Runoff multi-component segmentation method coupling nonlinear reservoir model and conductivity mass balance

Abstract

The invention relates to a runoff multi-component segmentation method coupling a nonlinear reservoir model and conductivity mass balance, comprising: obtaining observation data of a target basin, the observation data comprising runoff data and conductivity data; extracting a base flow recession section based on the observation data, analyzing the base flow recession section to obtain nonlinear reservoir parameters and conductivity related parameters, and respectively constructing a nonlinear reservoir model and a conductivity mass balance equation; dividing independent event segments on the basis of the base flow decline segment, performing runoff segmentation in each independent event segment through a non-linear reservoir model and a conductivity mass balance equation by taking conductivity flux estimation error minimization at the outlet of the target drainage basin as a target, obtaining rapid runoff and base flow, and respectively calculating an old water proportion and a new water proportion; and summarizing the fast runoff, the base flow, the old water and the new water segmented from each independent event segment to obtain a multi-component segmentation result of the target drainage basin. According to the invention, high-precision segmentation of runoff components can be realized.

Description

Technical Field

[0001] The present invention relates to the technical field of hydrology, water resources and environmental engineering, and particularly relates to a method for multi-component segmentation of runoff by coupling a non-linear reservoir model and a conductivity mass balance. Background Art

[0002] The segmentation of river runoff components is a hot and difficult point in the research of engineering hydrology. Accurately dividing runoff components is a prerequisite for deeply understanding the generation mechanism of basin runoff, increasing the accuracy of flood forecasting and evaluating the potential impact of climate change, and is also an important basis for better serving basin water resources management, non-point source pollution source identification, and aquatic ecosystem protection. Existing runoff segmentation techniques are mainly divided into two categories: the flow path segmentation method based on the non-linear reservoir model and the residence time segmentation method based on tracer mass conservation, but both have significant limitations, as follows:

[0003] (1) Segmentation method based on the non-linear reservoir model:

[0004] The method based on the non-linear reservoir model divides the total runoff into two categories, rapid runoff and base flow, by simulating the water storage-drainage process of the basin. Such methods only rely on runoff data, are easy to operate, and are widely used in hydrological research. However, its core defects are: ① single segmentation dimension: it can only distinguish the flow path of runoff (fast / slow), and cannot simultaneously analyze the residence time of water bodies (new / old water components); ② strong subjectivity: parameter calibration depends on empirical assumptions, and the segmentation results are easily affected by human interference.

[0005] (2) Segmentation method based on chemical mass conservation:

[0006] The method using isotopes, chemical ions or conductivity as tracers divides the new water and old water in runoff through a mass balance equation. The limitations of such methods are: ① high data requirements: multiple-dimensional tracers are required (N-1 independent tracers are required for N components), and in practical applications, only two components can often be segmented due to insufficient data; ② limited cost and universality: for example, stable hydrogen and oxygen isotopes require high-frequency sampling and high testing costs, and although conductivity can be monitored in-situ, its non-conservatism may introduce errors.

[0007] (3) Attempts and deficiencies in multi-component segmentation:

[0008] Existing technologies have attempted to achieve multi-component segmentation by coupling hydrological models and tracers, but there are still the following problems: ① high model complexity: it is necessary to generalize modules such as precipitation, evapotranspiration, and groundwater, introducing a large number of assumptions and parameters, resulting in difficult calibration; ② it is difficult to ensure water balance: the sum of simulated base flow and rapid runoff often deviates from the measured total runoff; ③ high application threshold: it is necessary to simultaneously obtain runoff, precipitation, and tracer data, which limits the universality of the method.

[0009] Generally speaking, the existing technologies have significant deficiencies in the following aspects: ①Segmentation group splitting: Traditional methods can only analyze the flow path or residence time separately and cannot synchronously split multiple components; ②Contradiction between data and cost: High-precision segmentation requires the support of multiple tracers, but in practical applications, it is difficult to balance the data acquisition cost and operation complexity; ③Conflict between physical basis and practicality: Although the coupled model has a clear physical basis, the complex characterization of the hydrological process limits the engineering application value; ④Absence of water balance: Existing coupled model methods are difficult to strictly ensure the consistency between the segmentation result and the measured runoff, affecting the reliability. Summary of the Invention

[0010] The object of the present invention is to provide a method for multi-component segmentation of runoff by coupling a non-linear reservoir model and conductivity mass balance, which breaks through the limitations of traditional methods in multi-component segmentation, data dependence and model complexity, and provides a high-precision, low-cost and easy-to-operate runoff multi-component analysis tool for hydrological research and engineering practice.

[0011] To achieve the above object, the present invention provides the following solutions:

[0012] A method for multi-component segmentation of runoff by coupling a non-linear reservoir model and conductivity mass balance, comprising:

[0013] Obtain the observation data of the target basin, wherein the observation data includes runoff data and conductivity data;

[0014] Based on the observation data, extract the base flow recession segment, analyze the base flow recession segment, obtain the non-linear reservoir parameters and conductivity-related parameters, and respectively construct a non-linear reservoir model and a conductivity mass balance equation;

[0015] Based on the base flow recession segment, divide the independent event segments, and through the non-linear reservoir model and the conductivity mass balance equation, with the goal of minimizing the estimation error of the conductivity flux at the outlet of the target basin, perform runoff segmentation within each independent event segment, obtain the quick runoff and the base flow, and respectively calculate the old water proportion and the new water proportion of the quick runoff and the base flow;

[0016] Summarize the quick runoff, base flow segmented from each independent event segment, and the calculated old water proportion and new water proportion of the quick runoff and the base flow to obtain the multi-component segmentation result of the target basin.

[0017] Optionally, obtaining the non-linear reservoir parameters includes:

[0018] Extract the recession segment without quick runoff interference in the base flow recession segment, and determine the non-linear reservoir parameters k and β by the bin regression method;

[0019] Obtaining the conductivity-related parameters includes:

[0020] Estimate the water-rock interaction correction term D through the conductivity change rate of the base flow recession segment;

[0021] Preset the vadose zone passive storage S through the parameter rate pa ;

[0022] Determine the effective precipitation conductivity EC based on the mean conductivity during the preset time period P .

[0023] Optionally, the non-linear reservoir model is:

[0024]

[0025] where Q b is the base flow, t is time, and the parameters The conversion relationships between and ω and the parameters k and β are: β = 2 - ω.

[0026] Optionally, the conductivity mass balance equation is:

[0027] Fast runoff dynamic conductivity equation:

[0028]

[0029] Base flow dynamic conductivity equation:

[0030]

[0031] where t is time, Δt is the time step, is the fast runoff conductivity at time t + Δt; R [t+Δt] is the infiltration recharge at time t + Δt; Q [t+Δt] is the total runoff at time t + Δt; is the base flow at time t + Δt; is the effective precipitation conductivity at time t + Δt; Q [t] is the total runoff at time t; is the base flow at time t; is the effective precipitation conductivity at time t; is the fast runoff conductivity at time t; S pa is the vadose zone passive storage; is the base flow conductivity at time t + Δt; k and β are non-linear reservoir parameters; D is the water-rock interaction correction term; is the base flow conductivity at time t.

[0032] Optionally, with the goal of minimizing the estimation error of the conductivity flux at the outlet of the target basin:

[0033]

[0034] where t is time, Δt is the time step, and AE [t+Δt] is the estimated error of the conductivity flux at the outlet of the target basin at time t + Δt; Q [t+Δt] is the total runoff at time t + Δt; EC [t+Δt] is the conductivity of the total runoff at time t + Δt; is the base flow at time t + Δt; is the conductivity of the quick runoff at time t + Δt; is the conductivity of the base flow at time t + Δt.

[0035] Optionally, after iteratively selecting the best base flow sequence with the goal of minimizing the estimated error of the conductivity flux at the outlet of the target basin, with the goal of minimizing the sum of the squared residuals of the conductivity flux, iteratively select the optimal value of the vadose zone passive storage volume S pa as:

[0036]

[0037] where SSE is the sum of the squared residuals, AE [t] is the estimated error of the conductivity flux at the outlet of the target runoff at time t, and nΔt is the length of the independent event segment sequence.

[0038] Optionally, calculating the proportion of old water and new water in the quick runoff and base flow includes:

[0039] The proportion of old water in the quick runoff Q fo :

[0040] The proportion of new water in the quick runoff Q fn : Q fn = Q f - Q fo ;

[0041] The proportion of old water in the base flow Q bo :

[0042] The proportion of new water in the base flow Q bn : Q bn = Q b - Q bo ;

[0043] where Q f is the quick runoff, EC f is the conductivity of the quick runoff, EC P is the conductivity of the effective precipitation, EC fo is the conductivity of the old water in the quick runoff, Q b is the base flow, ECb is the base flow conductivity, EC bo is the old water conductivity in the base flow.

[0044] Optionally, the multi-component segmentation result of the target basin satisfies the water balance ∑(Q bo +Q bn +Q fo +Q fn ) = Q 总径流 .

[0045] The beneficial effects of the present invention are as follows:

[0046] By coupling the non-linear reservoir model with the conductivity mass balance, the present invention innovatively proposes a reverse iteration algorithm, breaking through the limitations of traditional methods in terms of segmentation dimension, data dependence, and water balance. Its characteristics of low cost, high precision, and universality provide an efficient tool for watershed hydrological research, flood forecasting, and non-point source pollution management, and have significant academic value and engineering application prospects. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0048] Figure 1 is a flow chart of a multi-component runoff segmentation method coupling a non-linear reservoir model and a conductivity mass balance according to an embodiment of the present invention;

[0049] Figure 2 is the runoff volume and conductivity monitoring data of a single-event segment runoff segmentation example according to an embodiment of the present invention;

[0050] Figure 3 is a schematic diagram of the iterative segmentation process of a single-event segment according to an embodiment of the present invention;

[0051] Figure 4 is a segmentation result diagram of a single-event segment runoff segmentation example according to an embodiment of the present invention;

[0052] Figure 5 is a segmentation result diagram of all event segments according to an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0053] The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0054] To make the above objects, features, and advantages of the present invention more obvious and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0055] This embodiment provides a runoff multi-component segmentation method that couples a non-linear reservoir model and a conductivity mass balance, as Figure 1 shown, including:

[0056] Obtain the observed data of the target basin, where the observed data includes runoff data and conductivity data;

[0057] Based on the observed data, extract the base flow recession segment, analyze the base flow recession segment, obtain the non-linear reservoir parameters and conductivity-related parameters, and respectively construct a non-linear reservoir model and a conductivity mass balance equation;

[0058] Based on the base flow recession segment, divide the independent event segments. Through the non-linear reservoir model and the conductivity mass balance equation, with the goal of minimizing the estimation error of the conductivity flux at the outlet of the target basin, perform runoff segmentation within each independent event segment to obtain the quick runoff and base flow, and respectively calculate the old water proportion and new water proportion of the quick runoff and base flow;

[0059] Summarize the quick runoff, base flow segmented from each independent event segment, and the calculated old water proportion and new water proportion of the quick runoff and base flow to obtain the multi-component segmentation result of the target basin.

[0060] Specifically, by coupling the non-linear reservoir model and the conductivity mass balance in this embodiment, an inverse iteration algorithm is innovatively proposed, which breaks through the limitations of traditional methods in segmentation dimension, data dependence, and water balance. Its characteristics of low cost, high precision, and universality provide an efficient tool for watershed hydrological research, flood forecasting, and non-point source pollution management, and have significant academic value and engineering application prospects.

[0061] Furthermore, obtaining the non-linear reservoir parameters includes:

[0062] Extract the recession segment without quick runoff interference in the base flow recession segment, and determine the non-linear reservoir parameters k and β by the bin regression method;

[0063] Obtaining the conductivity-related parameters includes:

[0064] Estimate the water-rock interaction correction term D through the conductivity change rate of the base flow recession section;

[0065] Preset the vadose zone passive storage S through the parameter rate pa ;

[0066] Determine the effective precipitation conductivity EC based on the mean conductivity during the preset time period P .

[0067] Specifically, the determination of key parameters in this embodiment includes the following:

[0068] S1.1: Determination of non-linear reservoir parameters;

[0069] (1) Base flow recession analysis;

[0070] Extract the recession section without rapid runoff interference, and determine the non-linear reservoir parameters k and β through vs.log(Q b ) bin regression.

[0071] (2) Support for time-varying parameters;

[0072] For long sequence data, use the sliding window method to estimate k and β in segments to improve the adaptability of the model to seasonal or extreme events.

[0073] S1.2: Dynamic correction of conductivity;

[0074] (1) Water-rock interaction correction term (D);

[0075] Estimate D through the conductivity change rate of the base flow recession section to quantify the influence of evapotranspiration or solute exchange on the base flow conductivity.

[0076] (2) Vadose zone passive storage (S pa );

[0077] Optimize S through parameter calibration pa , reflecting the old water storage in the vadose zone that does not participate in runoff generation, and optimizing the estimation of rapid runoff conductivity.

[0078] (3) Effective precipitation conductivity (EC P );

[0079] Determine the conductivity value of effective precipitation based on the mean conductivity during the super-large flood peak period to avoid directly relying on precipitation monitoring data.

[0080] S1.3: Estimation of old water conductivity;

[0081] Interpolate the runoff conductivity at the end of the recession section to dynamically determine the end member value of old water conductivity and reduce the error caused by non-conservatism.

[0082] Furthermore, the non - linear reservoir model is constructed as follows:

[0083]

[0084] wherein, Q b is the base flow, t is time, and the conversion relationships of the parameters and ω with the parameters k and β are: β = 2 - ω.

[0085] The conductivity mass - balance equation is constructed as:

[0086] Fast - runoff dynamic conductivity equation:

[0087]

[0088] Base - flow dynamic conductivity equation:

[0089]

[0090] wherein, t is time, Δt is the time step, is the fast - runoff conductivity at time t+Δt; R [t+Δt] is the infiltration recharge at time t+Δt; Q [t+Δt] is the total runoff at time t+Δt; is the base flow at time t+Δt; is the effective precipitation conductivity at time t+Δt; Q [t] is the total runoff at time t; is the base flow at time t; is the effective precipitation conductivity at time t; is the fast - runoff conductivity at time t; S pa is the vadose - zone passive storage; is the base - flow conductivity at time t+Δt; k and β are non - linear reservoir parameters; D is the water - rock interaction correction term; is the base - flow conductivity at time t.

[0091] Furthermore, with the goal of minimizing the estimation error of the conductivity flux at the outlet of the target basin:

[0092]

[0093] wherein, t is time, Δt is the time step, AE [t+Δt] is the estimation error of the conductivity flux at the outlet of the target basin at time t+Δt; Q [t+Δt] is the total runoff at time t+Δt; EC [t+Δt] is the total - runoff conductivity at time t+Δt; is the base flow at time t+Δt; is the rapid runoff conductivity at time t+Δt; is the base flow conductivity at time t+Δt.

[0094] After iteratively selecting the optimal base flow sequence with the goal of minimizing the estimation error of the conductivity flux at the outlet of the target basin, and then with the goal of minimizing the sum of the squared residuals of the conductivity flux, iteratively select the optimal value of the vadose zone passive storage S pa as:

[0095]

[0096] where SSE is the sum of the squared residuals, AE [t] is the estimation error of the conductivity flux at the target runoff outlet at time t, and nΔt is the length of the independent event segment sequence.

[0097] Furthermore, calculating the proportion of old water and new water in the rapid runoff and base flow includes:

[0098] The proportion of old water in the rapid runoff Q fo :

[0099] The proportion of new water in the rapid runoff Q fn : Q fn =Q f -Q fo ;

[0100] The proportion of old water in the base flow Q bo :

[0101] The proportion of new water in the base flow Q bn : Q bn =Q b -Q bo ;

[0102] where Q f is the rapid runoff, EC f is the rapid runoff conductivity, EC P is the effective precipitation conductivity, EC fo is the old water conductivity in the rapid runoff, Q b is the base flow, EC b is the base flow conductivity, EC bo is the old water conductivity in the base flow.

[0103] The multi-component segmentation result of the target basin satisfies the water balance ∑(Q bo +Q bn +Q fo +Q fn ) = Q 总 runoff.

[0104] Specifically, this embodiment is based on the reverse deduction logic of "export → basin redistribution → effective precipitation", abandons the traditional "precipitation → runoff" forward modeling strategy, and realizes runoff segmentation by combining the nonlinear reservoir model and the conductivity mass balance equation. The specific steps are as follows:

[0105] S1.1: Core algorithm design;

[0106] (1) Simultaneous water balance equations;

[0107] The total runoff is decomposed into rapid runoff (Q f ) and base flow (Q b ), satisfying Q=Q f +Q b .

[0108] Vadose zone water balance equation The balance equation with base flow and nonlinear storage-exhaust equation Combined, the calculation formula for infiltration recharge (R) is derived

[0109] (2) Simultaneous conductivity mass balance equations;

[0110] The total runoff conductivity (EC) is decomposed into rapid runoff conductivity (EC f ) and base current conductivity (EC b ), satisfying QEC=Q f EC f +Q b EC b .

[0111] Introducing passive storage in the vadose zone (S pa ) and the rate of change of base flow conductivity caused by water-rock interaction (D), and the dynamic conductivity equation of rapid runoff and base flow is constructed

[0112] (3) Reverse iterative solution;

[0113] The goal is to minimize the conductivity flux estimation error at the basin outlet

[0114] Accurate division of runoff paths (fast runoff and base flow) is now possible.

[0115] The trapezoidal rule is used for integral approximation during the iteration process to ensure numerical stability.

[0116] S1.2: Water balance is strictly guaranteed;

[0117] Forcing satisfaction of ∑(Q bo +Q bn +Q fo +Q fn )=Q 总径流 through the reverse algorithm, the water volume deviation problem of the traditional coupling model is avoided.

[0118] During the iteration process, the upper and lower limits of the base flow value are constrained based on the non-linear reservoir model to ensure physical rationality. Among them is the base flow rate (theoretical lower limit) that decays forward without recharge within [t, t+Δt], and is the base flow rate (theoretical upper limit) that decays "backward"

[0119] Furthermore, the proportions of quick runoff, base flow, old water, and new water segmented from each independent event segment are summarized to obtain the multi-component segmentation result of the target runoff, specifically including:

[0120] S3.1: Event segment splitting and processing;

[0121] Taking the end moment of the base flow recession segment as the break point, the runoff sequence is segmented into multiple independent event segments for segmentation respectively to ensure the independence of the hydrological process within each event segment.

[0122] S3.2: Runoff segmentation process for a single event segment;

[0123] 1) Determine the initial values of relevant variables within each event segment, including: Assume that the front and back break points of the event segment are full base flow points, that is Assume that the initial conductivity of passive storage is equal to the base flow conductivity before the rising water, that is

[0124] 2) Under a specific passive storage (S pa ), starting from t = 0, with the goal of minimizing the absolute error (AE) between the measured value and the calculated value of the conductivity flux (Q [t+Δt] EC [t+Δt] ), the optimal value of is gradually iteratively optimized, and the iteration increment is taken as

[0125]

[0126] 3) At a specific moment (t = t+Δt), in order to ensure that the base flow recession process satisfies the non-linear reservoir relationship, the possible value range of is ​is the base flow (theoretical lower limit) resulting from positive recession during [t, t+Δt] without recharge, and is the base flow (theoretical upper limit) resulting from reverse "recession". To avoid possible contradictions caused by or , the upper and lower limits of the base flow value are taken as and respectively.

[0127] 4) After the optimization of the base flow at all times within the event segment is preferably completed, with the goal of minimizing the sum of squared residuals (SSE) of the conductivity flux, the optimal value of S pa is preferably determined, and the iterative increment is taken as

[0128]

[0129] 5) After the iteration is completed, based on the preferably determined S pa and sequences, the rapid runoff , infiltration recharge (R [t] ), effective precipitation (P *[t] ), base flow conductivity , rapid runoff conductivity , and the sequence values of the absolute error of solute flux (AE [t] ) are calculated, as well as the corresponding sum of squared residuals (SSE).

[0130] 6) After the flow paths (rapid runoff and base flow) are segmented, the rapid runoff volume, base flow volume, and their corresponding conductivity sequences can be obtained. Furthermore, the conductivity of the old water is estimated by linearly interpolating the conductivity at the end of the recession segment that is not affected by the rapid runoff, and it is assumed that the rapid runoff and the base flow have the same old water conductivity value. Finally, based on the conductivity mass balance equation, the proportion of old water in the rapid runoff (Q fo ) and the proportion of old water in the base flow (Q bo ) are segmented:

[0131]

[0132] Furthermore, based on the water conservation, the new water in the rapid runoff (Q fn ) and the part of the new water in the base flow (Q bn ) are calculated: Q fn =Q f -Q fo , Q bn =Q b -Q bo .

[0133] S3.3: Summarize the segmentation results;

[0134] Repeat the process of steps 1)-6) in S3.2 to achieve the runoff segmentation of all event segments. Summarize the segmentation results of each event segment to obtain the complete sequence values of the four-component (base flow, rapid runoff, old water, new water) segmentation results.

[0135] The following takes a typical small and medium-sized watershed in the continental United States (with an area of about 424 km 2 , and the climate type is temperate humid) as an example to apply this method to achieve the four-component runoff segmentation. This watershed has daily runoff (Q) and conductivity (EC) monitoring data, with a time span from October 9, 2019 to April 30, 2022, and the data completeness rate > 95%. The segmentation process is as shown in the appendix Figure 1 as follows, and the specific steps are as follows:

[0136] S1: Data preparation and preprocessing;

[0137] S1.1: Data acquisition and cleaning;

[0138] 1) Runoff data: Obtain daily flow data (unit: m 3 / s) from the United States Geological Survey (USGS) and convert it to runoff depth (mm / d).

[0139] 2) Conductivity data: Synchronously obtain daily EC data (μS / cm), and the missing values are interpolated through the power-law relationship between conductivity and runoff.

[0140] S1.2: Event segment division and recession segment extraction;

[0141] 1) Recession segment screening: Exclude all rising segments (dQ / dt ≥ 0) and sequences with a recession duration of less than 6 days; remove the data of the first 3 days and the last 2 days of each recession segment to exclude the interference of rapid runoff; exclude the abnormally fluctuating segments (ΔQ / Q > 30%), and finally retain 33 effective recession segments.

[0142] 2) Event segment division: Use the end time of each recession segment as the breakpoint to split the runoff sequence into different event segments for subsequent runoff segmentation.

[0143] S2: Determine key parameters;

[0144] S2.1: Nonlinear reservoir parameters (k, β);

[0145] Another representation of the nonlinear reservoir model is and Q b the power-law relationship between:

[0146]

[0147] where the parameter The conversion relationships between and ω and the parameters k and β are as follows: β = 2 - ω. Through vs. log(Q b ) scatter binning regression gives the values of and ω, and then k = 19.62 and β = 0.66 are calculated.

[0148] S2.2: Parameters related to conductivity;

[0149] 1) Effective precipitation conductivity (EC P ): Extract the EC values corresponding to the 3 largest flood peaks and take the average value EC P = 41 μS / cm.

[0150] 2) Water-rock interaction correction term (D): Calculate the average change rate of EC in the recession segment:

[0151] 3) Potential value range of vadose zone passive storage (S pa ): Estimate the upper limit of vadose zone passive storage based on the maximum flood peak flow rate, and determine its potential value range to be 0 - 217.7 mm.

[0152] S3: Reverse iterative segmentation process;

[0153] S3.1: Runoff segmentation for a single event segment (taking the event from April to May 2020 as an example);

[0154] 1) Event segment range: From April 22, 2020 (t = 0) to May 12, 2020 (t = 21 days), see Figure 2 .

[0155] 2) Determine the initial values of relevant variables within the event segment, including: Assume that the front and back break points of the event segment are full base flow points, that is

[0156] Assume that the initial conductivity of passive storage is equal to the base flow conductivity before the rising stage, that is

[0157] 3) Under a specific passive storage (S pa ), starting from t = 1, with a time step of Δt = 1 day, aiming to minimize the absolute error (AE) between the measured and calculated values of the conductivity flux (Q [t +Δt] EC [t+Δt] ), gradually iterate to optimize the best value, and the iteration increment is taken as

[0158]

[0159] 4) At a specific moment (t = t + Δt), to ensure that the base flow recession process satisfies the non-linear reservoir relationship, The possible value range of where is the base flow (theoretical lower limit) that decays forward from without recharge during [t, t + Δt], is the base flow (theoretical upper limit) that "recedes" backward from To avoid possible contradictions caused by or respectively take and as the upper and lower limits of the base flow value; see the schematic flow chart in Figure 3 .

[0160] 5) Repeat steps 2)-4) to complete the optimization of the 21-day base flow sequence.

[0161] 6) Traverse the S pa value range (0 - 217.7 mm), calculate the sum of squared residuals Optimize Spa = 8 mm (corresponding to the minimum SSE = 939).

[0162] 7) After the iteration ends, based on the optimized S pa and sequence, calculate the rapid runoff Base flow conductivity Rapid runoff conductivity The sequence values of Figure 4 .

[0163] 8) Estimate the conductivity of the old water by linearly interpolating the conductivity at the end of the recession segment. Based on the conductivity mass balance equation, divide the proportion of old water in the rapid runoff (Q fo ) and the proportion of old water in the base flow (Q bo ):

[0164]

[0165] 9) Furthermore, based on the water conservation, calculate the new water in the rapid runoff (Q fn ) and the part of the new water in the base flow (Q bn ): Q fn = Q f - Q fo , Q bn = Qb -Q bo The segmentation results are shown in Figure 4 。

[0166] S3.2: Runoff segmentation for the entire period;

[0167] According to the process of S3.1, the runoff segmentation of all event segments is realized. Summarize the segmentation results of each event segment to obtain the complete S pa 、 R [t] 、P *[t] 、 AE [t] 、the sequence values of SEE, and the results are shown in Figure 5 。

[0168] S4: Result verification and output;

[0169] Calculate the Kling Gupta Efficiency (KGE) between the simulated value and the measured value of runoff conductivity. KGE = 0.94, indicating that the segmentation result has excellent reproducibility for runoff conductivity. According to the above calculation results, summarize and output the results of multi-component runoff segmentation.

[0170] This embodiment fully demonstrates the entire process from data preparation, parameter calibration, inverse iterative calculation to result verification. Through actual data verification, this method realizes high-precision segmentation of the four components of runoff in small and medium-sized basins, strictly satisfies the water balance, and shows strong robustness to parameter fluctuations. This method can provide reliable technical support for flood forecasting, pollution source tracing, etc.

[0171] The method proposed in this embodiment can achieve the following technical effects:

[0172] (1) Breakthrough in segmentation accuracy and dimension. The inverse iterative algorithm is adopted to ensure that the segmentation result is completely consistent with the measured total runoff (ΣQ components = Q total runoff), improving the result accuracy; introduce the correction terms of water-rock interaction and vadose zone passive storage to reduce the influence of non-conservative conductivity on the segmentation accuracy; through the coupling of the non-linear reservoir model and the conductivity mass balance, realize the synchronous analysis of the four components of the flow path (fast / slow) and residence time (new / old) in the total runoff, breaking through the dimension limitation that traditional methods can only segment two components.

[0173] (2) Optimization of data cost and efficiency. Only rely on the runoff sequence and conductivity data at the basin outlet, avoiding the dependence on complex monitoring data such as precipitation and evapotranspiration; the inverse algorithm can reduce the number of iterations, and the segmentation efficiency is significantly improved. It only takes 2.8 minutes on average to perform daily segmentation of a 10-year-long runoff sequence at a single station.

[0174] (3) Reduced parameter sensitivity. The segmentation results are less sensitive to the parameters of passive storage and water-rock interaction (KGE fluctuates < 2% when varying ±10%), and the key parameters (k, β, EC P ) can be accurately estimated through a standardized process.

[0175] (4) Enhanced adaptability. It is applicable to basins lacking precipitation monitoring or data on complex surface processes. Especially in the context of the widespread popularization of long-term conductivity monitoring, it has the potential for large-scale promotion. It has been tested at 51 sites, covering arid, temperate, and snowmelt-dominated basins, as well as various geological types such as sedimentary rock, volcanic rock, and granite, verifying the wide applicability of the method.

[0176] In this embodiment, by coupling a non-linear reservoir model with a conductivity mass balance, an inverse iteration algorithm is innovatively proposed, breaking through the limitations of traditional methods in terms of segmentation dimension, data dependence, and water balance. Its characteristics of low cost, high precision, and universality provide an efficient tool for watershed hydrological research, flood forecasting, and non-point source pollution management, and have significant academic value and engineering application prospects.

[0177] The embodiments described above are only descriptions of the preferred embodiments of the present invention, and do not limit the scope of the present invention. Without departing from the design spirit of the present invention, various deformations and improvements made by those of ordinary skill in the art to the technical solutions of the present invention shall fall within the protection scope determined by the claims of the present invention.

Claims

1. A multi-component runoff segmentation method coupling nonlinear reservoir model and conductivity mass balance, characterized in that: include: Acquiring observation data of a target watershed, wherein the observation data includes runoff data and conductivity data; Extracting a baseflow recession segment based on the observed data, analyzing the baseflow recession segment, obtaining nonlinear reservoir parameters and conductivity-related parameters, and constructing a nonlinear reservoir model and a conductivity mass balance equation respectively; Based on the base flow decline segment, the independent event segment is divided, and the runoff is segmented in each independent event segment by using the nonlinear reservoir model and the conductivity mass balance equation, with the goal of minimizing the conductivity flux estimation error at the outlet of the target watershed, and the rapid runoff and base flow are obtained, and the old water proportion and new water proportion of the rapid runoff and base flow are calculated respectively; The rapid runoff and base flow segmented from each independent event segment, as well as the calculated old water proportion and new water proportion of the rapid runoff and base flow are summarized to obtain a multi-component segmentation result of the target watershed.

2. The runoff multi-component segmentation method for coupling nonlinear reservoir model and conductivity mass balance according to claim 1 is characterized in that: Acquiring the nonlinear reservoir parameters includes: Extracting the recession section without the interference of rapid runoff in the base flow recession section, and determining the nonlinear reservoir parameters k and β by binning regression method; Obtaining the conductivity-related parameters includes: Estimating the water-rock action correction term D by the conductivity change rate of the base flow recession section; Passive storage S of the air-filled zone is preset by parameter rate pa ; Determine the effective precipitation conductivity EC based on the average conductivity of the preset period P .

3. The runoff multi-component segmentation method for coupling nonlinear reservoir model and conductivity mass balance according to claim 2 is characterized in that: The nonlinear reservoir model is: Among them, Q b is the base flow, t is the time, and the parameter The transformation relationship between and ω and parameters k and β is: β=2-ω.

4. The runoff multi-component segmentation method for coupling nonlinear reservoir model and conductivity mass balance according to claim 2, characterized in that: The conductivity mass balance equation is: Rapid runoff dynamic conductivity equation: Base flow dynamic conductivity equation: Where t is time, Δt is the time step, is the rapid runoff conductivity at time t+Δt; R [t+Δt] is the infiltration recharge at time t+Δt; Q [t+Δt] is the total runoff at time t+Δt; is the base flow at time t+Δt; is the effective precipitation conductivity at time t+Δt; Q [t] is the total runoff at time t; is the base flow at time t; is the effective precipitation conductivity at time t; is the rapid runoff conductivity at time t; S pa Passive storage for the vadose zone; is the base flow conductivity at time t+Δt; k and β are nonlinear reservoir parameters; D is the water-rock effect correction term; is the base current conductivity at time t.

5. The runoff multi-component segmentation method for coupling nonlinear reservoir model and conductivity mass balance according to claim 1, characterized in that: The goal is to minimize the conductivity flux estimation error at the outlet of the target basin: Where t is time, Δt is the time step, and AE [t+Δt] is the conductivity flux estimation error at the outlet of the target basin at time t+Δt; Q [t+Δt] is the total runoff at time t+Δt; EC [t+Δt] is the total runoff conductivity at time t+Δt; is the base flow at time t+Δt; is the rapid runoff conductivity at time t+Δt; is the base current conductivity at time t+Δt.

6. The runoff multi-component segmentation method for coupling nonlinear reservoir model and conductivity mass balance according to claim 5, characterized in that: After iteratively selecting the best base flow sequence with the goal of minimizing the conductivity flux estimation error at the outlet of the target basin, iteratively selecting the passive storage capacity S of the vadose zone with the goal of minimizing the residual square sum of the conductivity flux pa The best value of is: Among them, SSE is the residual sum of squares, AE [t] is the conductivity flux estimation error at the target runoff outlet at time t, and nΔt is the length of the independent event segment sequence.

7. The runoff multi-component segmentation method for coupling nonlinear reservoir model and conductivity mass balance according to claim 1, characterized in that: The calculation of the old water percentage and new water percentage of the fast runoff and base flow includes: The proportion of old water in rapid runoff Q fo : The proportion of new water in rapid runoff Q fn :Q fn =Q f -Q fo ; The proportion of old water in base flow Q bo : The proportion of new water in base flow Q bn :Q bn =Q b -Q bo ; Among them, Q f For rapid runoff, EC f is the rapid runoff conductivity, EC P is the effective conductivity of precipitation, EC fo is the conductivity of old water in rapid runoff, Q b is the base flow, EC b is the base current conductivity, EC bo is the conductivity of old water in the base flow.

8. The runoff multi-component segmentation method for coupling nonlinear reservoir model and conductivity mass balance according to any one of claims 1 to 7, characterized in that: The multi-component segmentation result of the target watershed satisfies the water balance ∑(Q bo +Q bn +Q fo +Q fn )=Q 总径流 .

Images