A real-time calculation method for estuarine flux based on observational data and machine learning

Abstract

The present invention discloses a real-time calculation method for estuary flux based on observation data and machine learning, comprising the following steps: S1, real-time collection of high-frequency hydrological observation data of the section where the estuary enters the sea; S2, using the 3σ principle to determine the threshold value, and eliminating the abnormal values ​​of the hydrological observation data; S3, using the machine learning method to interpolate the missing water level and flow rate data during the low water level period; S4, using the cross-sectional area, flow rate, and the angle between the section and the water flow direction corresponding to a certain moment in the estuary to calculate the flow at that moment, and then combining the measured data of the hourly average concentration of carbon and nitrogen forms to obtain the hourly carbon and nitrogen material flux of the estuary; S5, according to the hourly carbon and nitrogen material flux of the estuary, statistically analyzing the carbon and nitrogen cumulative flux within a certain period of time; the method is flexible and easy to use, can significantly improve the calculation accuracy and effectiveness of the carbon and nitrogen material flux of the estuary entering the sea, has the ability to quickly respond to environmental changes, and is suitable for the environment of highly dynamic changes in the estuary.

Description

Technical Field

[0001] The present invention belongs to the technical field of material flux calculation, and in particular relates to a real-time calculation method for estuary flux based on observation data and machine learning. Background Art

[0002] Monitoring and research on the flux of substances such as carbon and nitrogen in rivers has become an important basis for scientifically assessing land-sea interactions and ecological environmental evolution.

[0003] At present, the methods for calculating the flux of rivers entering the sea at home and abroad can be divided into two categories: the first category is based on a small amount of low-frequency water quality monitoring data and basin hydrological data to estimate the material flux in a certain period of time. Although a variety of methods for estimating the period flux based on measured concentrations have been proposed, the estimation error is large due to the low frequency of monitoring data. The second category is the mechanism model method, which quantitatively describes the details of the emission, migration and transformation process of substances, comprehensively considers the influencing factors such as meteorology, hydrology and land use, and simulates and predicts the flux of elements such as carbon and nitrogen at the outlet of the basin. This type of model relies on a large amount of parameter data. Due to the limited local monitoring data, the simulation step size is mainly on the daily, monthly or annual scale, with large errors and few practical applications. These methods are mainly aimed at river sections and cannot reflect the complex migration and transformation process of the estuary area and the final material flux into the sea.

[0004] In the estuary area, due to the coupling of runoff and tides, the hydrodynamic conditions are highly dynamic, and the concentrations of substances such as carbon and nitrogen show complex changing characteristics with the periodic reciprocating motion of the tide. The material flux in the estuary area is usually estimated by the product of the average concentration of the period and the net flow, and the mechanism model based on algorithms such as convection-diffusion. The existing methods all have the disadvantages of small monitoring data and low model accuracy. Therefore, the accurate quantification of carbon and nitrogen fluxes requires high-frequency online monitoring of the flow and material content of the water section. In addition, due to the influence of tides, the water level in the estuary area changes greatly. At low water levels, the monitoring instruments are easily exposed to the water surface, resulting in missing monitoring data, and it is necessary to find a suitable data interpolation method. In short, based on high-frequency hydrological and water quality synchronous monitoring data, establishing a real-time calculation method for the flux of substances such as carbon and nitrogen into the sea is the key to solving the problem of quantitative evaluation of estuary flux into the sea. Summary of the invention

[0005] To solve the above problems, the present invention proposes a real-time calculation method for estuary flux based on observation data and machine learning. The method is flexible and easy to use, can significantly improve the calculation accuracy and effectiveness of the flux of carbon and nitrogen substances from the estuary to the sea, has the ability to quickly respond to environmental changes, and is suitable for the highly dynamic environment of the estuary.

[0006] To achieve the above object, the present invention adopts the following technical solutions:

[0007] A real-time calculation method for estuary flux based on observational data and machine learning, comprising the following steps:

[0008] S1. Real-time collection of high-frequency hydrological observation data at the estuary section;

[0009] S2. Use the 3σ principle to determine the threshold and eliminate abnormal values ​​of hydrological observation data;

[0010] S3, using machine learning methods to interpolate missing water level and flow velocity data during low water periods;

[0011] S4. Calculate the flow rate at a certain moment using the cross-sectional area, flow velocity, and angle between the cross section and the water flow direction at the estuary, and then combine the measured data of the hourly average concentration of carbon and nitrogen forms to obtain the hourly carbon and nitrogen flux at the estuary;

[0012] The specific process of step S4 is:

[0013] S41. By measuring the water level and flow rate of the river, the amount of water passing through the cross section of the river per unit time is calculated; wherein, based on a set of in-situ observations of different water levels and cross-sectional areas in a complete tidal cycle, the relationship between different water levels and cross-sectional areas is fitted, and the calculation formula is:

[0014] ,

[0015] In the formula, S t is the cross-sectional area of ​​the observation section at time t, in m 2 ;h t is the water level at time t, in m; a, b, c are parameters determined based on specific cross-sectional characteristics;

[0016] S42. Calculate the instantaneous runoff of the observation section at the estuary at time t. The calculation formula is:

[0017] ,

[0018] In the formula, Q t is the instantaneous runoff of the observation section at time t, in m 3 / s;S t is the cross-sectional area of ​​the observation section at time t, in m 2 ;u t is the flow velocity at time t, in m / s; sinθ is the angle between the cross section and the water flow direction;

[0019] S43. Calculate the hourly carbon and nitrogen flux at the estuary based on the instantaneous runoff and the hourly average concentration of carbon and nitrogen forms at the observation section at time t. The calculation formula is:

[0020] ,

[0021] In the formula, Ft is the hourly carbon and nitrogen flux in the estuary, in kg / h; Q t is the instantaneous runoff of the observation section at time t, in m 3 / s;c t is the hourly average concentration of carbon and nitrogen forms in the observation section at time t, in mg / L; 3.6 is the unit conversion factor;

[0022] S5. Based on the hourly carbon and nitrogen fluxes at the estuary, statistically analyze the cumulative carbon and nitrogen fluxes over a certain period of time.

[0023] Preferably, the hydrological observation data in step S1 include water level, flow velocity, angle between the cross section and the water flow direction, and hourly average concentration of carbon and nitrogen forms.

[0024] Preferably, the specific process of step S2 is: for the high-frequency hydrological observation data of the estuary section entering the sea, the average value and standard deviation of the hourly hydrological observation data are calculated respectively, and then the upper threshold and lower threshold of the hydrological observation data are calculated according to the 3σ principle, and the calculation formula is:

[0025] ,

[0026] ,

[0027] In the formula, x high is the upper threshold of hydrological observation data; x low is the lower threshold of hydrological observation data; is the average value of hourly hydrological observation data; is the standard deviation of hourly hydrological observation data;

[0028] If the hydrological observation data x at time u u Less than x low or greater than x high , then x u were identified as outliers and removed.

[0029] Preferably, the specific process of step S3 is:

[0030] S31. Calculate the water level change rate based on the water level data of other tidal stations upstream; for the water level h at time t, t , select the water level data at time t, the first n moments and the last n moments, and construct a linear regression equation:

[0031] ,

[0032] In the formula, h is the water level; t is the time; b is the intercept of the regression equation; k is the slope of the regression equation, and the slope k is the water level change rate; the water level data at time t, the previous n times, and the next n times are expressed as ht-n , h t-n+1 , …,h t , …, h t+n-1 , h t+n ;

[0033] S32. Build a random forest model with the publicly released water level data and water level change rate of nearby tidal stations as input variables and the water level and flow velocity data of the current station as output variables, and predict the missing water level and flow velocity data of the current station, using the predicted values ​​as interpolation values.

[0034] Preferably, the specific process of step S5 is:

[0035] According to the integral of the hourly carbon and nitrogen flux at the estuary, the cumulative carbon and nitrogen flux within a certain period of time is obtained. The calculation formula is:

[0036] ,

[0037] In the formula, F 总 is the cumulative flux of carbon and nitrogen in a certain period of time, which can be a day, month or year; F t It is the hourly carbon and nitrogen flux in the estuary, in kg / h.

[0038] After adopting the above technical scheme, the present invention has the following beneficial effects: first, the real-time calculation method of estuary flux based on observation data and machine learning proposed in the present invention can accurately count the hourly carbon and nitrogen material fluxes in the estuary, and can also calculate the cumulative carbon and nitrogen fluxes in a certain period of time (day, month or year) according to the actual statistical needs by day, month or year, especially in estuary areas significantly affected by tides, the accuracy of flux calculation is significantly improved by combining high-frequency observation data with machine learning methods; secondly, the method has real-time and flexibility, can quickly respond to environmental changes, and provide accurate estuary flux data in time to support decision makers to respond quickly; finally, the present invention not only helps to deeply understand the laws of material exchange between rivers and oceans, and then supports the formulation and implementation of more effective and faster environmental protection measures, but also provides new ideas and technical means for the accurate calculation of estuary fluxes, which has important scientific value and broad application prospects. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 It is a schematic diagram of the process of the present invention. DETAILED DESCRIPTION

[0040] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0041] like Figure 1 As shown, a real-time calculation method of estuary flux based on observation data and machine learning includes the following steps:

[0042] S1. Real-time collection of high-frequency hydrological observation data at the estuary section;

[0043] The hydrological observation data in step S1 include water level, flow velocity, angle between the cross section and the water flow direction, and hourly average concentration of carbon and nitrogen forms;

[0044] S2. Use the 3σ principle to determine the threshold and eliminate abnormal values ​​of hydrological observation data;

[0045] The specific process of step S2 is: for the high-frequency hydrological observation data of the estuary section, the average value and standard deviation of the hourly hydrological observation data are calculated respectively, and then the upper threshold and lower threshold of the hydrological observation data are calculated according to the 3σ principle. The calculation formula is:

[0046] ,

[0047] ,

[0048] In the formula, x high is the upper threshold of hydrological observation data; x low is the lower threshold of hydrological observation data; is the average value of hourly hydrological observation data; is the standard deviation of hourly hydrological observation data;

[0049] If the hydrological observation data x at time u u Less than x low or greater than x high , then x u were identified as outliers and removed;

[0050] S3, using machine learning methods to interpolate missing water level and flow velocity data during low water periods;

[0051] The specific process of step S3 is:

[0052] S31. Calculate the water level change rate based on the water level data of other tidal stations upstream; for the water level h at time t, t , select the water level data at time t, the first n moments and the last n moments, and construct a linear regression equation:

[0053] ,

[0054] In the formula, h is the water level; t is the time; b is the intercept of the regression equation; k is the slope of the regression equation, and the slope k is the water level change rate; the water level data at time t, the previous n times, and the next n times are expressed as h t-n , h t-n+1 , …,h t , …, h t+n-1 , h t+n ;

[0055] S32, using the publicly released water level data and water level change rate of nearby tidal stations as input variables and the water level and flow velocity data of the current station as output variables to build a random forest model, and predict the missing water level and flow velocity data of the current station, and use the predicted values ​​as interpolation values;

[0056] S4. Calculate the flow rate at a certain moment using the cross-sectional area, flow velocity, and angle between the cross section and the water flow direction at the estuary, and then combine the measured data of the hourly average concentration of carbon and nitrogen forms to obtain the hourly carbon and nitrogen flux at the estuary;

[0057] The specific process of step S4 is:

[0058] S41. By measuring the water level and flow rate of the river, the amount of water passing through the cross section of the river per unit time is calculated; wherein, based on a set of in-situ observations of different water levels and cross-sectional areas in a complete tidal cycle, the relationship between different water levels and cross-sectional areas is fitted, and the calculation formula is:

[0059] ,

[0060] In the formula, S t is the cross-sectional area of ​​the observation section at time t, in m 2 ;h t is the water level at time t, in m; a, b, c are parameters determined based on specific cross-sectional characteristics;

[0061] S42. Calculate the instantaneous runoff of the observation section at the estuary at time t. The calculation formula is:

[0062] ,

[0063] In the formula, Q t is the instantaneous runoff of the observation section at time t, in m 3 / s;S t is the cross-sectional area of ​​the observation section at time t, in m 2 ;u t is the flow velocity at time t, in m / s; sinθ is the angle between the cross section and the water flow direction;

[0064] S43. Calculate the hourly carbon and nitrogen flux at the estuary based on the instantaneous runoff and the hourly average concentration of carbon and nitrogen forms at the observation section at time t. The calculation formula is:

[0065] ,

[0066] In the formula, F t is the hourly carbon and nitrogen flux in the estuary, in kg / h; Q t is the instantaneous runoff of the observation section at time t, in m 3 / s;c t is the hourly average concentration of carbon and nitrogen forms in the observation section at time t, in mg / L; 3.6 is the unit conversion factor;

[0067] S5. Statistically analyze the cumulative carbon and nitrogen fluxes over a period of time based on the hourly carbon and nitrogen fluxes at the estuary;

[0068] The specific process of step S5 is:

[0069] According to the integral of the hourly carbon and nitrogen flux at the estuary, the cumulative carbon and nitrogen flux within a certain period of time is obtained. The calculation formula is:

[0070] ,

[0071] In the formula, F 总 is the cumulative flux of carbon and nitrogen in a certain period of time, which can be a day, month or year; F t It is the hourly carbon and nitrogen flux in the estuary, in kg / h.

[0072] The above is only a preferred specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any changes or substitutions that can be easily thought of by a person skilled in the art within the technical scope disclosed by the present invention should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

Claims

1. A real-time calculation method for estuary flux based on observation data and machine learning, characterized in that: The following steps are involved: S1. Real-time collection of high-frequency hydrological observation data at the estuary section; S2. Use the 3σ principle to determine the threshold and eliminate abnormal values ​​of hydrological observation data; S3, using machine learning methods to interpolate missing water level and flow velocity data during low water periods; The specific process of step S3 is: S31. Calculate the water level change rate based on the water level data of other tidal stations upstream; for the water level h at time t, t , select the water level data at time t, the first n moments and the last n moments, and construct a linear regression equation: ， In the formula, h is the water level; t is the time; d is the intercept of the regression equation; k is the slope of the regression equation, and the slope k is the water level change rate; the water level data at time t, the first n times, and the last n times are expressed as h t-n , h t-n+1 , …, h t ,…, h t+n-1 , h t+n ; S32, using the publicly released water level data and water level change rate of nearby tidal stations as input variables and the water level and flow velocity data of the current station as output variables to build a random forest model, and predict the missing water level and flow velocity data of the current station, and use the predicted values ​​as interpolation values; S4. Calculate the flow rate at a certain moment using the cross-sectional area, flow velocity, and angle between the cross section and the water flow direction at the estuary, and then combine the measured data of the hourly average concentration of carbon and nitrogen forms to obtain the hourly carbon and nitrogen flux at the estuary; The specific process of step S4 is: S41. By measuring the water level and flow rate of the river, the amount of water passing through the cross section of the river per unit time is calculated; wherein, based on a set of in-situ observations of different water levels and cross-sectional areas in a complete tidal cycle, the relationship between different water levels and cross-sectional areas is fitted, and the calculation formula is: ， In the formula, S t is the cross-sectional area of ​​the observation section at time t, in m 2 ;h t is the water level at time t, in m; a, b, c are parameters determined based on specific cross-sectional characteristics; S42. Calculate the instantaneous runoff of the observation section at the estuary at time t. The calculation formula is: ， In the formula, Q t is the instantaneous runoff of the observation section at time t, in m 3 / s;S t is the cross-sectional area of ​​the observation section at time t, in m 2 ;u t is the flow velocity at time t, in m / s; sinθ is the angle between the cross section and the water flow direction; S43. Calculate the hourly carbon and nitrogen flux at the estuary based on the instantaneous runoff and the hourly average concentration of carbon and nitrogen forms at the observation section at time t. The calculation formula is: ， In the formula, F t is the hourly carbon and nitrogen flux in the estuary, in kg / h; Q t is the instantaneous runoff of the observation section at time t, in m 3 / s;c t is the hourly average concentration of carbon and nitrogen forms in the observation section at time t, in mg / L; 3.6 is the unit conversion factor; S5. Based on the hourly carbon and nitrogen fluxes at the estuary, statistically analyze the cumulative carbon and nitrogen fluxes over a certain period of time.

2. A method for real-time calculation of estuary flux based on observation data and machine learning as claimed in claim 1, characterized in that: The hydrological observation data in step S1 include water level, flow velocity, angle between the cross section and the water flow direction, and hourly average concentration of carbon and nitrogen forms.

3. A method for real-time calculation of estuary flux based on observation data and machine learning as claimed in claim 1, characterized in that: The specific process of step S2 is: for the high-frequency hydrological observation data of the estuary section, the average value and standard deviation of the hourly hydrological observation data are calculated respectively, and then the upper threshold and lower threshold of the hydrological observation data are calculated according to the 3σ principle. The calculation formula is: ， ， In the formula, x high is the upper threshold of hydrological observation data; x low is the lower threshold of hydrological observation data; is the average value of hourly hydrological observation data; is the standard deviation of hourly hydrological observation data; If the hydrological observation data x at time u u Less than x low or greater than x high , then x u were identified as outliers and removed.

4. A method for real-time calculation of estuary flux based on observation data and machine learning as claimed in claim 1, characterized in that: The specific process of step S5 is: According to the integral of the hourly carbon and nitrogen flux at the estuary, the cumulative carbon and nitrogen flux within a certain period of time is obtained. The calculation formula is: ， In the formula, F 总 is the cumulative flux of carbon and nitrogen in a certain period of time, which can be a day, month or year; F t It is the hourly carbon and nitrogen flux in the estuary, in kg / h.

Images