construction method is consistent with step S41. If there is no forecast information, the runoff prediction sample is constructed empirically. The prediction value is constructed based on the last measured value or the average value of the most recent period. The specific value is determined according to formula (5) and Table 3:

[0110] y=ax+b (5)

[0111] Where y is the runoff input of the model, x is the latest observed value of the measured flow in the estuary area, and a and b are adjustment coefficients, which mainly reflect the change of continuity;

[0112] Table 3 Model runoff input adjustment coefficients

[0113] Flow value a b >=20000m3 / s 1.0 200 >12000m3 / s,<20000m3 / s 1.0 -700 <=12000m3 / s 1.0 -2200

[0114] If there is no measured data for empirical construction, the climatological value (multi-year average) is used to construct the runoff sample value;

[0115] Step S43: constructing a runoff aggregate sample by modifying the measured runoff value and generating a plurality of runoff prediction samples according to the above method to form a runoff aggregate sample.

[0116] The present invention constructs a set of runoff samples and inputs them into a numerical model of estuary saltwater intrusion based on data assimilation and with bottom friction calibration completed for calculation and processing, thereby generating data forecast results and predicting the intensity and duration of saltwater intrusion. The prediction has a high accuracy rate and can guide reservoir operation, ensure normal water supply during saltwater intrusion, and guarantee the safety of freshwater resource extraction.

[0117] See also Figure 2 The figure shows a system for realizing the above-mentioned operational ensemble numerical prediction method for estuary saltwater intrusion, including a model construction module 100, a parameter calibration module 200, a data assimilation module 300, a sample construction module 400 and a numerical prediction module 500.

[0118] The model construction module 100 is used to construct a numerical model of estuarine saltwater intrusion and set the relevant parameters of the numerical model. The parameter calibration module 200 is used to calculate the bottom friction parameters of the estuarine area and calibrate the numerical model of estuarine saltwater intrusion based on the calculation results. The data assimilation module 300 is used to use the optimal interpolation data assimilation technology to perform data assimilation on the numerical model of estuarine saltwater intrusion. The sample construction module 400 is used to construct a runoff collection sample based on the measured runoff in the estuarine area. The numerical prediction module 500 is used to input the runoff collection sample into the numerical model of estuarine saltwater intrusion for calculation and processing to obtain a numerical prediction result.

[0119] Each module in the system of the present invention may be implemented in whole or in part through software, hardware, or a combination thereof. Each of the modules may be embedded in or independent of a processor in a computer device in the form of hardware, or may be stored in a memory in the computer device in the form of software, so that the processor can call and execute the operations corresponding to each of the modules.

[0120] In order to better illustrate the operational ensemble numerical prediction method for estuary saltwater intrusion of the present invention, a more detailed and specific application embodiment is given below, which is implemented based on the Yangtze River Estuary area:

[0121] 1. Establishment of a numerical model for saltwater intrusion in the Yangtze River Estuary

[0122] 1.1 Ocean numerical model

[0123] The numerical model uses a three-dimensional numerical model of estuarine saltwater intrusion. This model is improved and developed based on the estuarine, coastal and ocean numerical model ECOM-si (Semi-implicit Estuarine, Costal and Ocean Model). It takes into account dynamic factors such as runoff, tides, wind stress, and bottom friction, and adopts an Arakawa C grid configuration in the horizontal direction. To better simulate the complex coastline of the estuary, a non-orthogonal curvilinear grid was developed. The model uses σ coordinates in the vertical direction to better characterize the changes in bottom shape with water depth. To eliminate the CFL criterion and increase the time step, the barotropic gradient force term in the momentum equation in the model is solved implicitly, and the continuity equation is solved using a semi-implicit method. The advection term model that produces slow processes uses an explicit difference method. To ensure the vertical stability and high resolution of the model, the turbulent diffusion and viscosity terms in the vertical direction are solved implicitly.

[0124] To better simulate saltwater intrusion at the estuary, the model has been improved. By introducing a dry-wet discrimination method to handle the dynamic boundary of the tidal flat, the model's computational accuracy for the sandbar shoals at the Yangtze River Estuary has been improved. The advection term in the material transport equation is solved using the third-order HSIMT-TVD (high-order spatial interpolation at the middle temporal level) format. This format significantly reduces numerical dissipation, eliminates numerical dispersion, and improves computational accuracy during material transport. With these improvements, the model can accurately simulate the hydrodynamic processes and saltwater intrusion phenomena at the Yangtze River Estuary. The improved model has been widely used in research on the hydrodynamic processes and saltwater intrusion at the Yangtze River Estuary, achieving numerous research results.

[0125] 1.2 Grid and water depth

[0126] The computational domain and grid of the model are as follows Figure 3As shown in a, the calculation range mainly covers the entire Yangtze River Estuary, Hangzhou Bay, and the northern Jiangsu waters. The easternmost point extends to 124.5°E longitude. The western river boundary is at the Datong Hydrological Station, which monitors the Yangtze River runoff in real time. The north extends to approximately 33.5°N latitude and the south extends to approximately 28°N latitude. The model uses a non-orthogonal curved grid, which effectively simulates the complex coastline of the Yangtze River Estuary. The grid is also encrypted in each branch channel. For example, the resolution is as high as 100m ( Figure 3 The model is divided into 10 vertical layers with a time step of 40 seconds. The wet-dry discrimination method is used to treat the dynamic boundary of the tidal flat, with a critical water depth of 0.2 m.

[0127] The model uses the measured water depth of the Yangtze River estuary in 2021 (such as Figure 4 As shown in the figure, the North Branch continues to silt up compared to the water depth in 2016. This will weaken the saltwater intrusion from the North Branch into the South Branch, reducing the saltwater intrusion from the South Branch and its impact on the water source.

[0128] 1.3 Initial Field

[0129] Since the water level field and flow field adjust quickly, their initial fields are set to 0.

[0130] The initial salinity field outside the Yangtze River estuary was obtained by digitizing the Bohai, Yellow Sea, and East China Sea Marine Atlas (Hydrology) (Marine Atlas Editorial Committee, 1992), while the initial salinity field inside the Yangtze River estuary was obtained by interpolation based on multi-year dry season measured data.

[0131] 1.4 Boundary conditions

[0132] The open boundary of the model offshore is driven by the tide level and residual water level. The tide level considers 16 partial tides (M2, S2, N2, K2, K1, O1, P1, Q1, U2, V2, T2, L2, 2N2, J1, M1, OO1) and is synthesized by the harmonic constants of each partial tide. The data are from the NAOTIDE database (http: / / www.miz.nao.ac.jp / staffs / nao99 / index_En.html).

[0133]

[0134] Among them, ζ is the actual total water level, H i is the amplitude, ω i is the angular frequency, V i +u i represents the correction angle, g i is the retardation angle, f is the node factor, and ζ0 is the boundary residual water level, which reflects the shelf circulation and is provided by a larger regional model calculation.

[0135] The upstream open boundary of the model runoff is set at Datong hydrological station and is given in the form of flux (average daily runoff).

[0136] Wind stress is also an important factor in the saltwater intrusion of estuaries, which is calculated by the wind field in the 10m layer of the sea surface according to the quadratic law formula.

[0137] 2. Calibration of bottom friction parameters

[0138] Bottom friction is the primary factor in tidal wave energy dissipation, significantly influencing the amplitude and phase distribution of tidal waves. Tidal wave motion is one of the fundamental forms of seawater motion, determining material transport and crucial for numerical simulations of salinity. The bottom friction coefficient, derived from boundary layer theory, varies with time during periodic motion.

[0139] In estuarine ocean numerical models, the bottom friction drag coefficient is generally solved using the logarithmic rate formula of the bottom friction drag coefficient in the bottom boundary layer (Batchelor, 1953; Dyer, 1980; Soulsby and Dyer, 1981; Lueck, 1988):

[0140] τ b =ρC d |u b |u b (2-1)

[0141]

[0142] Among them, τ b is the bottom friction; C d is the bottom friction drag coefficient; u b is the bottom flow velocity; z b is the bottom thickness; κ is the Karman constant (taken as 0.4 here); z0 is the bottom roughness.

[0143] Collins (1998) analyzed 192 sets of velocity data collected from the tidal flats of Loughor Estuary, Swansea Bay, and Wash Bay and found that less than 40% of the data exhibited a logarithmic distribution. Therefore, the logarithmic rate formula is not necessarily applicable in shallow and ultra-shallow estuaries. The North Branch of the Yangtze River Estuary is an ultra-shallow channel, with average water depths of 3.07, 3.44, and 5.07 m (Yellow Sea 85 datum) in the upper, middle, and lower sections, respectively. Furthermore, 39%, 23%, and 32% of the upper, middle, and lower sections of the North Branch have water depths below 2 m, respectively. Previous studies have often used the Shecai-Manning formula to calculate the bottom friction drag coefficient for shallow open channels (Dronkers, 1964; Ludwick, 1975; Dunbar et al., 1991; Bokhorst, 2003).

[0144]

[0145] In the formula, C is the Chow coefficient, whose value depends on the Manning roughness coefficient n (Chow, 1959).

[0146]

[0147] Where H is the water depth.

[0148] In the above formula, the magnitude of the Manning's roughness coefficient must be determined. Research has found that the Manning's roughness coefficient varies with water depth. Ree (1949), through extensive flume experiments, concluded that the Manning's roughness coefficient varies at different water depths. Qian Ning (1959), analyzing measured data from various hydrological stations in the lower Yellow River, concluded that the Manning's roughness coefficient decreases with rising water levels. Chow (1959) believed that in most open channels, the Manning's roughness coefficient decreases with increasing water depth. Knight (1981) and Knight (2009), through field measurements of the Conwy Estuary, Main River, Severn River, Waiwakaiho River, and Tomebamba River, found that the Manning's roughness coefficient varies significantly at different water levels and decreases with rising water levels. Cheng (1993) found that when the Manning's roughness coefficient was assigned according to the principle that "the Manning's roughness coefficient decreases with increasing water depth," the model calculation results fit the measured data well.

[0149] Based on the above, and according to previous studies (Cheng et al., 1993; Bokhorst, 2003; Knight et al., 2009), the Manning roughness coefficient n decreases with increasing water depth. After extensive testing and experimentation, the Manning roughness coefficient for the North Branch was assigned according to the rules in Table 2-1.

[0150] Table 2-1 Values ​​of the Manning Roughness Coefficient of the Northern Branch

[0151] Water depth (m) Manning's roughness D<2 0.030 2≤D<3 0.024 3≤D<10 0.014 D≥10 0.010

[0152] The traditional bottom friction drag coefficient adopts the quadratic law formula and the Xie Cai-Manning coefficient that varies with water depth, and the model calculation results (such as Figure 5 As shown in the figure, the bottom friction drag coefficient calculated using the modified Xie Cai-Manning formula (Test 2) can well simulate the North Branch backflow phenomenon. However, the North Branch backflow calculated using the logarithmic rate formula (Test 1) is significantly understated. Therefore, the accuracy of the bottom friction drag coefficient calculation has a significant impact on the accuracy of the North Branch backflow simulation.

[0153] The Yangtze River Estuary is the source of inland saltwater intrusion, mainly from the backflow of the North Branch saltwater tide. Therefore, the Xie Cai-Manning formula is used to calculate the bottom friction drag coefficient of the North Branch to better simulate the backflow of the North Branch saltwater tide and improve the accuracy of saltwater intrusion forecast.

[0154] 3. Data Assimilation

[0155] Based on the online monitoring network of saltwater intrusion in the Yangtze River Estuary, after data cleaning and quality control, the optimal interpolation data assimilation technology is applied to carry out online salinity observation data assimilation three times a day to improve the accuracy of the initial field of numerical forecasts, thereby improving the accuracy of forecast results.

[0156] 3.1 Setting of assimilation stations

[0157] 3.1.1 Position Selection

[0158] The assimilated data are based on the salinity data observed by the online monitoring network for saltwater intrusion in the Yangtze River Estuary. Some stations are selected from the online monitoring network for data assimilation.

[0159] The selection principles for assimilation stations are: select as many as possible; be sufficiently far away from key forecast points; and not be affected by unconsidered inland river runoff.

[0160] According to the above principles, the current analysis shows that there are 10 stations that cannot be used for data assimilation (Table).

[0161]

[0162]

[0163] For example, see Figure 6 Wusong Water Plant Station is located at the Huangpu River estuary. The model does not consider the flow of the Huangpu River into the river, so this station is not used for assimilation. Figure 7 ,The Pudong Airport observation point is also affected by the nearby river and is not used for assimilation.

[0164] 3.1.2 Virtual Site Settings

[0165] The observation stations in some areas are too sparse, which is not conducive to the assimilation of the model initial field. The assimilation of salinity gradient differences can be achieved by setting virtual stations.

[0166] The virtual station location is the midpoint of the line connecting the two reference stations, and its data value is configured in the form of a polynomial. Salinity value of the virtual station:

[0167] z=ax+by

[0168] Where a and b are coefficients, and x and y are the salinity of the selected site.

[0169] Virtual site configuration coefficient table

[0170] coefficient value a 0.5 b 0.5

[0171] For example, the observation station in the middle section of the northern branch is missing, and the salinity gradient difference between Sanjiao Port and Qinglong Port is assimilated by setting three virtual stations.

[0172] 3.1.3 Assimilation of Stations

[0173] After comprehensive station selection and virtual station configuration, 23 assimilated stations were finally determined, as shown in the following table:

[0174] X Y Z name 1 383280.4508 3509352.869 0 Sanjo Port 2 358694.929 3518422.53 0 Qingsan Middle School 2 3 346401.5 3522957 0 Qingsan Middle School 1 4 370987.705 3513887.7 0 Qingsan Middle School 3 5 334109.4068 3527492.188 0 Qinglong Port 6 326867.0937 3516682.035 0 Chongming Head 7 326816.478 3523781.163 0 Qingniu Hydrological Station 8 316016.0055 3514070.185 0 Baimaosha 9 337259.0007 3496591.105 0 Taicang Petrochemical 10 343749.0145 3488569.359 0 River Center 11 349596.8838 3484046.354 0 Stone cave entrance 12 347271.8027 3500290.002 0 South Gate 13 367446.8023 3490150.977 0 Fort Town 14 375289.9512 3473304.313 0 Changxing Island 15 390470.8284 3463833.758 0 Hengsha Island 16 376581.3635 3481405.629 0 Green Grass Sand B 17 369173.7125 3486527.466 0 Green Grass Sand C 18 388321.6799 3477371.43 0 Hengsha North 19 403804.8783 3471456.455 0 Beigangwai 20 354384.5553 3489479.692 0 Xiabiantansha 21 413282.9164 3467274.819 0 Hengsha East 22 377074.7479 3487002.321 0 Liuyang 23 390036.8501 3473183.262 0 Gongqingwei

[0175] 3.2 Data Quality Control

[0176] The purpose of data assimilation is to improve the accuracy of the initial numerical field by assimilating observational data. Therefore, the observational data used must be accurate and high-quality. Quality control is essential to producing high-quality marine science data. The purpose of quality control is to strictly control data quality and ensure its authority.

[0177] The quality control methods for salinity data include format inspection, correlation inspection, missing inspection, range inspection, continuity inspection, visualization graph drawing inspection and comprehensive analysis inspection. The quality control process is as follows: Figure 8 As shown. Data element range test values ​​are shown in Table 3-1. Quality control methods are implemented through computer-automated quality control and manual review. Computers are used to perform unified standardization and range and logic verification of large quantities of data, combined with manual review such as visual mapping.

[0178] Table 3-1 Salinity element test value parameters

[0179] parameter Range Testing Gradient testing Peak test salinity [0，41] 5 1

[0180] (1) Format check

[0181] File name verification: Check whether the data files are named according to standard file naming rules.

[0182] Data record format inspection: Check whether the data record arrangement sequence, starting position, length, record type identification, data storage type, etc. comply with the corresponding format regulations. If not, errors will occur when reading the data, which should be corrected before other quality control inspections.

[0183] (2) Missing test

[0184] Check whether a certain observation data record is missing data. If it is missing data, no further checks will be performed on the record.

[0185] (3) Range test

[0186] Based on the characteristics of the observed element, the normal value range of the element is determined. If it exceeds this range, the data is considered abnormal. The range test method is divided into the following aspects according to the method of determining the normal value range parameters of the element.

[0187] a) Extreme value range test

[0188] In general, when the observed value of a certain element in a fixed area exceeds the statistical extreme value range of the element in that area for many years (generally not less than 20 years), the data is suspicious. i Formula (3-1) should be satisfied; otherwise, the data is suspicious and further analysis should be conducted to determine whether it is correct.

[0189] X min ≤x i ≤X max (3-1)

[0190] Where: X min is the minimum value of the factor's statistics over many years; x i is the observed value; X max It is the maximum value of the feature statistics over many years.

[0191] b) Empirical range test

[0192] The value range of the factors obtained from personal experience or literature is used as the quality control parameter.

[0193] c) Grubbs Criterion

[0194] When using the Grubbs criterion to perform quality control on element observations, the data should satisfy formula (3-2), otherwise the data is suspicious.

[0195]

[0196] Where: x i is the observed value; is the mean of the observed values; σ is the standard deviation of the data series; G is the Grubbs critical value, which is calculated by formula (3-3).

[0197]

[0198] In the formula: n is the number of data sequences; t is the critical value of the single boundary test t distribution with n-2 degrees of freedom and a / n significance level, which is generally obtained by looking up the general function or the t distribution critical value table. a is generally taken as 0.05 or 0.01.

[0199] (4) Continuity test

[0200] (a) Gradient test

[0201] The ocean observation elements have continuity within a certain time or space range, and the change values ​​of observation elements with close time or adjacent locations should be within a certain range, otherwise the data is considered abnormal. The specific method is:

[0202] Assume that the current observation value x i , the previous non-missing value x adjacent to it in time or space i-1 , continuity test parameter H g , then formula (3-4) should be satisfied, otherwise x i and x i-1 The gradient of change is large, and both data are suspicious. Further analysis is needed to confirm whether they are abnormal.

[0203] |x i -x i-1 |≤H g (3-4)

[0204] Where: H g It is a gradient test parameter, which is determined based on factors such as feature type, observation time interval, spatial distance, observation time and area.

[0205] (b) Peak test

[0206] The changes of ocean observation elements within a certain spatial or temporal range are limited. If an observation value is significantly different from the surrounding observation values ​​and a large mutation occurs, it is judged as an outlier.

[0207] Assume that the current observation value is x i , the first correct observation adjacent to it in time or space is x i-1 and x i+1 , then x is required i Satisfy formula (3-5), otherwise it is considered that x i suspicious.

[0208] |x i -(x i-1 +x i+1 ) / 2|≤H j (3-5)

[0209] Where: H j It is the peak test parameter, which is determined according to factors such as feature type, observation time interval or spatial interval, observation time and area.

[0210] (c) Constant test

[0211] While observation instruments have sufficient sensitivity and accuracy, oceanographic observation elements are affected by fluid dynamics and will not remain constant over a certain time or space. If they do, the data is questionable. The specific method is as follows.

[0212] Find the maximum value V of an element in a certain period of time (or a profile) max and minimum value V min , formula (3-6) should be satisfied, otherwise the data of this section (section) is abnormal.

[0213] V max -V min ≥H h (3-6)

[0214] Where: V max is the maximum value of the observation data sequence or the maximum value of a certain observation section; V min is the maximum value of the observation data sequence or the minimum value of a certain observation section; H h To maintain constant test parameters, the parameters are determined based on factors such as feature type, observation time, data accuracy and region.

[0215] (5) Correlation test

[0216] Data anomalies are checked through the interrelationships between elements. For example, the recorded values ​​at fixed or hourly intervals within a day cannot exceed the daily extreme value; the data for the same element at the same observation time but at different observation frequencies should be equal; the maximum wave height should be greater than or equal to the average wave height; the extreme high (low) tide height should not be lower (higher) than the hourly tide height at the adjacent time; and the wind speed and wave height levels should meet the corresponding relationship.

[0217] (6) Visual graphics drawing test

[0218] Observational elements change continuously within a certain spatial and temporal range. Drawing visual graphs can intuitively display out-of-range anomalies, sudden changes, spikes, and missing values, effectively assisting manual review. For example, by drawing a time series process curve for each element, spikes are displayed as outliers. When drawing data series process curves for the same element at different observation frequencies, observations at the same moment should overlap; otherwise, the data record at that moment is abnormal.

[0219] When necessary, perform manual inspection of visual graphics.

[0220] 3.3 Assimilation method

[0221] Improvements in numerical forecasting rely heavily on the accuracy of initial conditions and the robustness of models. In recent years, with the continuous improvement of numerical models, the demand for accurate initial model values ​​has intensified. Data assimilation technology is an effective way to obtain accurate initial model values. Common data assimilation methods include dynamic approximation, optimal interpolation, Kalman filtering, and variational assimilation.

[0222] The optimal interpolation method is a widely used and mature data assimilation method internationally. It considers the statistical characteristics of background fields and observation errors, encompassing the inherent relationships between observation, forecast, and analysis. It determines the optimal weights in a least-squares sense, thereby statistically minimizing analysis errors. The Japan Meteorological Agency's weather and ocean forecast systems, as well as the ocean data assimilation system within the ENSO forecast system of the U.S. National Center for Environmental Protection and Development (NCEP), all use the optimal interpolation method as one of their observational data assimilation schemes.

[0223] The assimilation scheme of the numerical forecast system adopts the optimal interpolation assimilation method. The basic algorithm of this method is:

[0224] x a =x b +K(yH[x b ])

[0225] K=BH T (HBH T +O) -1

[0226] Among them, x a and x a are the analysis field and background field, respectively, K is the optimal weight, y is the observation data, H is the linear operator for interpolating the background field value to the observation position, the superscript T represents the transpose of the matrix, B is the background field error covariance matrix, and O is the observation field error covariance matrix.

[0227] 3. Construction of runoff aggregate sample

[0228] The boundary conditions that need to be adjusted for the Yangtze River Estuary saltwater intrusion numerical model over time are runoff boundary conditions and wind field upper boundary conditions. Considering the uncertainty of upstream water inflow and wind field, it is necessary to use ensemble forecasting technology to improve the coverage of numerical simulation results.

[0229] The magnitude of runoff determines the severity of saltwater intrusion at the estuary, and accurately determining its value is crucial for numerical saltwater intrusion forecasts. The Datong Hydrological Station, 620 km from Xuliujing at the Yangtze River estuary, provides daily runoff data, providing the basis for the model's determination of the river's open boundary for the momentum equation. The upstream open boundary of the numerical forecast model for saltwater intrusion at the Yangtze River estuary is set at the Datong Hydrological Station. Runoff there affects saltwater intrusion at the estuary approximately one week later, naturally accounting for the lag in its impact.

[0230] However, the Datong hydrological station is still 620 km away from Xuliujing at the Yangtze River estuary. The changes in runoff during this period will affect the runoff into the sea and the accuracy of saltwater intrusion forecasts. The main factors affecting runoff are the East Route of the South-to-North Water Diversion Project and other river diversion and drainage projects (Qiu Cheng, 2014). The East Route of the South-to-North Water Diversion Project is located in Yangzhou and diverts water to the north through the Grand Canal. The first phase of the project is 500m3 / s, the second phase project is 800m 3 / s. In extremely dry years, pumping water along the Yangtze River below Datong will reduce the runoff of the Yangtze River by 500-700m 3 / s. While specific values ​​for water diversion and drainage along the river haven't been given for normal years, the amount of water pumped is much smaller than in dry years, so it can generally be assumed that diversion and drainage are essentially balanced. Therefore, in the Yangtze River Estuary numerical forecast model, the runoff boundary condition is based on the measured runoff in the estuary region, minus the water diversion volume from the East Route Project of the South-to-North Water Diversion Project.

[0231] 4.1 Short-term runoff prediction sample

[0232] The model's runoff is input from the Datong Station on the Yangtze River. However, the Datong runoff still has a lag of 5-7 days before it reaches the Yangtze River Estuary. Therefore, the current Datong runoff is the Yangtze River runoff at the Yangtze River Estuary 5-7 days from now, while the Datong runoff 5-7 days ago is the current Yangtze River runoff at the Yangtze River Estuary.

[0233] There is still a certain gap between the Datong flow value and the final runoff input to the Yangtze River Estuary. This gap is mainly due to the fact that there is actually diversion and drainage along the way from Datong Station to the Yangtze River Estuary, and the model does not set up diversion and drainage along the way. Therefore, it is necessary to directly reflect the diversion and drainage along the way in the Datong input of the model. According to the above analysis, it is believed that the reduction in extremely dry years is 500-700m 3 / s, water diversion and drainage are basically balanced in normal years.

[0234] The flow input of the large diameter before the start of the model calculation is mainly given based on the measured flow at the large diameter station:

[0235] y=ax+b

[0236] Among them, y is the runoff input of the model, x is the measured flow of Datong, and a and b are adjustment coefficients, which mainly reflect the drainage along the way.

[0237] Table of runoff input adjustment factors for the model

[0238] Datong flow value a b >=20000m3 / s 1.0 0 >12000m3 / s,<20000m3 / s 1.0 -700 <=12000m3 / s 1.0 -2000

[0239] 4.2 Construction of runoff prediction samples

[0240] The runoff input after the model calculation start time, that is, the input for predicted runoff, needs to be constructed based on other information because there is no measured data.

[0241] 4.2.1 Constructing runoff prediction samples based on forecast information

[0242] The Forecast Center of the Yangtze River Commission's Hydrological Bureau will release forecast information on large-scale runoff flow. When the forecast information is available, the runoff input will be constructed based on this forecast information.

[0243] The large-flow flow information released by the Forecast Center of the Yangtze River Commission's Hydrological Bureau includes short-term forecasts (daily forecasts for the past five days), medium-term forecasts (daily forecasts for every ten days), and long-term forecasts (monthly characteristic forecasts).

[0244]

[0245] The construction method is consistent with the measured part:

[0246] y=ax+b

[0247] Among them, y is the runoff input of the model, x is the measured flow of Datong, and a and b are adjustment coefficients, which mainly reflect the drainage along the way.

[0248] Table of runoff input adjustment factors for the model

[0249] Datong flow value a b >=20000m3 / s 1.0 0 >12000m3 / s,<20000m3 / s 1.0 -700 <=12000m3 / s 1.0 -2000

[0250] There are differences in confidence levels between short-term, medium-term, and long-term forecast products. The confidence level of short-term forecasts is higher than that of medium-term forecasts, which in turn is higher than that of long-term forecasts. Therefore, when constructing runoff samples at a specific moment, the forecast product with the highest confidence level is given priority.

[0251] 4.2.2 Empirical construction of runoff prediction samples

[0252] In the absence of runoff forecast information, runoff prediction samples can be predicted using empirical construction. Considering that runoff changes are continuous, the runoff prediction value in a short period of time can be estimated based on the latest observation value.

[0253] Construct a forecast based on the last observed value (or the average value over the most recent period, such as 24 hours).

[0254] y=ax+b

[0255] Among them, y is the runoff input of the model, x is the latest observation value of the measured flow in Datong, and a and b are adjustment coefficients, which mainly reflect the changes in continuity.

[0256] Table of runoff input adjustment factors for the model

[0257]

[0258]

[0259] In the absence of long-term forecast information on a monthly scale, climatological values ​​(multi-year averages) can be used to construct runoff sample values ​​over a longer period of time.

[0260] 4.3 Runoff aggregate sample construction scheme

[0261] The historical data of the runoff sample is calculated based on the observation value at the corresponding time as y=a1x+b1 (where y is the historical data value of the sample and x is the observation value at the corresponding time); the configuration items are the historical runoff coefficient a1 and the historical runoff offset b1.

[0262] The forecast data is constructed based on the average observation value of the last 24 hours through y=a2x+b2 (where y is the forecast value and x is the average observation value). The time is 6 hours later and 9999 hours later. The configuration items are historical runoff coefficient a2 and historical runoff offset b2.

[0263] Configure the constant value of runoff in the format of time + value. The configured data replaces the calculated data at that time. There may be multiple values.

[0264] The runoff sample set is constructed as follows Figure 9 shown.

[0265] V. Operational Numerical Forecasting

[0266] The runoff samples are fed into a numerical model for saltwater intrusion at the estuary, where they are processed and numerically predicted. Uncertainty in the Yangtze River runoff is the primary source of uncertainty in the calculations of the model. Developing an ensemble forecasting technique for runoff, using multiple conditions and case studies with varying probability of occurrence, can more comprehensively cover the potential for saltwater intrusion and effectively avoid under-reporting in actual forecasts.

[0267] The basic principles, main features, and advantages of the present invention are shown and described above. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The above embodiments and descriptions are merely illustrative of the principles of the present invention. Various changes and modifications may be made to the present invention without departing from the spirit and scope of the present invention. Such changes and modifications are intended to fall within the scope of the present invention. The scope of protection claimed in the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for operational ensemble numerical prediction of estuarine saltwater intrusion, characterized by: include: Constructing a numerical model of saltwater intrusion at the estuary and setting relevant parameters of the numerical model; Calculating and processing the bottom friction parameters of the estuary area, and calibrating the estuary saltwater intrusion numerical model according to the calculation results; The optimal interpolation data assimilation technology is used to perform data assimilation processing on the estuary saltwater intrusion numerical model; Based on the measured runoff in the estuary area, a runoff aggregate sample is constructed; as well as Inputting the runoff aggregate sample into the estuary saltwater intrusion numerical model for calculation and processing to obtain a numerical prediction result; The runoff aggregate sample is constructed based on the measured runoff in the estuary area, including: To construct a short-term runoff prediction sample, the measured runoff volume in the estuary area is subtracted from the drainage volume along the way. The specific value is determined according to formula (4) and Table 2: y=ax+b (4) Among them, y is the runoff input of the model, x is the measured flow in the estuary area, and a and b are adjustment coefficients, which mainly reflect the drainage along the way; Table 2 Model runoff input adjustment coefficients The runoff prediction samples are constructed based on the forecast information of the authoritative agencies. If there is no forecast information, the runoff prediction samples are constructed based on experience. The prediction value is constructed based on the last measured value or the average value of the most recent period. The specific value is determined according to formula (5) and Table 3: y=ax+b (5) Where y is the runoff input of the model, x is the latest observed value of the measured flow in the estuary area, and a and b are adjustment coefficients, which mainly reflect the change of continuity; Table 3 Model runoff input adjustment coefficients If there is no measured data for empirical construction, the climatological state value is used to construct the runoff sample value; Construct a runoff aggregate sample, and generate multiple runoff prediction samples according to the above by modifying the measured runoff value to form a runoff aggregate sample.

2. The operational ensemble numerical prediction method for estuarine saltwater intrusion according to claim 1, characterized in that: The method of constructing a numerical model of saltwater intrusion at the estuary and setting relevant parameters of the numerical model of saltwater intrusion at the estuary includes: A numerical model of saltwater intrusion in estuaries was constructed, taking into account runoff, tides, wind stress, and bottom friction dynamic factors. A non-orthogonal curvilinear grid was used horizontally and σ coordinates were used vertically. Set the calculation area and perform local encryption on the grid in key areas; Set vertical stratification, time step and critical water depth; Set the initial field values ​​of water level field, flow field and salinity field; and Set the tide level, residual water level, runoff and wind field boundary conditions.

3. The operational ensemble numerical prediction method for estuary saltwater intrusion according to claim 1, characterized in that: The calculating and processing of the bottom friction parameters of the estuary area and the calibration processing of the estuary saltwater intrusion numerical model according to the calculation results include: The bottom friction parameters of the estuary area include the Manning roughness coefficient and the bottom friction drag coefficient. The Manning roughness coefficient of the estuary area is assigned according to Table 1: Table 1 Values ​​of Manning's roughness coefficient Calculate the Xiecai coefficient C according to formula (1): Where H is the water depth, n is the Manning roughness coefficient; The bottom friction drag coefficient C in the estuary area is calculated according to formula (2): d : Where C is the Xie Cai coefficient, g is the acceleration due to gravity; The Manning roughness coefficient and bottom friction drag coefficient of the estuary area are used to calibrate the estuary saltwater intrusion numerical model.

4. The operational ensemble numerical prediction method for estuary saltwater intrusion according to claim 1, characterized in that: The optimal interpolation data assimilation technology is used to perform data assimilation processing on the estuary saltwater intrusion numerical model, including: Select the actual measurement sites for data assimilation. If the sites in some areas are too sparse, set up virtual sites. Conduct quality control on the measured data of selected sites; The optimal interpolation method shown in formula (3) is used to assimilate data and produce the initial field of the model; Among them, x a and x b are the analysis field and background field, respectively, K is the optimal weight, y is the observation data, H is the linear operator for interpolating the background field value to the observation position, the superscript T represents the transpose of the matrix, B is the background field error covariance matrix, and O is the observation field error covariance matrix.

5. The operational ensemble numerical prediction method for estuary saltwater intrusion according to claim 4, characterized in that: The quality control includes format inspection, relevance inspection, missing inspection, range inspection, continuity inspection, visual graphics drawing inspection and comprehensive analysis inspection.

6. A system for implementing the operational ensemble numerical prediction method for estuarine saltwater intrusion according to any one of claims 1 to 5, characterized in that: include: A model construction module, wherein the model construction module is used to construct a numerical model of estuarine saltwater intrusion and set relevant parameters of the numerical model of estuarine saltwater intrusion; A parameter calibration module, wherein the parameter calibration module is used to calculate the bottom friction parameters of the estuary area and calibrate the estuary saltwater intrusion numerical model according to the calculation results; A data assimilation module, wherein the data assimilation module is used to perform data assimilation processing on the estuary saltwater intrusion numerical model by using an optimal interpolation data assimilation technology; A sample construction module, wherein the sample construction module is used to construct a runoff aggregate sample based on the measured runoff in the estuary area; as well as The numerical prediction module is used to input the runoff set samples into the estuary saltwater intrusion numerical model for calculation and processing to obtain numerical prediction results.

Images