CN113609783B - Upward forecasting system and method of salt tide coupled with large-scale circulation climate information - Google Patents

Abstract

The invention relates to a salt tide upward forecasting system and method coupled with large-scale circulation climate information. The system includes a data input module, a model forecast module and a data output module. The salt tide upward forecasting system coupled with large-scale circulation climate information obtains the hydrometeorological data of the historically affected salt tide upward for the areas that need to predict the area affected by the salt tide upward movement, and establishes the salt tide upward forecast based on the key influencing factors selected from the data. model, and verify the salt tide upward forecast model, and the chloride content output by the salt tide upward forecast model that has passed the verification is used as the forecast data of the salt tide upward forecast in a certain area; the salt tide upward forecast system coupled with the large-scale circulation climate information obtains The forecast data has high precision, which solves the technical problems that the existing salt tide upward forecast model has high data requirements, high cost of information acquisition, and the interference of the cross-correlation of forecast factors, resulting in low forecast precision.

Description

Upward forecasting system and method of salt tide coupled with large-scale circulation climate information

technical field

The invention relates to the technical field of environmental data models, in particular to a salt tide upward forecasting system and method coupled with large-scale circulation climate information.

Background technique

As a typical water environment problem in estuary areas, salt tide uptrend occurs in countries or regions along the coast or along the river. Severe upward salt tide will not only change the water quality in the river, destroy the ecological balance of water, but also disturb the water supply order of surrounding cities and restrict local social and economic development. Therefore, it is of great practical significance for regional water resources management to study the key influencing factors and time lag of the uptake of salt tide, and to establish a medium and long-term forecast mechanism for the uptake of salt tide.

Existing methods for upcasting of salt tide in estuaries include empirical formula method, physical experiment method, and numerical simulation method. Among them: the empirical formula method mainly establishes the empirical formula between the salt tide and its influencing factors, and derives the mathematical relationship between the two based on the statistics and analysis of a large number of measured data. For example, in Estuarine Dynamics and Morphology, David Plantell summarizes a chloride simulation equation suitable for partially mixed estuaries. coefficient, longitudinal mixing coefficient, runoff, and the distance from the point to the estuary, but this empirical formula method ignores the time factor of the horizontal convection-diffusion equation. The physical experiment method is mainly the method of physical model or field test, which restores and simulates the physical process. For example, through a field test in the Twitt Estuary in the United Kingdom, the response law of chloride content to runoff and tide is analyzed, and it is found that the Richardson number of the estuary reaches the empirical maximum value during the ebb tide period, and the minimum value during the high tide period. In "Modaomen Salt Tide Intrusion and Salt Suppression Technology", Lu Chen, He Yong, etc. carried out the physical model test of the salt tide upstream of the Modaomen waterway. The changing characteristics of the salt tide in the Modaomen waterway. Numerical simulation method mainly establishes and solves mathematical models through computer methods. For example, the unstructured finite volume nearshore ocean three-dimensional model FVCOM was used to establish a numerical model for the uptake of the salt tide from the Modaomen channel to Lingdingyang, and the temporal and spatial variation of the chloride content in the Modaomen channel and its vertical stratification were studied and analyzed. Phenomenon.

However, most of the above three methods have higher requirements on application conditions or data, and the operation is more complicated, which limits the research on the upcasting of salt tide in data-deficient areas. In addition, the main factors affecting the uptake of salt tide include upstream water, tides, sea level, wind, and estuary topography. At present, the above-mentioned three methods have not considered the application of large-scale circulation climatic factors to the mid- and long-term forecasting of salt tide upward, and most of them have constructed salt tide upward forecast models based on daily-scale runoff and tidal changes. In general, there are few studies on the simulation of chloride content on other time scales, which leads to the low accuracy of the simulated salt tide.

SUMMARY OF THE INVENTION

The embodiments of the present invention provide a salt tide upward forecasting system and method coupled with large-scale circulation climate information, which are used to solve the problem that the existing salt tide upward forecasting model has high data requirements, high information acquisition costs, and the existence of cross-correlation of forecasting factors. interference, which makes the technical problem of low forecast accuracy.

In order to achieve the above purpose, the embodiments of the present invention provide the following technical solutions:

A salt tide upward forecasting system coupled with large-scale circulation climate information, comprising a data input module, a model forecast module and a data output module;

The data input module is used to obtain the hydrometeorological data that affects the uptake of the salt tide, and obtains the key influencing factors that affect the uptake of the salt tide through the random forest importance analysis of the obtained hydrometeorological data;

The model forecasting module is used for establishing a salt tide upward forecasting model by using a random forest algorithm according to the key influencing factors, and verifying the salt tide upward forecasting model to obtain an optimal salt tide upward forecasting model;

The data output module is used for outputting the optimal model for the upward forecasting of the salt tide to predict the chlorine content of the area affected by the upward salt tide.

Preferably, the data input module includes a data acquisition sub-module, a data analysis sub-module and a screening sub-module;

The data acquisition sub-module is used to obtain the historical hydrological factors, large-scale circulation climate factors and chlorine content of the estuary area of a certain place, and obtain the hydrometeorological data of the salt tide upstream;

The data analysis sub-module is used to perform random importance analysis on the hydrometeorological data affecting the upward salt tide through the estimation error and Gini impurity of the sample set outside the bag, and obtain each factor and the chlorine content at different time scales. between the first important degree set and the second important degree set;

The screening sub-module is used to screen each factor in the first importance degree set and the second importance degree in descending order of importance, to obtain the corresponding first factor set and second importance degree. factor set; and select the same factor from the first factor set and the second factor set as the key influencing factor affecting the uptake of salt tide.

Preferably, the hydrological factors include flow and tide level, and the large-scale circulation climate factors include the Pacific Interdecadal Oscillation and the Southern Oscillation Index.

Preferably, the screening sub-module is configured to sort each factor in the first importance degree set and the second importance degree in descending order of importance, and filter out the top third of the importance degree. Two factors are correspondingly constructed into a first factor set and a second factor set.

Preferably, the model prediction module includes a data set classification submodule, a model establishment submodule and a model verification submodule;

The data set classification submodule is used to obtain all the key influencing factors to construct a data set, and divide the data set into a training set and a verification set;

The model building submodule is used to perform model training on the training set by using the random forest algorithm, and establish a salt tide upward forecasting model;

The verification model sub-module is used to input the verification set into the salt tide upward forecasting model, and output the Nash efficiency coefficient and determination coefficient corresponding to the verification set, if both the Nash efficiency coefficient and the determination coefficient are If it is greater than the coefficient threshold, the optimal model for upcasting of salt tide is obtained.

Preferably, the data set classification sub-module is used to use 80% of the data set as a training set and 20% of the data set as a validation set.

Preferably, the coefficient threshold is 0.5.

The present invention also provides a salt tide upward forecasting method coupled with large-scale circulation climate information, comprising the following steps:

Obtain the hydrometeorological data affecting the uptake of the salty tide, and analyze the importance of the acquired hydrometeorological data through random forests to obtain the key influencing factors that affect the uptake of the salty tide;

According to the key influencing factors, a random forest algorithm is used to establish a salt tide upward forecast model, and the salt tide upward forecast model is verified to obtain an optimal salt tide upward forecast model;

Outputting the optimal model for the upward forecasting of the salt tide to predict the chlorine content of the area affected by the upward salt tide.

Preferably, the steps of obtaining the key influencing factors affecting the upward trace of the salt tide through random forest importance analysis on the obtained hydrometeorological data include:

Obtain the historical hydrological factors, large-scale circulation climatic factors and chlorine content of a certain estuary area, and obtain the hydrometeorological data of the salt tide upstream:

Through the estimation error and Gini impurity of the out-of-bag sample set, the random importance analysis was carried out on the hydrometeorological data affecting the uptake of the salt tide, respectively, and the first importance set and the chloride content at different time scales were obtained. The second important degree set;

After sorting each factor in the first importance degree set and the second importance degree degree according to the importance degree from large to small, the corresponding first factor set and the second factor set are obtained; and from the first factor set and the second factor set; The factor set and the second factor set select the same factor as the key influencing factor affecting the uptake of the salt tide.

Preferably, the step of obtaining the optimal model for upcasting of salt tide includes:

All the key impact factors will be obtained to construct a data set, and the data set will be divided into a training set and a validation set;

The random forest algorithm is used to perform model training on the training set, and a salt tide upward forecasting model is established;

Input the verification set into the salt tide upward forecast model, and output the Nash efficiency coefficient and determination coefficient corresponding to the verification set. If both the Nash efficiency coefficient and the determination coefficient are greater than the coefficient threshold, the salt tide upward forecast is obtained. optimal model.

It can be seen from the above technical solutions that the embodiments of the present invention have the following advantages: the salt tide upward forecasting system and method coupled with large-scale circulation climate information includes a data input module, a model forecast module and a data output module, and the data input module is used for Obtain the hydrometeorological data that affects the uptake of the salty tide, and analyze the importance of the obtained hydrometeorological data through random forests to obtain the key influencing factors affecting the uptake of the salty tide; The data output module is used to output the optimal model of salt tide upward forecast to predict the chlorine content in the area affected by the salt tide upward forecast. The salt tide upward forecasting system coupled with large-scale circulation climate information obtains the hydrometeorological data of the historically affected salt tide upward for the areas that need to predict the area affected by the salt tide upward movement, and establishes the salt tide upward forecast based on the key influencing factors selected from the data. model, and verify the salt tide upward forecast model. The chloride content output by the salt tide upward forecast model that has passed the verification is used as the forecast data of the salt tide upward forecast in a certain area; the salt tide upward forecast system coupled with the large-scale circulation climate information obtains The forecast data has high precision, which solves the technical problems that the existing salt tide upward forecast model has high data requirements, high information acquisition cost, and the interference of the cross-correlation of forecast factors, resulting in low forecast precision.

Description of drawings

In order to explain the embodiments of the present invention or the technical solutions in the prior art more clearly, the following briefly introduces the accompanying drawings that need to be used in the description of the embodiments or the prior art. Obviously, the accompanying drawings in the following description are only These are some embodiments of the present invention. For those of ordinary skill in the art, other drawings can also be obtained based on these drawings without any creative effort.

FIG. 1 is a frame diagram of a salt tide upward forecasting system coupled with large-scale circulation climate information according to an embodiment of the present invention.

FIG. 2 is a frame diagram of a random forest algorithm adopted by the salt tide upward forecasting system coupled with large-scale circulation climate information according to an embodiment of the present invention.

FIG. 3 is a frame diagram of a data input module of a salt tide upward forecasting system coupled with large-scale circulation climate information according to an embodiment of the present invention.

FIG. 4 shows the key influencing factors of the salt tide upward forecasting system coupled with the large-scale circulation climate information according to the embodiment of the present invention.

FIG. 5 is a frame diagram of a model forecasting module of a salt tide upward forecasting system coupled with large-scale circulation climate information according to an embodiment of the present invention.

FIG. 6 is a monthly forecast flow chart of the model forecast module of the salt tide upward forecast system coupled with the large-scale circulation climate information according to the embodiment of the present invention.

FIG. 7 is a monthly-scale salt tide forecast diagram based on the hydrological influence factor forecast of the data output module of the salt tide upward forecasting system coupled with the large-scale circulation climate information according to the embodiment of the present invention.

FIG. 8 is a forecast diagram of monthly-scale salt tide forecast based on the combination of hydrology and large-scale circulation climate influence factors of the data output module of the salt tide upward forecasting system coupled with large-scale circulation climate information according to an embodiment of the present invention.

Detailed ways

In order to make the purpose, features and advantages of the present invention more obvious and understandable, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the following The described embodiments are only some, but not all, embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

The embodiments of the present application provide a salt tide upward forecasting system and method coupled with large-scale circulation climate information, which are used to solve the problem that the existing salt tide upward forecasting model has high data requirements, high information acquisition costs, and the existence of cross-correlation of forecasting factors. interference, which makes the technical problem of low forecast accuracy.

Example 1:

FIG. 1 is a frame diagram of the salt tide upward forecasting system coupled with large-scale circulation climate information according to the embodiment of the present invention, and FIG. 2 is the salt tide upward forecast system used in the coupled large-scale circulation climate information according to the embodiment of the present invention. Random forest algorithm framework diagram.

As shown in FIG. 1 and FIG. 2 , an embodiment of the present invention provides a salt tide upward forecasting system coupled with large-scale circulation climate information, including a data input module 10, a model forecast module 20 and a data output module 30;

The data input module 10 is used to obtain the hydrometeorological data affecting the uptake of the salt tide, and obtain the key influencing factors affecting the uptake of the salt tide by analyzing the acquired hydrometeorological data through random forest importance;

The model forecasting module 20 is used for establishing a salt tide upward forecasting model by using a random forest algorithm according to key influencing factors, and verifying the salt tide upward forecasting model to obtain an optimal salt tide upward forecasting model;

The data output module 30 is used for outputting the optimal model for the upward forecasting of the salt tide to predict the chlorine content of the area affected by the upward salt tide.

In the embodiment of the present invention, the data input module 10 mainly acquires the hydrometeorological data affecting the uptake of the salt tide in a certain area in history, and uses random forest importance analysis on the acquired hydrometeorological data to screen out the key influencing factors that affect the uptake of the salt tide .

In the embodiment of the present invention, the model forecast module 20 mainly uses the selected key influencing factors to establish a salt tide upward forecast model through random forest algorithm, and then verifies the obtained salt tide upward forecast model, and the verified salt tide upward forecast model is verified. The model is regarded as the optimal model for upcasting of salt tide.

In the embodiment of the present invention, the data output module 30 mainly outputs the obtained chlorine content of the salt tide upward forecast optimal model output, and the output chlorine content of the salt tide upward forecast optimal model is to predict the salt tide in a certain area. Backward forecast data.

The present invention provides a salt tide upward forecasting system coupled with large-scale circulation climate information, comprising a data input module, a model forecast module and a data output module. The data input module is used to obtain hydrometeorological data affecting the salt tide upward movement, and to obtain The hydrometeorological data obtained through the random forest importance analysis obtains the key influencing factors that affect the upcasting of the salt tide; the model forecast module is used to establish the upcasting forecast model of the salty tide by using the random forest algorithm according to the key influencing factors, and to verify the upward forecasting model of the salty tide. , to obtain the optimal model for the upward forecast of the salt tide; the data output module is used to output the optimal model of the upward forecast of the salt tide to predict the chlorine content in the area affected by the upward salt tide. The salt tide upward forecasting system coupled with large-scale circulation climate information obtains the hydrometeorological data of the historically affected salt tide upward for the areas that need to predict the area affected by the salt tide upward movement, and establishes the salt tide upward forecast based on the key influencing factors selected from the data. model, and verify the salt tide upward forecast model. The chloride content output by the salt tide upward forecast model that has passed the verification is used as the forecast data of the salt tide upward forecast in a certain area; the salt tide upward forecast system coupled with the large-scale circulation climate information obtains The forecast data has high precision, which solves the technical problems that the existing salt tide upward forecast model has high data requirements, high information acquisition cost, and the interference of the cross-correlation of forecast factors, resulting in low forecast precision.

Fig. 3 is a frame diagram of the data input module of the salt tide upward forecasting system coupled with large-scale circulation climate information according to the embodiment of the present invention, and Fig. 4 is the salt tide upward forecast coupled with large-scale circulation climate information according to the embodiment of the present invention key factors of the system.

As shown in FIG. 3 and FIG. 4 , in an embodiment of the present invention, the data input module 10 includes a data acquisition sub-module 11, a data analysis sub-module 12 and a screening sub-module 13;

The data collection sub-module 11 is used to obtain the historical hydrological factors, large-scale circulation climate factors and chlorine content of the estuary area of a certain place, and obtain the hydrometeorological data of the salt tide upstream;

The data analysis sub-module 12 is used to perform random importance analysis on the hydrometeorological data affecting the uptake of the salt tide through the estimation error and Gini impurity of the sample set outside the bag, and obtain the relationship between each factor and the chloride content at different time scales. The first important degree set and the second important degree set;

The screening sub-module 13 is used to screen each factor in the first importance degree set and the second importance degree degree in descending order of importance, to obtain the corresponding first factor set and second factor set; and from The first factor set and the second factor set select the same factor as the key influencing factor affecting the uptrend of salt tide;

Among them, the hydrological factors include flow and tidal level, etc., and the large-scale circulation climate factors include the Pacific Interdecadal Oscillation and the Southern Oscillation Index.

It should be noted that the screening sub-module 13 is mainly used to sort each factor in the first importance degree set and the second importance degree in descending order of importance, and then filter out the top two-thirds of the importance degree. The factors are correspondingly constructed into a first factor set and a second factor set. Among them, the data input module 10 comprehensively and meticulously discussed the telecorrelation relationship and time lag between 14 large-scale circulation climate factors and the upward movement of the salt tide in a certain area based on the random forest importance assessment, and determined the upward movement of the salt tide at different time scales. combination of predictors.

In the embodiment of the present invention, the data input module 10 is mainly composed of data such as flow, tidal level and other hydrological factors, large-scale circulation factors such as the Pacific decadal oscillation, the Southern Oscillation Index, and chlorine content obtained through the data acquisition sub-module 11. The hydrometeorological data of the salt tide upstream; then, through the random forest importance analysis of the data analysis sub-module 12, the correlation between each factor and the chloride content at different time scales such as day, month, and year in the data is explored. Finally, through the screening sub-module 13, the same and the top 2/3 factors are selected from the first and second important degree sets as the key influencing factors affecting the uptake of the salt tide.

It should be noted that the higher the importance, the closer the correlation between the factor and the chlorine content. Therefore, the key influencing factors that affect the strength of the salt tide upstream activity in a certain estuary area can be obtained. The obtained key influencing factors are used as the input data of the salt tide upward forecasting model, so that the forecast data output by the obtained salt tide upward forecasting optimal model has high precision. The optimal model for the upward forecast of salt tide takes into account the correlation between the large-scale circulation climatic factors and the upward movement of salt tide, which enhances the capture and extraction of the upward forecast of salt tide by the upward forecast model of salt tide. Therefore, the obtained optimal model for upcasting of salt tide can also be effectively applied to mid- and long-term forecasting of salt tide in data-deficient areas.

In the embodiment of the present invention, the importance evaluation of the estimation error of the out-of-bag sample set is also called the replacement importance evaluation. Explain the more important the role of explanatory variables in the model simulation process. Its calculation formula is as follows:

In the formula, VIM(x _i ) _OOB is the importance index of the ith explanatory variable evaluated according to the estimation error of the out-of-bag sample set; n is the number of decision trees; OOBerr _j and OOBerr′ _j are the jth tree before and after replacement, respectively. Out-of-bag sample set estimation error for decision trees. The explanatory variables refer to each influencing factor used to evaluate the importance.

In the embodiment of the present invention, Gini impurity refers to the frequency of randomly extracting sub-items from a data set, and the sub-items are incorrectly marked. For randomly drawn sub-items X ₁ , X ₂ ,...,X _m , the Gini impurity (ie, how often a sub-item is mislabeled) is:

In the formula, f _i is the probability that the random sample X _i is correctly marked.

In the formula, IG _t (f) and IG _r (f) are the Gini impurity corresponding to the two new nodes after node splitting; n is the number of decision trees; k is the explanatory variable x _i in the jth decision tree. The number of times m nodes appear.

Fig. 5 is a frame diagram of the model forecast module of the salt tide upward forecasting system coupled with large-scale circulation climate information according to the embodiment of the present invention, and Fig. 6 is the salt tide upward forecast coupled with large-scale circulation climate information according to the embodiment of the present invention The monthly forecast flow chart of the system model forecast module.

As shown in FIG. 5, in one embodiment of the present invention, the model prediction module 20 includes a data set classification submodule 21, a model establishment submodule 22 and a model verification submodule 23;

The data set classification submodule 21 is used to construct a data set by obtaining all the key influencing factors, and divide the data set into a training set and a verification set;

establishing a model sub-module 22, which is used to perform model training on the training set by using the random forest algorithm, and establish a salt tide upward forecasting model;

The verification model sub-module 23 is used to input the verification set into the salt tide upward forecast model, and output the Nash efficiency coefficient and determination coefficient corresponding to the verification set. If the Nash efficiency coefficient and the determination coefficient are both greater than the coefficient threshold, the optimal salt tide upward forecast is obtained. Model.

It should be noted that the coefficient threshold is preferably 0.5. The data set classification submodule is used to take 80% of the data and data sets from the salt tide as the training set, and 20% of the data and data sets from the salt tide as the validation set. Among them, the model forecast module 20 uses the random forest algorithm to establish a salt tide upward forecast model coupled with large-scale circulation climate information.

As shown in FIG. 6 , in the embodiment of the present invention, the model forecasting module 20 adopts the random forest algorithm to construct a salt tide upward forecasting model in a certain estuary area under different time scales, and gradually inputs the key influencing factors in the training set according to their importance. In the salt tide upward forecast model based on random forest algorithm, combined with the existing chlorine content data, the salt tide upward forecast model is trained and verified, and the input variable combination with the best fitting degree is finally selected as the most suitable salt tide upward forecast model. The forecast factors of the excellent model are used to forecast the salt tide upward. In this embodiment, taking the monthly-scale forecast of the upward salt tide as an example, the establishment and application of the upward forecast model of the salt tide are as follows: different input variable scenarios are set for the upward forecast model of the salt tide, wherein: scenario 1 inputs hydrological factors, scenario 1 2 input hydrological coupled large-scale circulation climate factors, and scenario 3 input large-scale circulation climate factors. Input the three scenarios into the salt tide upward forecast model respectively, and through training and verification, the salt tide upward forecast model achieves the best fitting effect under the three scenarios. The fitting effect of the model is determined by the Nash efficiency coefficient (NSE) and the Coefficient (R ² ) was evaluated. The optimal model for upcasting of salt tide obtained by training was applied to the upcasting of salt tide, and the forecasting accuracy under three scenarios was compared.

FIG. 7 is a monthly-scale salt tide forecast diagram based on the hydrological influence factor forecast of the data output module of the salt tide upward forecasting system coupled with large-scale circulation climate information according to the embodiment of the present invention, and FIG. The monthly-scale salt tide forecast map based on the combination of hydrology and large-scale circulation climate influence factors of the data output module of the salt tide upward forecasting system of the scale circulation climate information.

In the embodiment of the present invention, if the salt tide upward forecasting system coupled with large-scale circulation climate information takes the monthly scale forecast as an example, the hydrological influencing factors such as flow and tidal level of the Modaomen waterway from 2005 to 2015 and the interdecadal hydrological factors of the Pacific Ocean are selected. Based on the data of large-scale circulation factors such as oscillation and Southern Oscillation Index, through random forest importance analysis, the degree of correlation between the uptake of the salt tide in the Modaomen Channel and its influencing factors at different time scales was explored, so as to identify the influence of the salt tide in the region. The key influencing factors of the backtracking. The salt tide upward forecast model is trained and verified by using the salt tide upward data and data sets, of which 80% of the salt tide upward data and data sets are used as the training set of the salt tide upward forecast model, and 20% of the salt tide upward data and data sets are used as the training set of the salt tide upward forecast model. As the validation set of the salt tide upward forecasting model. As shown in Figures 7 and 8, during the verification process, the salt tide upward forecast model based on the combined forecast of hydrology and large-scale circulation climate factors has the best fitting effect, and the obtained NSE and R2 values are 0.80 and 0.82, respectively. Therefore, coupling the large-scale circulation climate information can significantly improve the forecasting effect of the model, which is of great practical significance for guiding the upcasting of salt tides in data-deficient areas.

The salt tide upward forecasting system coupled with the large-scale circulation climate information provided by the invention can incorporate the large-scale circulation climate factors into the salt tide upward mid- and long-term forecast, indicating that the application of the large-scale circulation climate factors in the salt tide medium and long-term forecast under the influence of teleconnection has the advantages of Feasibility and effectiveness; in order to effectively avoid the redundancy of forecast information and the cross-correlation between forecast factors, the salt tide upward forecast system determines the forecast elements and delay time of salt tide upward at different time scales through comparison and selection. , and the implementation of the salt tide upward forecasting system has low data requirements and is easy to operate.

Embodiment 2:

The present invention also provides a salt tide upward forecasting method coupled with large-scale circulation climate information, comprising the following steps:

Obtain the hydrometeorological data that affects the uptake of the salty tide, and analyze the importance of the obtained hydrometeorological data through random forests to obtain the key influencing factors that affect the uptake of the salty tide;

According to the key influencing factors, the random forest algorithm is used to establish the salt tide upward forecast model, and the salt tide upward forecast model is verified, and the optimal model of the salt tide upward forecast is obtained;

Output the optimal model for upcasting of salt tide to predict the chloride content in the area affected by the upcast salt tide.

In the embodiment of the present invention, the steps of obtaining the key influencing factors affecting the upward trace of the salt tide by analyzing the acquired hydrometeorological data through random forest importance include:

Obtain the historical hydrological factors, large-scale circulation climatic factors and chlorine content of a certain estuary area, and obtain the hydrometeorological data of the salt tide upstream:

Through the estimation error and Gini impurity of the out-of-bag sample set, the random importance analysis was carried out on the hydrometeorological data affecting the uptrend of the salt tide, and the first and second importance degrees between each factor and the chloride content at different time scales were obtained. important degree set;

Sort each factor in the first and second importance degree sets according to their importance in descending order to obtain the corresponding first and second factor sets; and from the first factor set and the second factor set The same factors are selected as the key influencing factors affecting the uptake of salt tide.

In the embodiment of the present invention, the steps of obtaining the optimal model for the upcasting of salt tide include:

All key impact factors will be obtained to construct a dataset, and the dataset will be divided into training set and validation set;

The random forest algorithm is used to train the model on the training set, and the salt tide upward forecasting model is established;

Input the validation set into the salt tide upward forecast model, and output the Nash efficiency coefficient and determination coefficient corresponding to the validation set. If the Nash efficiency coefficient and the determination coefficient are both greater than the coefficient threshold, the optimal model of salt tide upward forecast is obtained.

It should be noted that the steps in the method of the second embodiment correspond to the module settings in the system of the first embodiment. The content of each module has been described in detail in the system of the first embodiment, and the details of the method of the second embodiment will not be described again. The content of the steps is explained in detail.

The above embodiments are only used to illustrate the technical solutions of the present invention, but not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that: The recorded technical solutions are modified, or some technical features thereof are equivalently replaced; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims (6)

1. a salt tide upward forecasting system coupled with large-scale circulation climate information, is characterized in that, comprises data input module, model forecast module and data output module; The data input module is used to obtain the hydrometeorological data that affects the uptake of the salt tide, and obtains the key influencing factors that affect the uptake of the salt tide by analyzing the acquired hydrometeorological data through the importance of random forests; The model forecasting module is used for establishing a salt tide upward forecasting model by using a random forest algorithm according to the key influencing factors, and verifying the salt tide upward forecasting model to obtain an optimal salt tide upward forecasting model; The data output module is used for outputting the optimal model for the upward forecasting of the salt tide to predict the chlorine content in the area affected by the upward salt tide; The data input module includes a data acquisition sub-module, a data analysis sub-module and a screening sub-module; The data acquisition sub-module is used to obtain the historical hydrological factors, large-scale circulation climate factors and chlorine content of the estuary area of a certain place, and obtain the hydrometeorological data of the salt tide upstream; The data analysis sub-module is used to carry out an important random forest analysis on the hydrometeorological data that affects the upstream salt tide through the estimation error and Gini impurity of the out-of-bag sample set, and obtain each factor and chlorine content at different time scales. between the first important degree set and the second important degree set; The screening sub-module is used to screen each factor in the first importance degree set and the second importance degree in descending order of importance, to obtain the corresponding first factor set and the second importance degree. factor set; and select the same factor from the first factor set and the second factor set as the key influencing factor affecting the uptake of salt tide; The model prediction module includes a data set classification sub-module, a model-building sub-module and a verification model sub-module; The data set classification submodule is used to obtain all the key influencing factors to construct a data set, and divide the data set into a training set and a verification set; The model building submodule is used to perform model training on the training set by using the random forest algorithm, and establish a salt tide upward forecasting model; The verification model sub-module is used to input the verification set into the salt tide upward forecasting model, and output the Nash efficiency coefficient and determination coefficient corresponding to the verification set, if both the Nash efficiency coefficient and the determination coefficient are If it is greater than the coefficient threshold, the optimal model for upcasting of salt tide is obtained. 2. The salt tide upward forecasting system coupled with large-scale circulation climate information according to claim 1, wherein the hydrological factors include flow and tidal level, and the large-scale circulation climate factors include the Pacific Interdecadal Oscillation and the Southern Wave moving index. 3. The salt tide upward forecasting system coupled with large-scale circulation climate information according to claim 1, characterized in that, the screening submodule is used for respectively evaluating the first important degree set and the second important degree set. After the factors of , are sorted in descending order of importance, the first two-thirds of the factors are screened out and correspondingly constructed into the first factor set and the second factor set. 4. The salt tide upward forecasting system coupled with large-scale circulation climate information according to claim 1, wherein the data set classification submodule is used to use 80% of the data sets as a training set, and 20% of the data sets as a validation set. 5 . The salt tide upward forecasting system coupled with large-scale circulation climate information according to claim 1 , wherein the coefficient threshold is 0.5. 6 . 6. A salt tide upward forecasting method coupled with large-scale circulation climate information, characterized in that it comprises the following steps: Obtain the hydrometeorological data affecting the uptake of the salty tide, and analyze the importance of the acquired hydrometeorological data through random forests to obtain the key influencing factors that affect the uptake of the salty tide; According to the key influencing factors, a random forest algorithm is used to establish a salt tide upward forecast model, and the salt tide upward forecast model is verified to obtain an optimal salt tide upward forecast model; outputting the optimal model for the upward forecasting of the salt tide to predict the chlorine content of the area affected by the upward salt tide; The steps to obtain the key influencing factors affecting the uptake of the salt tide by analyzing the acquired hydrometeorological data through the importance of random forests include: Obtain the historical hydrological factors, large-scale circulation climatic factors and chlorine content of a certain estuary area, and obtain the hydrometeorological data of the salt tide upstream: Based on the estimation error and Gini impurity of the out-of-bag sample set, the random forest importance analysis was carried out on the hydrometeorological data affecting the uptake of the salt tide, respectively, and the first importance degree set and the chloride content between each factor at different time scales were obtained. The second important degree set; After sorting each factor in the first importance degree set and the second importance degree degree according to the importance degree from large to small, the corresponding first factor set and the second factor set are obtained; and from the first factor set and the second factor set; The factor set and the second factor set select the same factor as the key influencing factor affecting the uptake of the salt tide; The steps to obtain the optimal model for upcasting of salt tide include: All the key impact factors will be obtained to construct a dataset, and the dataset will be divided into a training set and a validation set; The random forest algorithm is used to perform model training on the training set, and a salt tide upward forecasting model is established; Input the verification set into the salt tide upward forecast model, and output the Nash efficiency coefficient and determination coefficient corresponding to the verification set. If both the Nash efficiency coefficient and the determination coefficient are greater than the coefficient threshold, the salt tide upward forecast is obtained. optimal model.

Images