Method for monitoring and evaluating long-term hydrological effect of grassed ditch

Abstract

The invention provides a method for monitoring and evaluating the long-term hydrological effect of a grassed ditch. The method comprises the following steps: screening by using a data recoverability evaluation model to obtain a future representative rainfall event of the grassed ditch to be evaluated; collecting meteorological parameters and hydrological performance parameters of representative rainfall events; calculating by utilizing a grassed swale instant saturated permeability coefficient inversion method to obtain grassed swale instant saturated permeability coefficient values under a representative rainfall event, and forming a second training data set with the meteorological parameter values; based on the second training data set, establishing an instant saturated permeability coefficient dynamic prediction model; predicting by using the real-time saturated permeability coefficient dynamic prediction model to obtain a predicted value of the real-time saturated permeability coefficient of the grassed swale under a new rainfall event except the representative rainfall event; and inputting the predicted value of the immediate saturated permeability coefficient of the grassed swale into the hydrological model of the grassed swale to obtain a hydrological performance parameter simulation result. The method provided by the invention can accurately simulate the long-term hydrological performance of the grassed swallow under different climate conditions.

Description

Technical Field

[0001] The present invention belongs to the technical field of municipal facility management services, and in particular relates to a method for monitoring and evaluating the long-term hydrological effects of grass-planting ditches. Background Art

[0002] Grassed ditches are one of the most widely used municipal facilities for stormwater management both domestically and internationally. Accurately predicting the changing patterns of their runoff control capacity (i.e., their hydrological effects) is a crucial foundation for planning, design, and operation and maintenance. The saturated permeability coefficient is the primary controlling factor in the hydrological effects of grassed ditches, but existing monitoring or calibration methods cannot quantitatively assess their heterogeneity along the ditches and their long-term variations. Current research on the long-term hydrological effects of grassed ditches faces the following challenges: 1. Spatial uncertainty: Traditional grass-planted ditches are generally long, and the heterogeneity of their fill material leads to variations in the saturated permeability coefficient along the ditches. Single-point measurements cannot accurately represent the permeability performance of the entire system. Currently, there is no reliable method to calculate the equivalent value of the saturated permeability coefficient of a single facility, resulting in a large discrepancy between conventional hydrological effect simulation results and actual facility performance.

[0003] 2. Temporal uncertainty: The saturated permeability coefficient of grass-planted ditches exhibits long-term decline and periodic fluctuations due to runoff erosion, water evaporation, and vegetation root growth. However, existing infiltration theoretical models and hydrological effect simulation software do not take this into account. There is also a lack of quantitative analysis and prediction methods for long-term changes in the saturated permeability coefficient, which cannot meet the needs of long-term or even full life cycle assessment of the hydrological effects of grass-planted ditches.

[0004] 3. Methodological uncertainty: Existing studies generally study the long-term hydrological effects of grass-planted gullies by inferring the instantaneous saturated permeability coefficient from the inlet and outlet curves of the grass-planted gullies during rainfall events. However, in current practice, there is a lack of methods to screen relevant meteorological and hydrological performance data of grass-planted gullies during rainfall events, resulting in poor reliability of research results. Summary of the Invention

[0005] The technical problem to be solved by the present invention is to provide a method for monitoring and evaluating the long-term hydrological effects of grass-planted ditches, which can accurately simulate the long-term hydrological performance of grass-planted ditches under different climatic conditions.

[0006] In order to solve the above technical problems, the present invention adopts the following technical solutions: The present invention provides a method for monitoring and evaluating the long-term hydrological effects of a grass-planted ditch, comprising the following steps: Step 10: Using the data accessibility assessment model, evaluate and screen future rainfall events in the area where the grass-planting ditch to be assessed is located to obtain representative rainfall events; Step 20: When a representative rainfall event occurs, meteorological parameters and hydrological performance parameters of the representative rainfall event are collected; the instantaneous saturated permeability coefficient value of the grass-planted ditch under the representative rainfall event is calculated using the inversion method of the instantaneous saturated permeability coefficient of the grass-planted ditch, and the instantaneous saturated permeability coefficient value of the grass-planted ditch under the representative rainfall event is combined with the meteorological parameter values ​​of the representative rainfall event to form a second training data set; Step 30: Based on the second training data set, an XGBoost algorithm is used to establish a dynamic prediction model for instantaneous saturated permeability coefficient; Step 40, using the dynamic prediction model of the instantaneous saturated permeability coefficient, predicting the instantaneous saturated permeability coefficient of the grass-planted ditch under a new rainfall event other than the representative rainfall event, to obtain a predicted value of the instantaneous saturated permeability coefficient of the grass-planted ditch; Step 50: input the predicted value of the instantaneous saturated permeability coefficient of the grass-planted ditch into the grass-planted ditch hydrological model to obtain the simulation result of the hydrological performance parameters of the grass-planted ditch to be evaluated.

[0007] As a further improvement of the present invention, the meteorological parameters include the number of sunny days before rainfall, temperature, humidity, rainfall, rainfall duration, cumulative rainfall and seasonal factors; the hydrological performance parameters include the depth of water accumulation in the grass-planting ditch and the outflow flow of the grass-planting ditch.

[0008] As a further improvement of the present invention, the instantaneous saturated permeability dynamic prediction model is a dynamic prediction model of the instantaneous saturated permeability of the grass-planted ditch with respect to the previous number of sunny days, temperature, humidity, rainfall, rainfall duration, accumulated rainfall and seasonal factors.

[0009] As a further improvement of the present invention, the process of constructing the data admissibility evaluation model includes: The instantaneous saturated permeability coefficient of the grass-planted ditch is calculated using the inversion method of the instantaneous saturated permeability coefficient of the grass-planted ditch under no less than 100 historical rainfall events in the past five years in the area where the grass-planted ditch is located. The instantaneous saturated permeability coefficient of the grass-planted ditch is calculated and combined with the meteorological parameter values ​​of the historical rainfall events to form the first training data subset; Resampling the first training data subset to randomly generate N*M groups of short data sequences containing N rainfall event groups with a total of M categories; calculating the average value of the meteorological parameters corresponding to each group of short data sequences, and performing level mapping on the average values ​​of the meteorological parameters calculated from the N*M groups of data sequences to classify them into three categories: low, medium, and high; wherein N is a multiple of 1000, and M is an integer not greater than the total number of historical rainfall events; Based on N*M groups of short data sequences, the XGBoost algorithm is used to establish dynamic prediction models for the instantaneous saturated permeability coefficient, and the relative error compared with the dynamic prediction model for the instantaneous saturated permeability coefficient established based on the first training data subset is calculated; if the relative error is less than 10%, the group of short data sequences is considered representative; otherwise, the group of short data sequences is considered unrepresentative; Based on the rank mapping results of N*M groups of short data sequences and the representative classification results, a data feasibility assessment model is established using the Bernoulli Naive Bayes method.

[0010] As a further improvement of the present invention, step 10 specifically includes: Calculate the average values ​​of meteorological parameters of historical rainfall events in the area where the grass-planting ditch to be evaluated is located, perform level mapping on the calculated average values ​​of meteorological parameters, and classify them into three categories: low, medium, and high. Then update the mapping results into the data accessibility assessment model. The data accessibility assessment model is used to calculate the probability that the future rainfall event in the area where the grass-planting ditch is located is representative of the meteorological parameter values ​​under the meteorological forecast rainfall event. The first rainfall event of each month is assumed to be representative. If it is an extreme rainfall event, it is postponed to the next one. If the probability of the data sequence being representative increases after adding the meteorological parameter values ​​of a rainfall event, then the rainfall event is representative; otherwise, the rainfall event is not representative.

[0011] As a further improvement of the present invention, the method for constructing the hydrological model of the grass-planted ditch is as follows: based on equations (1) and (2), considering the time-varying characteristics of the saturated permeability coefficient, proposing the consistency hypothesis of the vertical variation of the soil moisture content in the ditch body and introducing the static pressure head parameter, the hydrological model of the grass-planted ditch is established; Formula (1) Formula (2) Where, f represents the infiltration rate; K s ( x , t ) represents the saturated permeability coefficient, x Indicates the influencing factors, t Indicates time; Z f represents the penetration path length; ΔP Indicates the pressure difference across the permeation unit; K ( θ ) represents the unsaturated permeability coefficient; K s represents the saturated permeability coefficient; θ Indicates the volumetric moisture content of soil; θ r Indicates the residual volume moisture content of soil; θ s represents the saturated volumetric moisture content of soil; α represents the empirical fitting parameter or curve shape parameter; n represents the empirical fitting parameter or curve shape parameter.

[0012] As a further improvement of the present invention, in the grass-planted ditch hydrological model, the calculation process of the wet stage includes: As the rainfall progresses, surface runoff continues to flow in. The moisture content of the packing layer continues to rise under the combined action of top water inflow and bottom drainage until it reaches the saturated moisture content. The outflow from the bottom of the packing layer serves as the inflow of the drainage layer. Specifically: Formula (3) Where, It represents the surface water volume of the grass-planted ditch at the end of unit time when there is no infiltration at the current moment; It represents the surface water volume of the grass-planting ditch before the current moment; Indicates rainfall intensity; Represents a unit time interval; Formula (4) Where, Indicates the water depth in the grass-planting ditch at the end of unit time when there is no infiltration at the current moment; It represents the upper surface area of ​​the filler layer when the aquifer is rectangular; Formula (5) Where, Indicates the infiltration rate of the packing layer at the end of unit time when there is no infiltration; It represents the saturated permeability coefficient of the packing layer; Indicates the thickness of the filler layer; Formula (6) Where, It represents the volume of water accumulated in the grass-planted ditch after infiltration per unit time at the infiltration rate without infiltration; It represents the overflow volume of the grass-planted ditch per unit time; make , set the minimum error range and perform iterative calculation to obtain the volume and depth of water in the grass-planting ditch; Formula (7) Formula (8) Where, Indicates the infiltration amount of the filler layer when the bottom is not drained at the current moment; Indicates the infiltration amount of the filler layer relative to the residual moisture content before the current moment; Indicates the moisture content of the packing layer when the bottom is not drained at the current moment; Formula (9) Where, It indicates the unsaturated permeability coefficient of the packing layer when the bottom is not drained at the current moment; Indicates the residual volume water content of the packing layer; Indicates the saturated volume water content of the packing layer; Indicates the saturated permeability coefficient of the packing layer; Formula (10) Where, Indicates the water outflow rate at the bottom of the packing layer at the current moment; Will As the water inflow rate of the drainage layer, the same iterative method as that of the filler layer is adopted to obtain the water volume and depth at each moment on the surface of the drainage layer; Formula (11) Formula (12) Where, It represents the water outflow rate at the bottom of the packing layer calculated based on the current moment when the bottom is not drained, and the infiltration rate of the packing layer after drainage; It indicates the water discharge rate at the bottom of the packing layer calculated based on the current moment when the bottom is not drained, and the moisture content of the packing layer after drainage; It represents the water outflow rate at the bottom of the packing layer calculated based on the current moment when the bottom is not drained, and the infiltration rate of the drainage layer after drainage; represents the upper surface area of ​​the drainage layer; make , set the minimum error range and perform iterative calculation to obtain the infiltration volume and moisture content of the packing layer at the current moment; The same iterative method as that for the filler layer is adopted to calculate the infiltration rate and moisture content of the drainage layer.

[0013] As a further improvement of the present invention, in the grass-planted ditch hydrological model, the calculation process of the drainage stage includes: When the rainfall process has not yet ended and the drainage layer is full, the grass-planted ditch begins to drain water outward through the perforated drainage pipe. Assuming that the permeability of the drainage layer is large enough to discharge the water from the filler layer in time, the outflow rate of the drainage pipe per unit time is equal to the inflow rate at the bottom of the filler layer minus the infiltration rate from the bottom of the grass-planted ditch into the natural soil. The average infiltration rate from the bottom to the natural soil is considered to be approximately equal to the saturated permeability coefficient of the natural soil. The outflow rate is calculated using formula (13): Formula (13) Where, It represents the average outflow rate at the outlet of the drainage pipe of the grass-planting ditch; Indicates the saturated permeability coefficient of the natural soil at the bottom of the grass-planting ditch; represents the surface area of ​​the drainage layer; Use formula (14) to calculate the pipe flow velocity along the drainage blind pipe and the local head loss: Formula (14) Formula (15) Where, The pipe flow velocity represents the loss of head along the drainage blind leg and the local head loss; H0 represents the total head including the travel head; Indicates the correction coefficient, generally taken as 1.0; Indicates the head loss, including the head loss along the way and the local head loss; represents the resistance coefficient along the way; Indicates the length of the pipeline; Indicates the pipe diameter; Indicates the cross-sectional length of the water flow in the pipe; Indicates the length of the pipeline; represents the local resistance coefficient; Indicates the flow velocity in the pipe; After the rainfall process ends, the surface runoff also stops, and the grass-planted ditch enters the emptying stage. At the beginning, there is still water on the surface of the grass-planted ditch, and the infiltration meets the saturated infiltration. The infiltration rate of the packing layer is calculated using formula (7); after the surface water disappears, the packing layer begins to gradually empty in an unsaturated state. The emptying rate of the packing layer is calculated using formula (16): Formula (16) Where, Indicates the emptying rate of the packing layer, represents the unsaturated permeability coefficient of the packing layer, Indicates the moisture content of the packing layer at the current moment; At this time, the drainage layer is still in a saturated state. When the moisture content of the drainage layer at the current moment is less than the saturated moisture content of the drainage layer, the outlet stops discharging water, and the water in the drainage layer slowly seeps into the underground soil. When the moisture content of the filler layer and the drainage layer drops to their respective field moisture contents, the rainfall infiltration process ends.

[0014] As a further improvement of the present invention, in step 20, the instantaneous saturated permeability coefficient value of the grass-planted ditch of a representative rainfall event is calculated using the instantaneous saturated permeability coefficient inversion method of the grass-planted ditch, specifically including: Taking the water accumulation depth in the grass ditch or the outflow flow of the grass ditch under a representative rainfall event as the observation variable, the initial assumed instantaneous saturated permeability coefficient value of the grass ditch is substituted into the grass ditch hydrological model to obtain the simulation result of the observation variable; the ensemble Kalman filter algorithm is used to perform data assimilation on the observation results and the simulation results to generate an estimated value of the instantaneous saturated permeability coefficient, which is then substituted into the grass ditch hydrological model for simulation calculation. After multiple iterations, the converged instantaneous saturated permeability coefficient value of the grass ditch is obtained.

[0015] As a further improvement of the present invention, in step 20, the data assimilation is performed using an ensemble Kalman filter algorithm, which specifically includes: Based on the state transfer equation shown in formula (17) and the observation equation shown in formula (18): Formula (17) Formula (18) Where, represents the predicted value of the kth set of parameters of the grass-growing ditch hydrological model at time i+1; represents the updated value of the kth set of parameters of the grass-planted ditch hydrological model at time i; Represents the predicted value of the kth set of state variables at time i+1; represents the updated value of the kth set number of the state variable at time i; represents the prediction operator; Indicates the driving data of the grass-growing ditch hydrological model; represents the independent white noise of the parameters of the grass-growing ditch hydrological model; represents the independent white noise of the state variables of the grass-growing ditch hydrological model; It represents the kth set value of the outflow or water depth of the grass-planting ditch hydrological simulation at time i+1; h represents the observation operator; Represents the error term, which has a mean of 0 and a variance of Normal distribution; Assimilation is performed using equations (19) to (21): Formula (19) Formula (20) Formula (21) Where, represents the predicted value of the kth set number at the i+1th time; represents the updated value of the kth set at the i-th moment; represents the white noise of the kth set; represents the observation value of the kth set; represents the observation error of the kth set; represents the Kalman gain, which is the weight relationship between the predicted value and the observed value, and is calculated using Equations (22) to (24): Formula (22) Formula (23) Formula (24) Where, represents the covariance matrix of the predicted state variables; Represents the covariance matrix of the error of the observed variable prediction value; represents the ensemble mean of the predicted state variables; represents the ensemble mean of the predicted values ​​of the observed variables; N represents the number of ensembles.

[0016] Compared with the prior art, the technical solution of the present invention has the following beneficial effects: (1) The present invention provides a method for monitoring and evaluating the long-term hydrological effects of grass-planted ditches. The method uses a data collectibility assessment model to perform data sampling, thereby improving the predictability of sampling and realizing lightweight monitoring from more to less. The method uses an instant saturated permeability inversion method for grass-planted ditches to invert the instant saturated permeability, thereby realizing lightweight monitoring from difficult to easy. The monitored meteorological data and the inverted value of the instant saturated permeability are substituted into the instant saturated permeability dynamic prediction model to calibrate the model variable parameters, and the calibrated model is used to predict the instant saturated permeability under future rainfall events, thereby realizing simulation from static to dynamic. The predicted value of the instant saturated permeability under future rainfall events is substituted into the grass-planted ditch hydrological model to simulate the hydrological performance and realize the long-term hydrological effect evaluation of the grass-planted ditch.

[0017] (2) The present invention provides a method for monitoring and evaluating the long-term hydrological effects of grass-planted ditches. A data accessibility evaluation model based on a Bayesian network is established. The data accessibility evaluation model is used to screen out representative measured data groups for training the instant saturated permeability coefficient prediction model, thereby achieving predictability and lightweight data sampling. While ensuring prediction accuracy, the measurement workload is effectively reduced, thereby improving the practicality of the method for evaluating the long-term hydrological effects of grass-planted ditches.

[0018] (3) The present invention provides a method for monitoring and evaluating the long-term hydrological effects of grass-planted ditches, proposes the concept of the instantaneous saturated permeability coefficient of grass-planted ditches under rainfall events, and uses the water depth or outflow flow in the grass-planted ditches under rainfall events, which are relatively easy to measure, as the observation variable, to establish an instantaneous saturated permeability coefficient inversion method that couples the grass-planted ditch hydrological model with the ensemble Kalman filter algorithm; the instantaneous saturated permeability coefficient value of the grass-planted ditch under a representative rainfall event is calculated using the instantaneous saturated permeability coefficient inversion method, which solves the problem that the saturated permeability coefficient of the grass-planted ditch under rainfall events varies greatly along the course, the measured value of a local single point cannot represent the overall performance, and the error of the simulation calculation using the simple average value of the measured values ​​of multiple points is large, thereby improving the accuracy of the dynamic prediction model of the instantaneous saturated permeability coefficient obtained by training based on the instantaneous saturated permeability coefficient of the grass-planted ditch.

[0019] (4) The present invention provides a method for monitoring and evaluating the long-term hydrological effects of grass-planting ditches. The inversion result of the instant saturated permeability coefficient is used as the dependent variable, and the measured data of the meteorological parameters are used as the independent variable. The XGBoost algorithm is used to establish a dynamic prediction model for the instant saturated permeability coefficient. Compared with the traditional machine learning model, the demand for training samples is smaller while ensuring the performance of the model, which is suitable for the situation where the sample size is not large in the present invention. Secondly, the XGBoost model is more interpretable, and the decision-making process of the model can be understood through feature importance and tree structure visualization. In the method of the present invention, the importance of each feature can be intuitively given by using this method, and some features with less importance can be deleted as appropriate without affecting the overall prediction ability of the model.

[0020] (5) The present invention provides a method for monitoring and evaluating the long-term hydrological effects of grass-planted gullies. Based on the Darcy formula and the van Genuchten-Mualem soil-water characteristic curve equation, a grass-planted gully hydrological model is established by considering the time-varying characteristics of the saturated permeability coefficient. The static pressure head parameter is introduced into the coupling process between the Darcy formula and the soil-water characteristic curve equation, so that the grass-planted gully hydrological model can simulate the infiltration of water in the grass-planted gully during rainfall events. The hypothesis of the consistency of the vertical variation of the soil moisture content in the ditch is proposed, so that the grass-planted ditch hydrological model can simulate the infiltration of the ditch soil when it is unsaturated in the initial stage of rainfall, thereby reducing the difficulty of calculating the position of the wetting front under the unsaturated state. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 A flow chart of the method for monitoring and evaluating the long-term hydrological effects of grass-planted ditches provided by the present invention; Figure 2 Schematic diagram of the structure of the data admissibility assessment model in the method of the present invention; Figure 3 This is a flow chart of the instant saturated permeability inversion method for the grass-planted ditch in the method of the present invention; Figure 4 This is a calculation flow chart of the grass-planted ditch hydrological model in the method of the present invention. DETAILED DESCRIPTION

[0022] The technical solution of the present invention is described in detail below.

[0023] The embodiment of the present invention provides a method for monitoring and evaluating the long-term hydrological effects of a grass-planted ditch. Figure 1 As shown, the following steps are included: Step 10: Use the data accessibility assessment model to evaluate and screen future rainfall events in the area where the grass-planting ditch to be assessed is located, and obtain representative rainfall events.

[0024] Step 20, when a representative rainfall event occurs, collect meteorological parameters and hydrological performance parameters of the representative rainfall event. The instantaneous saturated permeability coefficient value of the grass-planted ditch under the representative rainfall event is calculated using the inversion method of the instantaneous saturated permeability coefficient of the grass-planted ditch, and is combined with the meteorological parameter values ​​of the representative rainfall event to form a second training data set. Preferably, the meteorological parameters include the number of sunny days before the rainfall, temperature, humidity, rainfall, rainfall duration, cumulative rainfall, and seasonal factors. The hydrological performance parameters include the depth of water accumulation in the grass-planted ditch and the outflow flow of the grass-planted ditch.

[0025] Step 30: Based on the second training data set, an XGBoost algorithm is used to establish a dynamic prediction model for the instantaneous saturated permeability coefficient.

[0026] Step 40 , using the dynamic prediction model of the instantaneous saturated permeability coefficient, predict the instantaneous saturated permeability coefficient of the grass-planted ditch under a new rainfall event other than the representative rainfall event, and obtain a predicted value of the instantaneous saturated permeability coefficient of the grass-planted ditch.

[0027] Step 50: input the predicted value of the instantaneous saturated permeability coefficient of the grass-planted ditch into the grass-planted ditch hydrological model to obtain the simulation result of the hydrological performance parameters of the grass-planted ditch to be evaluated.

[0028] Preferably, the method for constructing the grass-planted ditch hydrological model in step 50 is: Based on the Darcy formula shown in formula (1) and the van Genuchten-Mualem soil-water characteristic curve equation shown in formula (2), the time-varying characteristics of the saturated permeability coefficient are considered, the vertical variation consistency hypothesis of the ditch soil moisture content is proposed, and the static pressure head parameter is introduced to establish a hydrological model for the grass-planted ditch.

[0029] Formula (1) Formula (2) Where, f represents the infiltration rate;K s ( x , t ) represents the saturated permeability coefficient, x Indicates the influencing factors, t Indicates time; Z f represents the penetration path length; ΔP Indicates the pressure difference across the permeation unit; K ( θ ) represents the unsaturated permeability coefficient; K s represents the saturated permeability coefficient; θ Indicates the volumetric moisture content of soil; θ r Indicates the residual volume moisture content of soil; θ s represents the saturated volumetric moisture content of soil; α represents the empirical fitting parameter or curve shape parameter; n represents the empirical fitting parameter or curve shape parameter.

[0030] The construction of the grass-planted ditch hydrological model mainly includes the wetting process and the drainage process, such as Figure 4 shown.

[0031] The calculation process of the wetting process includes: As the rainfall progresses, surface runoff continues to flow in, and the moisture content of the filler layer continues to rise under the combined action of top water inflow and bottom drainage until it reaches the saturated moisture content. The outflow from the bottom of the filler layer serves as the inflow of the drainage layer.

[0032] Specifically: Formula (3) Where, It represents the surface water volume of the grass-planted ditch at the end of unit time when there is no infiltration at the current moment; It represents the surface water volume of the grass-planting ditch before the current moment; Indicates rainfall intensity; Indicates a unit time interval.

[0033] Formula (4) Where, Indicates the water depth in the grass-planting ditch at the end of unit time when there is no infiltration at the current moment; It represents the upper surface area of ​​the filler layer when the aquifer is rectangular.

[0034] Formula (5) Where, Indicates the infiltration rate of the packing layer at the end of unit time when there is no infiltration; It represents the saturated permeability coefficient of the packing layer; Indicates the thickness of the filler layer.

[0035] Formula (6) Where, It represents the volume of water accumulated in the grass-planted ditch after infiltration per unit time at the infiltration rate without infiltration; It represents the overflow volume of the grass-planted ditch per unit time.

[0036] make , set the minimum error range and perform iterative calculations to obtain the volume and depth of water accumulation in the grass-planting ditch.

[0037] Formula (7) Formula (8) Where, Indicates the infiltration amount of the filler layer when the bottom is not drained at the current moment; Indicates the infiltration amount of the filler layer relative to the residual moisture content before the current moment; Indicates the moisture content of the packing layer when the bottom is not drained at the current moment.

[0038] Formula (9) Where, It indicates the unsaturated permeability coefficient of the packing layer when the bottom is not drained at the current moment; Indicates the residual volume water content of the packing layer; Indicates the saturated volume water content of the packing layer; It represents the saturated permeability coefficient of the packing layer.

[0039] Formula (10) Where, Indicates the water outflow rate at the bottom of the packing layer at the current moment.

[0040] Will As the water inflow rate of the drainage layer, the same iterative method as the filler layer is used to obtain the water volume and depth on the surface of the drainage layer at each moment.

[0041] Formula (11) Formula (12) Where, It represents the water outflow rate at the bottom of the packing layer calculated based on the current moment when the bottom is not drained, and the infiltration rate of the packing layer after drainage; It indicates the water discharge rate at the bottom of the packing layer calculated based on the current moment when the bottom is not drained, and the moisture content of the packing layer after drainage; It represents the water outflow rate at the bottom of the packing layer calculated based on the current moment when the bottom is not drained, and the infiltration rate of the drainage layer after drainage; Represents the upper surface area of ​​the drainage layer.

[0042] make , set the minimum error range and perform iterative calculations to obtain the infiltration volume and moisture content of the filler layer at the current moment.

[0043] The same iterative method as that for the filler layer is used to calculate the infiltration rate and moisture content of the drainage layer.

[0044] The calculation process of the drainage process includes: When the rainfall process has not yet ended and the drainage layer is full, the grass-planted ditch begins to drain water outward through the perforated drainage pipe. Assuming that the permeability of the drainage layer is large enough to discharge the water from the filler layer in time, the outflow rate of the drainage pipe per unit time is equal to the inflow rate at the bottom of the filler layer minus the infiltration rate from the bottom of the grass-planted ditch into the natural soil. The average infiltration rate from the bottom to the natural soil is considered to be approximately equal to the saturated permeability coefficient of the natural soil. The outflow rate is calculated using formula (13): Formula (13) Where, It represents the average outflow rate at the outlet of the drainage pipe of the grass-planting ditch; Indicates the saturated permeability coefficient of the natural soil at the bottom of the grass-planting ditch; represents the surface area of ​​the drainage layer; Use formula (14) to calculate the pipe flow velocity along the drainage blind pipe and the local head loss: Formula (14) Formula (15) Where, The pipe flow velocity represents the loss of head along the drainage blind leg and the local head loss; H0 represents the total head including the travel head; Indicates the correction coefficient, generally taken as 1.0; Indicates the head loss, including the head loss along the way and the local head loss; represents the resistance coefficient along the way; Indicates the length of the pipeline; Indicates the pipe diameter; Indicates the cross-sectional length of the water flow in the pipe; Indicates the length of the pipeline; represents the local resistance coefficient; Indicates the flow velocity in the pipe.

[0045] After the rainfall process ends, the surface runoff also stops, and the grass-planted ditch enters the emptying stage. At the beginning, there is still water on the surface of the grass-planted ditch, and the infiltration meets the saturated infiltration. The infiltration rate of the packing layer is calculated using formula (7); after the surface water disappears, the packing layer begins to gradually empty in an unsaturated state. The emptying rate of the packing layer is calculated using formula (16): Formula (16) Where, Indicates the emptying rate of the packing layer, represents the unsaturated permeability coefficient of the packing layer, Indicates the moisture content of the packing layer at the current moment.

[0046] At this time, the drainage layer is still in a saturated state. When the moisture content of the drainage layer at the current moment is less than the saturated moisture content of the drainage layer, the outlet stops discharging water, and the water in the drainage layer slowly seeps into the underground soil. When the moisture content of the filler layer and the drainage layer drops to their respective field moisture contents, the rainfall infiltration process ends.

[0047] Considering that the saturated permeability coefficient of the filling layer of the grass-planted ditch is a key factor affecting the hydrological effect of the grass-planted ditch, it is also the most important input parameter in the grass-planted ditch hydrological model. However, the grass-planted ditch is generally long and its saturated permeability coefficient varies along the ditch. Therefore, the saturated permeability coefficient value measured at a single point cannot reflect the overall infiltration performance of the grass-planted ditch. The commonly used method of taking the average value of multiple-point measurements to evaluate the overall infiltration performance of the grass-planted ditch also has a large error. Therefore, the results obtained by simulating the hydrological effect using the measured value of the saturated permeability coefficient are unreliable. At the same time, the existing grass-planted ditch hydrological model usually assumes that the saturated permeability coefficient value is constant and does not change with time, or assigns a constant attenuation coefficient to characterize the long-term decreasing characteristics of the saturated permeability coefficient. However, in a changing environment, under the influence of factors such as rainfall, temperature, humidity and plant growth, the saturated permeability coefficient of the grass-planted ditch shows a dynamic change trend.

[0048] Preferably, the embodiment of the present invention uses the saturated permeability coefficient value that can characterize the overall infiltration performance under a given rainfall event, which is inverted from the measured data of the hydrological performance of the grass-planted ditch, as the instantaneous saturated permeability coefficient under the rainfall, based on the along-the-way variation and time-varying characteristics of the saturated permeability coefficient of the grass-planted ditch.

[0049] In step 20, the instantaneous saturated permeability coefficient of the grass-planted ditch under a representative rainfall event is calculated using the inversion method of the instantaneous saturated permeability coefficient of the grass-planted ditch, specifically: Using the water depth or outflow rate within the grassed ditch during a representative rainfall event as the observed variable, the initially assumed instantaneous saturated permeability coefficient of the grassed ditch was substituted into the grassed ditch hydrological model to obtain simulation results for the observed variable. The ensemble Kalman filter algorithm was used to assimilate the observed and simulated results to generate an estimated instantaneous saturated permeability coefficient. This was then substituted into the grassed ditch hydrological model for simulation calculations. After multiple iterations, the converged instantaneous saturated permeability coefficient was obtained, which is the equivalent value of the grassed ditch's overall infiltration performance under this rainfall event.

[0050] Among them, the ensemble Kalman filter algorithm is used for data assimilation, such as Figure 3 As shown, specifically including: Construct the state transfer equation shown in formula (17) and the observation equation shown in formula (18): Formula (17) Formula (18) Where, represents the predicted value of the kth set of parameters of the grass-growing ditch hydrological model at time i+1; represents the updated value of the kth set of parameters of the grass-planted ditch hydrological model at time i; Represents the predicted value of the kth set of state variables at time i+1; represents the updated value of the kth set number of the state variable at time i; represents the prediction operator, which here refers to the grass-growing ditch hydrological model; Indicates the driving data of the grass-growing ditch hydrological model, mainly rainfall data; Indicates that the parameters of the grass-growing ditch hydrological model are independent white noise, all of which have a mean of 0 and a specific variance. Normal distribution; Indicates that the state variables of the grass ditch hydrological model are independent white noises, all of which have a mean of 0 and a specific variance. Normal distribution; It represents the kth set value of the outflow or water depth of the grass-planting ditch hydrological simulation at time i+1; h Represents the observation operator, that is, the conversion relationship between state variables and observation variables; Represents the error term, which has a mean of 0 and a variance of Normal distribution.

[0051] Assimilation is performed using equations (19) to (21): Formula (19) Formula (20) Formula (21) Where, represents the predicted value of the kth set number at the i+1th time; represents the updated value of the kth set at the i-th moment; represents the white noise of the kth set; represents the observation value of the kth set; represents the observation error of the kth set; represents the Kalman gain, which is the weight relationship between the predicted value and the observed value, and is calculated using Equations (22) to (24): Formula (22) Formula (23) Formula (24) Where, represents the covariance matrix of the predicted state variables; Represents the covariance matrix of the error of the observed variable prediction value; represents the ensemble mean of the predicted state variables; represents the ensemble mean of the predicted values ​​of the observed variables; N represents the number of sets; T represents the matrix transpose.

[0052] Preferably, step 10 specifically includes: Meteorological parameters (number of preceding sunny days, temperature, humidity, rainfall, rainfall duration, cumulative rainfall, and seasonal factors) from historical rainfall events in the area where the grass-planted gullies to be evaluated were averaged. These average values ​​were then mapped and categorized into three categories: low, medium, and high. Indicators were categorized by establishing these three categories. Cluster analysis was performed on the data series characteristics based on the values ​​of each indicator. The category range was calculated by considering the amplitude of the indicator observed across all data series (maximum minus minimum). The overall minimum value was the lower limit of the "low" range, and the overall maximum value was the upper limit of the "high" range. The amplitude of the indicator was divided by three, and the minimum value was added to the resulting value to obtain the upper limit of the "low" range. The maximum value was subtracted from the same value to obtain the lower limit of the "high" range. The range from the upper limit of the "low" range to the lower limit of the "high" range was the "medium" range.

[0053] The mapping results are updated into the data admissibility evaluation model. The data admissibility evaluation model is based on the Bayesian network structure, such as Figure 2 As shown in the figure, the input nodes represent the levels corresponding to the data series' characteristic values, including the number of events, average preceding sunny days, average temperature, average humidity, average rainfall, average rainfall duration, average cumulative rainfall, and seasonality index. The output nodes represent the probability that the data series is representative. Each node is discretized and the conditional probability is calculated. Node 1 is classified as the number of rainfall events. Nodes 2 through 8 are similarly categorized as "low," "medium," and "high." Node 9 is classified as "yes" or "no."

[0054] The data accessibility assessment model is used in combination with the meteorological parameter values ​​under the rainfall events in the meteorological forecast to calculate the probability that the future rainfall events in the area where the grass-planting ditch is located are representative.

[0055] The first rainfall event of each month is assumed to be representative. If it is an extreme rainfall event, it is postponed to the next one. If the probability of the data sequence being representative increases after adding the meteorological parameter values ​​of a rainfall event, then the rainfall event is representative; otherwise, the rainfall event is not representative.

[0056] The construction process of the data admissibility assessment model includes: The instantaneous saturated permeability coefficient inversion method of the grass-planted ditch was used to calculate the instantaneous saturated permeability coefficient of the grass-planted ditch under no less than 100 historical rainfall events in at least 5 years in the area where the modeled grass-planted ditch was located. The instantaneous saturated permeability coefficient of the grass-planted ditch was obtained together with the meteorological parameter values ​​of all historical rainfall events (number of sunny days before rainfall, temperature, humidity, rainfall, rainfall duration, accumulated rainfall and seasonal factors) to form the first training data subset.

[0057] Resample the first training data subset to randomly generate N*M short data sequences containing N (a multiple of 1000) rainfall event groups and M (an integer not greater than the total number of historical rainfall events) categories. Calculate the average meteorological parameter corresponding to each short data sequence. Then, perform a level mapping on the average meteorological parameter values ​​calculated from the N*M data sequences, classifying them into three categories: low, medium, and high.

[0058] Based on N*M groups of short data sequences, the XGBoost algorithm is used to establish dynamic prediction models of instantaneous saturated permeability coefficient, and the relative error compared with the dynamic prediction model of instantaneous saturated permeability coefficient established based on the first training data subset is calculated. If the relative error is less than 10%, the group of short data sequences is considered representative; otherwise, the group of short data sequences is considered unrepresentative.

[0059] Based on the rank mapping results of N*M groups of short data sequences and the representative classification results, an adoptability assessment model is established using the Bernoulli Naive Bayes method.

[0060] Among them, the main calculation process of the Bernoulli Naive Bayes method is as follows: Data preprocessing: The features in the dataset (N x M groups) (average number of sunny days before the event, average temperature, average humidity, average rainfall, average rainfall duration, average cumulative rainfall, and average seasonal factor) were converted into binary features (0 or 1), where 0 indicates that the feature value is outside the high, medium, or low range, and 1 indicates that the feature value is within the high, medium, or low range. The dataset was divided into training and test sets.

[0061] Calculate the prior probability: For each category C k , calculate the prior probability P(C k ):

[0062] Where, k Represents the category index.

[0063] Calculate the conditional probability: for each feature x i and each category C k , calculate the conditional probability P( x i | C k ):

[0064]

[0065] Where α is a smoothing parameter, which is used to prevent the probability from being 0.

[0066] Calculate the prior probability:

[0067] The class with the highest posterior probability is selected as the predicted class. For example, given a rainfall time group, the probability of the prediction being representative is 0.8 and the probability of it being unrepresentative is 0.2, then the predicted class is representative.

[0068] Preferably, in step 30, the process of constructing the model using the XGBoost algorithm is as follows: Data preparation: Convert the training data into a format suitable for XGBoost input, such as Excel spreadsheets or txt files.

[0069] Data partitioning: Divide the training data into training and test sets, usually randomly, for example, 70% training set and 30% test set.

[0070] Model initialization: Select the number-based model (gbtree), set the objective function to minimize the squared error between the predicted value and the true value, and initialize the model parameters, including the learning rate, maximum tree depth, L2 regularization parameter, etc.

[0071] Model training: The model is trained using the training set. During training, the model gradually adjusts parameters based on the objective function and optimization algorithm to minimize the training error. Select an appropriate evaluation metric (such as root mean square error or logarithmic loss) by setting the eval_metric parameter. Monitor the model's performance on the validation set during training to prevent overfitting. Enable early stopping during training to terminate training if performance on the validation set does not improve after several consecutive rounds.

[0072] Parameter optimization: Optimize model parameters using methods such as grid search, random search, or Bayesian optimization to improve model performance. Reduce model complexity and improve model generalization by analyzing feature importance or using regularization methods for feature selection.

[0073] A specific example is provided below.

[0074] A grass-planted ditch that has been in operation for more than 15 years in Utrecht, the Netherlands was selected as the application object.

[0075] The data accessibility assessment model was used to screen out 35 rainfall events in the Utrecht area of ​​the Netherlands between 2020 and 2021, and seven meteorological data including the number of sunny days before the rainfall, temperature, humidity, rainfall, rainfall duration, cumulative rainfall and seasonal factors, as well as the outflow data of the grass-planted ditch were measured. The data sequence group composed of the measured data was judged as a representative data sequence by the data accessibility assessment model, and its representative probability was 92.4%.

[0076] The outflow from the grassed ditch during each of the 35 rainfall events was used as the observation variable. The instantaneous saturated permeability coefficient for each rainfall event was calculated using the inversion method. The training data consisted of the outflow from the grassed ditch, the number of sunny days before the rainfall, temperature, humidity, rainfall, rainfall duration, cumulative rainfall, seasonal factors, and the instantaneous saturated permeability coefficient for each of the 35 rainfall events.

[0077] Based on the training data, a dynamic prediction model of instant saturated permeability coefficient was established using the XGBoost algorithm.

[0078] The instant saturated permeability coefficient dynamic prediction model was used to predict the instant saturated permeability coefficient of the grass-planted ditch under 12 rainfall events measured in 2023, and the predicted values ​​of the instant saturated permeability coefficient of the grass-planted ditch under the 12 rainfall events were obtained.

[0079] The predicted values ​​of the instantaneous saturated permeability coefficient of the grass-growing ditch under 12 rainfall events were substituted into the grass-growing ditch hydrological model to simulate the outflow flow of the grass-growing ditch under 12 rainfall events.

[0080] The results show that the Nash efficiency coefficient (NSE) of the simulated and measured outflow values ​​for the grass-growing ditch ranges from 0.71 to 0.95, with an average of 0.83. The relative error (RE) ranges from 2.43% to 13.28%, with an average of 7.13%. This indicates that the method used in this embodiment of the present invention achieves good overall results for long-term simulation of outflow from the grass-growing ditch and is feasible for practical application.

[0081] The embodiment method of the present invention provides a long-term hydrological effect monitoring and evaluation method for grass-planted ditches by monitoring the long-term data of grass-planted ditches, combining data assimilation and machine learning algorithms. The method can be applied to grass-planted ditches in any area and any scenario, and improves the effectiveness of hydrological monitoring of grass-planted ditches and the accuracy of long-term simulation of hydrological models.

[0082] The basic principles, main features, and advantages of the present invention are shown and described above. Those skilled in the art will appreciate that the present invention is not limited to the specific embodiments described above. The specific embodiments and descriptions in the specification are intended only to further illustrate the principles of the present invention. The basic principles, main features, and advantages of the present invention are shown and described above without departing from the spirit and scope of the present invention. Those skilled in the art will appreciate that various changes and modifications may be made, and such changes and modifications are intended to fall within the scope of the invention as claimed.

Claims

1. A method for monitoring and evaluating the long-term hydrological effects of grass-planting gullies, characterized in that: The following steps are involved: Step 10: Using the data accessibility assessment model, evaluate and screen future rainfall events in the area where the grass-planting ditch to be assessed is located to obtain representative rainfall events; Step 20: When a representative rainfall event occurs, meteorological parameters and hydrological performance parameters of the representative rainfall event are collected; the instantaneous saturated permeability coefficient value of the grass-planted ditch under the representative rainfall event is calculated using the inversion method of the instantaneous saturated permeability coefficient of the grass-planted ditch, and the instantaneous saturated permeability coefficient value of the grass-planted ditch under the representative rainfall event is combined with the meteorological parameter values ​​of the representative rainfall event to form a second training data set; Step 30: Based on the second training data set, an XGBoost algorithm is used to establish a dynamic prediction model for instantaneous saturated permeability coefficient; Step 40, using the dynamic prediction model of the instantaneous saturated permeability coefficient, predicting the instantaneous saturated permeability coefficient of the grass-planted ditch under a new rainfall event other than the representative rainfall event, to obtain a predicted value of the instantaneous saturated permeability coefficient of the grass-planted ditch; Step 50: input the predicted value of the instantaneous saturated permeability coefficient of the grass-planted ditch into the grass-planted ditch hydrological model to obtain the simulation result of the hydrological performance parameters of the grass-planted ditch to be evaluated.

2. The long-term hydrological effect monitoring and evaluation method of grass-planted ditch according to claim 1 is characterized in that: The meteorological parameters include the number of sunny days before rainfall, temperature, humidity, rainfall, rainfall duration, cumulative rainfall and seasonal factors; the hydrological performance parameters include the depth of water accumulation in the grass-planting ditch and the outflow flow of the grass-planting ditch.

3. The long-term hydrological effect monitoring and evaluation method of grass-planted ditch according to claim 2 is characterized in that: The instantaneous saturated permeability coefficient dynamic prediction model is a dynamic prediction model of the instantaneous saturated permeability coefficient of the grass-planted ditch with respect to the previous number of sunny days, temperature, humidity, rainfall, rainfall duration, accumulated rainfall and seasonal factors.

4. The long-term hydrological effect monitoring and evaluation method of grass-planted ditch according to claim 1 is characterized in that: The process of constructing the data admissibility assessment model includes: The instantaneous saturated permeability coefficient of the grass-planted ditch is calculated using the inversion method of the instantaneous saturated permeability coefficient of the grass-planted ditch under no less than 100 historical rainfall events in the past five years in the area where the grass-planted ditch is located. The instantaneous saturated permeability coefficient of the grass-planted ditch is calculated and combined with the meteorological parameter values ​​of the historical rainfall events to form the first training data subset; Resampling the first training data subset to randomly generate N*M groups of short data sequences containing N rainfall event groups with a total of M categories; calculating the average value of the meteorological parameters corresponding to each group of short data sequences, and performing level mapping on the average values ​​of the meteorological parameters calculated from the N*M groups of data sequences to classify them into three categories: low, medium, and high; wherein N is a multiple of 1000, and M is an integer not greater than the total number of historical rainfall events; Based on N*M groups of short data sequences, the XGBoost algorithm is used to establish dynamic prediction models for the instantaneous saturated permeability coefficient, and the relative error compared with the dynamic prediction model for the instantaneous saturated permeability coefficient established based on the first training data subset is calculated; if the relative error is less than 10%, the group of short data sequences is considered representative; otherwise, the group of short data sequences is considered unrepresentative; Based on the rank mapping results of N*M groups of short data sequences and the representative classification results, a data feasibility assessment model is established using the Bernoulli Naive Bayes method.

5. The long-term hydrological effect monitoring and evaluation method of grass-planted ditch according to claim 4 is characterized in that: The step 10 specifically includes: Calculate the average values ​​of meteorological parameters of historical rainfall events in the area where the grass-planting ditch to be evaluated is located, perform level mapping on the calculated average values ​​of meteorological parameters, and classify them into three categories: low, medium, and high. Then update the mapping results into the data accessibility assessment model. The data accessibility assessment model is used to calculate the probability that the future rainfall event in the area where the grass-planting ditch is located is representative of the meteorological parameter values ​​under the meteorological forecast rainfall event. The first rainfall event of each month is assumed to be representative. If it is an extreme rainfall event, it is postponed to the next one. If the probability of the data sequence being representative increases after adding the meteorological parameter values ​​of a rainfall event, then the rainfall event is representative; otherwise, the rainfall event is not representative.

6. The long-term hydrological effect monitoring and evaluation method of grass-planted ditch according to claim 1 is characterized in that: The construction method of the grass-planted ditch hydrological model is as follows: based on equations (1) and (2), considering the time-varying characteristics of the saturated permeability coefficient, proposing the consistency hypothesis of the vertical variation of the ditch soil moisture content and introducing the static pressure head parameter, the grass-planted ditch hydrological model is established; Formula (1) Formula (2) Where, f represents the infiltration rate; K s ( x , t ) represents the saturated permeability coefficient, x Indicates the influencing factors, t Indicates time; Z f represents the penetration path length; ΔP Indicates the pressure difference across the permeation unit; K ( θ ) represents the unsaturated permeability coefficient; K s represents the saturated permeability coefficient; θ Indicates the volumetric moisture content of soil; θ r Indicates the residual volume moisture content of soil; θ s represents the saturated volumetric moisture content of soil; α represents the empirical fitting parameter or curve shape parameter; n represents the empirical fitting parameter or curve shape parameter.

7. The long-term hydrological effect monitoring and evaluation method of grass-planted ditch according to claim 6 is characterized in that: In the grass-planted ditch hydrological model, the calculation process of the wet stage includes: As the rainfall progresses, surface runoff continues to flow in. The moisture content of the packing layer continues to rise under the combined action of top water inflow and bottom drainage until it reaches the saturated moisture content. The outflow from the bottom of the packing layer serves as the inflow of the drainage layer. Specifically: Formula (3) Where, It represents the surface water volume of the grass-planted ditch at the end of unit time when there is no infiltration at the current moment; It represents the surface water volume of the grass-planting ditch before the current moment; Indicates rainfall intensity; Represents a unit time interval; Formula (4) Where, Indicates the water depth in the grass-planting ditch at the end of unit time when there is no infiltration at the current moment; It represents the upper surface area of ​​the filler layer when the aquifer is rectangular; Formula (5) Where, Indicates the infiltration rate of the packing layer at the end of unit time when there is no infiltration; It represents the saturated permeability coefficient of the packing layer; Indicates the thickness of the filler layer; Formula (6) Where, It represents the volume of water accumulated in the grass-planted ditch after infiltration per unit time at the infiltration rate without infiltration; It represents the overflow volume of the grass-planted ditch per unit time; make , set the minimum error range and perform iterative calculation to obtain the volume and depth of water in the grass-planting ditch; Formula (7) Formula (8) Where, Indicates the infiltration amount of the filler layer when the bottom is not drained at the current moment; Indicates the infiltration amount of the filler layer relative to the residual moisture content before the current moment; Indicates the moisture content of the packing layer when the bottom is not drained at the current moment; Formula (9) Where, It indicates the unsaturated permeability coefficient of the packing layer when the bottom is not drained at the current moment; Indicates the residual volume water content of the packing layer; Indicates the saturated volume water content of the packing layer; Indicates the saturated permeability coefficient of the packing layer; Formula (10) Where, Indicates the water outflow rate at the bottom of the packing layer at the current moment; Will As the water inflow rate of the drainage layer, the same iterative method as that of the filler layer is adopted to obtain the water volume and depth at each moment on the surface of the drainage layer; Formula (11) Formula (12) Where, It represents the water outflow rate at the bottom of the packing layer calculated based on the current moment when the bottom is not drained, and the infiltration rate of the packing layer after drainage; It indicates the water discharge rate at the bottom of the packing layer calculated based on the current moment when the bottom is not drained, and the moisture content of the packing layer after drainage; It represents the water outflow rate at the bottom of the packing layer calculated based on the current moment when the bottom is not drained, and the infiltration rate of the drainage layer after drainage; represents the upper surface area of ​​the drainage layer; make , set the minimum error range and perform iterative calculation to obtain the infiltration volume and moisture content of the packing layer at the current moment; The same iterative method as that for the filler layer is adopted to calculate the infiltration rate and moisture content of the drainage layer.

8. The long-term hydrological effect monitoring and evaluation method of grass-planted ditch according to claim 6 is characterized in that: In the grass-planted ditch hydrological model, the calculation process of the drainage stage includes: When the rainfall process has not yet ended and the drainage layer is full, the grass-planted ditch begins to drain water outward through the perforated drainage pipe. Assuming that the permeability of the drainage layer is large enough to discharge the water from the filler layer in time, the outflow rate of the drainage pipe per unit time is equal to the inflow rate at the bottom of the filler layer minus the infiltration rate from the bottom of the grass-planted ditch into the natural soil. The average infiltration rate from the bottom to the natural soil is considered to be approximately equal to the saturated permeability coefficient of the natural soil. The outflow rate is calculated using formula (13): Formula (13) Where, It represents the average outflow rate at the outlet of the drainage pipe of the grass-planting ditch; Indicates the saturated permeability coefficient of the natural soil at the bottom of the grass-planting ditch; represents the surface area of ​​the drainage layer; Use formula (14) to calculate the pipe flow velocity along the drainage blind pipe and the local head loss: Formula (14) Formula (15) Where, The pipe flow velocity represents the loss of head along the drainage blind leg and the local head loss; H0 represents the total head including the travel head; Indicates the correction coefficient, generally taken as 1.0; Indicates the head loss, including the head loss along the way and the local head loss; represents the resistance coefficient along the way; Indicates the length of the pipeline; Indicates the pipe diameter; Indicates the cross-sectional length of the water flow in the pipe; Indicates the length of the pipeline; represents the local resistance coefficient; Indicates the flow velocity in the pipe; After the rainfall process ends, the surface runoff also stops, and the grass-planted ditch enters the emptying stage. At the beginning, there is still water on the surface of the grass-planted ditch, and the infiltration meets the saturated infiltration. The infiltration rate of the packing layer is calculated using formula (7); after the surface water disappears, the packing layer begins to gradually empty in an unsaturated state. The emptying rate of the packing layer is calculated using formula (16): Formula (16) Where, Indicates the emptying rate of the packing layer, represents the unsaturated permeability coefficient of the packing layer, Indicates the moisture content of the packing layer at the current moment; At this time, the drainage layer is still in a saturated state. When the moisture content of the drainage layer at the current moment is less than the saturated moisture content of the drainage layer, the outlet stops discharging water, and the water in the drainage layer slowly seeps into the underground soil. When the moisture content of the filler layer and the drainage layer drops to their respective field moisture contents, the rainfall infiltration process ends.

9. The long-term hydrological effect monitoring and evaluation method of grass-planted ditch according to claim 1 is characterized in that: In step 20, the instantaneous saturated permeability coefficient value of the grass-planted ditch of a representative rainfall event is calculated using the instantaneous saturated permeability coefficient inversion method, which specifically includes: Taking the water accumulation depth in the grass ditch or the outflow flow of the grass ditch under a representative rainfall event as the observation variable, the initial assumed instantaneous saturated permeability coefficient value of the grass ditch is substituted into the grass ditch hydrological model to obtain the simulation result of the observation variable; the ensemble Kalman filter algorithm is used to perform data assimilation on the observation results and the simulation results to generate an estimated value of the instantaneous saturated permeability coefficient, which is then substituted into the grass ditch hydrological model for simulation calculation. After multiple iterations, the converged instantaneous saturated permeability coefficient value of the grass ditch is obtained.

10. The long-term hydrological effect monitoring and evaluation method of grass-planted ditch according to claim 9 is characterized in that: In step 20, the data assimilation is performed using the ensemble Kalman filter algorithm, which specifically includes: Based on the state transfer equation shown in formula (17) and the observation equation shown in formula (18): Formula (17) Formula (18) Where, represents the predicted value of the kth set of parameters of the grass-growing ditch hydrological model at time i+1; represents the updated value of the kth set of parameters of the grass-planted ditch hydrological model at time i; Represents the predicted value of the kth set of state variables at time i+1; represents the updated value of the kth set number of the state variable at time i; represents the prediction operator; Indicates the driving data of the grass-growing ditch hydrological model; represents the independent white noise of the parameters of the grass-growing ditch hydrological model; represents the independent white noise of the state variables of the grass-growing ditch hydrological model; It represents the kth set value of the outflow or water depth of the grass-planting ditch hydrological simulation at time i+1; h represents the observation operator; Represents the error term, which has a mean of 0 and a variance of Normal distribution; Assimilation is performed using equations (19) to (21): Formula (19) Formula (20) Formula (21) Where, represents the predicted value of the kth set number at the i+1th time; represents the updated value of the kth set at the i-th moment; represents the white noise of the kth set; represents the observation value of the kth set; represents the observation error of the kth set; represents the Kalman gain, which is the weight relationship between the predicted value and the observed value, and is calculated using Equations (22) to (24): Formula (22) Formula (23) Formula (24) Where, represents the covariance matrix of the predicted state variables; Represents the covariance matrix of the error of the observed variable prediction value; represents the ensemble mean of the predicted state variables; represents the ensemble mean of the predicted values ​​of the observed variables; N represents the number of ensembles.

Images