CN117786614B - Real-time flood forecast rainfall error correction method and device considering time lag influence - Google Patents

Abstract

The present invention discloses a method and device for correcting rainfall errors in real-time flood forecasts that considers the influence of time lag. The method comprises: collecting measured rainfall and outlet section runoff data of different historical floods, and calculating the average confluence time of the basin; using the measured and calculated surface average rainfall sequence as the input of the basin hydrological model to obtain a simulated runoff sequence, and calculating the runoff error sequence in combination with the measured runoff sequence; using the ARMA model to analyze the runoff error sequence to obtain a prediction sequence, thereby obtaining a new runoff error sequence; calculating the hysteresis sensitivity of the simulated runoff output by the hydrological model to the surface average rainfall, and obtaining a hysteresis sensitivity matrix; based on the new runoff error sequence and the hysteresis sensitivity matrix, calculating the surface average rainfall error according to the least squares principle, correcting the surface average rainfall sequence, and using the corrected surface average rainfall to perform runoff simulation forecasting. The method solves the insufficient correction of precipitation errors caused by time lag and improves the accuracy of flood forecasting.

Description

Real-time flood forecast rainfall error correction method and device considering time lag effect

Technical Field

The invention relates to a method for correcting rainfall errors in real-time flood forecasting, and belongs to the technical field of real-time correction of flood forecasting.

Background Art

Flood forecasting, as the main non-structural measure to mitigate flood risks, is widely adopted by engineering technicians and managers from a practical perspective. How to further ensure the accuracy of flood forecasting has always been a major concern for hydrological forecasters. However, due to the uncertainty of input data, model structure and model parameter estimation, the forecast results are usually not accurate enough, which has prompted researchers to further improve the prediction accuracy from multiple perspectives, including the development of real-time error correction technology.

Precipitation is the main driving input of hydrological model forecasts, and the uncertainty of its observation data is an important source of uncertainty in flood forecasts. Improving the accuracy of flood forecasts by minimizing precipitation uncertainty has become an important research topic in real-time flood forecast error correction. The precipitation error correction technology based on the dynamic system response curve method can effectively improve the accuracy of flood forecasts by establishing a system response relationship between precipitation input and model output, and reversely solving the precipitation input error based on the model output calculation error to correct the precipitation.

However, in the actual application of this method, the hysteresis response relationship between rainfall and runoff caused by the confluence effect in the actual basin is ignored. This actual hysteresis response relationship will lead to the lack of information for error correction, which is specifically manifested in that the latest monitored precipitation observation values cannot be corrected, and thus affects the update effect of flood forecast. Therefore, how to improve the accuracy of flood forecasting through rainfall error correction based on the full consideration of the hysteresis relationship between rainfall and runoff in the actual basin is an urgent problem to be solved in the field of flood forecast error correction technology.

Summary of the invention

Purpose of the invention: In view of the shortcomings of the prior art, the present invention proposes a real-time flood forecast rainfall error correction method and device taking into account the influence of time lag, which can fully consider the time lag characteristics of rainfall-runoff formation in the actual basin, solve the shortcomings of precipitation error correction caused by time lag, and help improve the accuracy of flood forecasting.

Technical solution: A real-time flood forecast rainfall error correction method considering the effect of time lag, comprising the following steps:

(1) Collect the measured rainfall data and outlet runoff data of different flood events in the basin history, calculate the average rainfall, count the time difference between the peak time of the average rainfall in different events and the peak time of the corresponding runoff process, and calculate the average time difference to obtain the average confluence time step δ of the basin;

(2) The surface average rainfall sequence P = [p ₁ ,p ₂ ,…,p _m ] calculated based on the real-time monitoring rainfall data of the basin is used as the input of the basin hydrological model, where p _m represents the mth surface average rainfall. The simulated runoff sequence Qc = [qc ₁ ,qc ₂ ,…,qc _n ] is obtained using the basin hydrological model, where qc _n represents the nth simulated runoff, n>m. Then, based on the measured runoff sequence Qo = [qo ₁ ,qo ₂ ,…,qo _m ] monitored in real time, the runoff error sequence ΔQ = Qo-Qc = [qo ₁ -qc ₁ ,qo ₂ -qc ₂ ,…,qo _m -qc _m ] = [Δq ₁ ,Δq ₂ ,…,Δq _m ] is calculated;

(3) The ARMA model is used to analyze the runoff error sequence and obtain the predicted runoff error sequence [Δq _m+1 , Δq _m+2 , …, Δq _m+δ ], and the original runoff error sequence is integrated with the predicted sequence to obtain a new runoff error sequence ΔQ' = [Δq _1+δ , Δq _2+δ , …, Δq _m+δ ];

(4) Apply disturbance to each rainfall in the area average rainfall sequence P, together with the original area average rainfall sequence, as the input of the basin hydrological model, calculate and output the corresponding simulated runoff, calculate the lag sensitivity of the simulated runoff sequence to the area average rainfall, and obtain the lag sensitivity matrix S;

(5) Based on the new runoff error sequence ΔQ' and the lagged sensitivity matrix S, the surface average rainfall error sequence ΔP = [Δp ₁ , Δp ₂ , …, Δp _m ] is calculated according to the least squares principle. The surface average rainfall sequence P is corrected according to ΔP, and the corrected surface average rainfall sequence is input into the basin hydrological model to re-simulate the runoff forecast.

Furthermore, in step (1), the average rainfall on the surface is calculated using the Thiessen polygon method.

Furthermore, in step (2), the watershed hydrological model adopts the Xin'anjiang model.

Furthermore, in step (3), the ARMA model is expressed as:

Where: j is the autoregressive order, k is the moving average order, φ _i is the autoregressive coefficient, is the moving average coefficient, ε _t and ε _ti are white noises that satisfy Gaussian distribution at time t and time ti, and Δq _ti is the ti-th runoff error.

Furthermore, the autoregressive order j and the moving average order are determined by the AIC criterion, expressed as:

AIC＝Nln(MSE)+2K

Where: N is the actual number of samples, K is the number of model parameters, and MSE is the mean square error.

Furthermore, the step (4) comprises:

Apply the disturbance term dp ₁ to p ₁ in the average rainfall sequence P = [p ₁ ,p ₂ ,…, _pm ] to obtain the sequence P+dp ₁ = [p ₁ +dp ₁ ,p ₂ ,…, _pm ], and use the two rainfall sequences as the input of the basin hydrological model to obtain the simulated runoff Qc(P) = [qc ₁ ,qc ₂ ,…,qc _n ] and The lag sensitivity of the output runoff series to the area average rainfall p ₁ is:

Apply the disturbance term dp ₂ to p ₂ in the average rainfall sequence P = [p ₁ ,p ₂ ,…, _pm ] to obtain the sequence P+dp ₂ = [p ₁ ,p ₂ +dp ₂ ,…, _pm ], and use the two rainfall sequences as the input of the basin hydrological model to obtain the simulated runoff Qc(P) = [qc ₁ ,qc ₂ ,…,qc _n ] and The lag sensitivity of the output runoff series to the area average rainfall _p2 is:

Repeat the above steps until the hysteresis sensitivity calculation of the output runoff sequence to all values in the surface average rainfall sequence is completed, and finally the hysteresis sensitivity matrix S is obtained:

Furthermore, in step (5), the precipitation error sequence ΔP is calculated according to the least squares principle, and the expression is:

ΔP＝(S ^T S) ^-1 S ^T ΔQ'

The original surface average rainfall sequence is corrected using the solved surface average rainfall error sequence:

P'＝P+ΔP＝[p ₁ +Δp ₁ , p ₂ +Δp ₂ ,…, p _m +Δp _m ]

P' is the corrected surface average rainfall series.

The present invention also provides a real-time flood forecast rainfall error correction device taking into account the influence of time lag, comprising:

The average confluence time calculation module is used to collect the measured rainfall data and outlet section runoff data of different flood events in the basin history, calculate the average rainfall, count the time difference between the peak time of the average rainfall in different events and the peak time of the corresponding runoff process, and calculate the average value of the time difference to obtain the average confluence time step δ of the basin;

a runoff error sequence calculation module, for using the surface average rainfall sequence P = [p ₁ ,p ₂ ,…, _pm ] calculated based on the real-time monitoring rainfall data of the watershed as the input of the watershed hydrological model, _wherepm represents the mth surface average rainfall, and using the watershed hydrological model to obtain the simulated runoff sequence Qc = [qc ₁ ,qc ₂ ,…,qc _n ], whereqc _n represents the nth simulated runoff, n>m, and then calculating the runoff error sequence ΔQ = Qo-Qc = [qo ₁ -qc ₁ ,qo ₂ -qc ₂ ,…,qo _m -qc _m ] = [ _Δq 1 ,Δq ₂ ,…,Δq _m ] based on the real-time monitored measured runoff sequence Qo = [qo ₁ ,qo ₂ ,…,qo _m ];

The runoff error sequence prediction and update module is used to analyze the runoff error sequence using the moving autoregressive average model ARMA to obtain the predicted runoff error sequence [Δq _m+1 ,Δq _m+2 ,…,Δq _m+δ ], and integrate the original runoff error sequence with the predicted sequence to obtain a new runoff error sequence ΔQ'=[Δq _1+δ ,Δq _2+δ ,…,Δq _m+δ ];

The hysteresis sensitivity calculation module is used to apply disturbance to each rainfall in the surface average rainfall sequence P, together with the original surface average rainfall sequence, as the input of the basin hydrological model, calculate and output the corresponding simulated runoff, calculate the hysteresis sensitivity of the simulated runoff sequence to the surface average rainfall, and obtain the hysteresis sensitivity matrix S;

The rainfall error correction module is used to calculate the surface average rainfall error sequence ΔP = [Δp ₁ , Δp ₂ , …, Δp _m ] based on the new runoff error sequence ΔQ' and the lag sensitivity matrix S according to the least squares principle, correct the surface average rainfall sequence P according to ΔP, and input the corrected surface average rainfall sequence into the basin hydrological model to re-carry out runoff simulation forecast.

The present invention also provides a computer device, comprising: one or more processors; a memory; and one or more programs, wherein the one or more programs are stored in the memory and are configured to be executed by the one or more processors, and when the programs are executed by the processors, the steps of the real-time flood forecast rainfall error correction method considering the effect of time lag as described above are implemented.

The present invention also provides a computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the steps of the real-time flood forecast rainfall error correction method considering the time lag effect as described above are implemented.

Compared with the prior art, the present invention has the following beneficial effects:

The present invention overcomes the deficiency of incomplete precipitation correction caused by failure to consider the rainfall-runoff lag relationship when correcting rainfall errors in flood forecasts based on dynamic system response curves. This invention is of great significance for perfecting the theory of implementing flood forecasting based on dynamic system response curves and improving the accuracy of real-time flood forecasting.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG1 is a flow chart of a method for correcting rainfall errors in real-time flood forecasting taking into account the influence of time lag according to the present invention.

DETAILED DESCRIPTION

The technical solution of the present invention is further described below in conjunction with the accompanying drawings.

As shown in FIG1 , a real-time flood forecast rainfall error correction method considering the time lag effect includes the following steps:

Step (1), collect the measured rainfall data and outlet section runoff data of different flood events in the basin history, calculate the average rainfall, count the time difference between the peak time of the average rainfall in different events and the peak time of the corresponding runoff process, calculate the average value of the time difference to obtain the average confluence time step δ of the basin;

In an embodiment of the present invention, hourly measured rainfall data and outlet section runoff data of 13 rain gauges in different historical flood events in the Dapoling Basin from 1991 to 2010 are collected, and the surface average rainfall is calculated using the Thiessen polygon method. The time difference between the peak time of the surface average rainfall in different events and the peak time of the corresponding runoff process is counted, and the average value of the time difference is calculated to obtain the basin average confluence time step δ of 9 hours.

Step (2): based on the real-time monitoring rainfall data, the surface average rainfall sequence P = [ _p1 , _p2 , ..., _pm ] is calculated, where m is the number of real-time monitoring rainfalls. P is generally a time series, that is, P is composed of the surface average rainfall at the first moment _p1 to the surface average rainfall at the mth moment _pm . This sequence is also called the surface average rainfall sequence or the surface average rainfall sequence. P is input into the watershed hydrological model to obtain the calculated runoff (also called simulated runoff) sequence Qc = [ _qc1 , _qc2 , ..., _qcn ], where n is the number of calculated runoffs, n>m. Through the watershed hydrological model, the runoff for a period of time in the past (1 to m) and the future (m to n) can be calculated and simulated from the known precipitation (1 to m), where [ _qcm , qcm ₊₁ , ..., _qcn ] is the predicted runoff. The measured runoff sequence based on real-time monitoring is Qo = [ _qo1 , _qo2 , ..., _qom ] Calculate the runoff error sequence ΔQ = Qo-Qc = [qo ₁ -qc ₁ ,qo ₂ -qc ₂ ,…,qo _m -qc _m ] = [Δq ₁ ,Δq ₂ ,…,Δq _m ];

Real-time correction means that after collecting the observed average rainfall, the model is immediately run to simulate the runoff. At the same time, the model is corrected through the past monitored runoff (1~m) to achieve the process of updating the future calculated runoff (m~n). The time point of the entire operation is at m, 1~m is the past, and m~n is the future.

In the embodiment of the present invention, the basin hydrological model adopts the Xin'anjiang model. Taking the 19960707 flood as an example, the surface average rainfall P = [p ₁ , p ₂ , …, p ₅₂ ] calculated based on the real-time monitoring rainfall data is used as the input of the Xin'anjiang model, and the simulated runoff Qc = [qc ₁ , qc ₂ , …, qc _n ] is calculated by the model. The measured runoff monitored in real time is Qo = [qo ₁ , qo ₂ , …, qo ₅₂ ], and the difference between the measured runoff and the simulated runoff is calculated as the runoff error: ΔQ = Qo-Qc = [qo ₁ -qc ₁ , qo ₂ -qc ₂ , …, qo ₅₂ -qc ₅₂ ] = [Δq ₁ , Δq ₂ , …, Δq ₅₂ ], and some calculation results are shown in Table 1.

Table 1. Average rainfall, simulated runoff, measured runoff, and runoff error series

Step (3), using the moving autoregressive average model ARMA to analyze the runoff error sequence, obtain the predicted runoff error sequence [Δq _m+1 , Δq _m+2 , …, Δq _m+δ ], integrate the original runoff error sequence with the predicted sequence to obtain a new runoff error sequence ΔQ' = [Δq _1+δ , Δq _2+δ , …, Δq _m+δ ];

Due to the existence of the average runoff time δ in the basin, the surface average rainfall _pt at time t will not affect the calculated runoff until t+δ and later, which is the so-called delayed response. Therefore, when the surface average rainfall error is inferred by calculating the runoff error, the surface average precipitation at time 1 to m is corrected, which requires a runoff error of 1+δ to m+δ. Since the measured runoff is only observed at time m, only the runoff error of 1+δ to m can be calculated, while the part of m+1 to m+δ can be obtained through ARMA prediction.

Among them, the ARMA model is expressed as:

Where: j is the autoregressive order, k is the moving average order, φ _i is the autoregressive coefficient, is the moving average coefficient, _εt and _εti are white noises satisfying Gaussian distribution at time t and time ti;

Integrating the original runoff error sequence with the predicted sequence means splicing the two sequences together and getting the part we need from them, i.e., from Δq _1+δ to Δq _m+δ .

In the embodiment of the present invention, the ARMA model is used to analyze the runoff error sequence to obtain a predicted runoff error sequence [Δq ₅₃ , Δq ₅₄ , …, Δq ₆₁ ], and the original runoff error sequence and the predicted sequence are concatenated to obtain a new runoff error sequence ΔQ'=[Δq ₁₀ , Δq ₁₁ , …, Δq ₆₁ ].

The present invention determines the autoregressive order j and the moving average order k based on the AIC criterion. Different (j, k) combinations obtain different ARMA model expressions. Different expressions have different fits for ΔQ=[Δq ₁ , Δq ₂ ,…, Δq _m ] (expressed by MSE mean square error). The best combination of (j, k) is determined by minimizing AIC (highest fit). The AIC criterion is expressed as:

AIC＝Nln(MSE)+2K

Where: N is the actual number of samples, K is the number of model parameters, and MSE is the mean square error.

In the embodiment of the present invention, the autoregressive order j=3 and the moving average order k=1 are determined based on the AIC criterion, the AIC calculated value is 449.29, and the multi-step runoff error [Δq ₅₃ ,Δq ₅₄ ,…Δq ₆₁ ]=[31.43,70.41,102.84,125.84,134.45,130.23,117.09,100.87,86.79] is predicted based on the ARMA model. The original runoff error sequence and the predicted sequence are concatenated to obtain a new runoff error sequence ΔQ'.

Step (4), applying disturbance to each rainfall in the surface average rainfall sequence, together with the original surface average rainfall sequence, as the input of the basin hydrological model, using the model to calculate the corresponding simulated runoff output, calculating the lag sensitivity of the simulated runoff sequence to the surface average rainfall, and obtaining the lag sensitivity matrix S;

Specifically, first, a disturbance term dp ₁ (such as 0.001) is applied to p ₁ in the area average rainfall sequence P = [p ₁ , p ₂ , …, p _m ] to obtain the sequence P + dp ₁ = [p ₁ + dp ₁ , p ₂ , …, p _m ], and the two rainfall sequences are respectively used as inputs of the basin hydrological model (the Xin'anjiang model in this embodiment), and the corresponding simulated runoff Qc (P) = [qc ₁ , qc ₂ , …, qc _n ] and According to the following formula, the hysteresis sensitivity of the output runoff series to the average rainfall p ₁ is:

Correspondingly, the disturbance term dp ₂ (such as 0.001) is applied to p ₂ in the area average rainfall sequence P = [p ₁ ,p ₂ ,…, _pm ] to obtain the sequence P+dp ₂ = [p ₁ ,p ₂ +dp ₂ ,…, _pm ], and the two rainfall sequences are used as the input of the Xin'anjiang model to obtain the corresponding simulated runoff Qc(P) = [qc ₁ ,qc ₂ ,…,qc _n ] and Then, according to the following formula, the hysteresis sensitivity of the runoff sequence to the average rainfall _p2 is:

Repeat the above steps until the hysteresis sensitivity calculation of the output runoff sequence to all values in the surface average rainfall sequence is completed, and finally the hysteresis sensitivity matrix S is obtained:

In the embodiment of the present invention, the specific calculation results of the lag sensitivity matrix S are shown in Table 2.

Table 2 Hysteresis sensitivity matrix S

Step (5), based on the new runoff error sequence ΔQ' and the lagged sensitivity matrix S, the surface average rainfall error ΔP = [Δp ₁ , Δp ₂ , ..., Δp _m ] is calculated according to the least squares principle, and the surface average rainfall P is corrected; the corrected surface average rainfall is input into the basin hydrological model to re-simulate the runoff forecast.

The precipitation error ΔP is calculated based on the least squares principle, and the expression is:

ΔP＝(S ^T S) ^-1 S ^T ΔQ'

The original surface average rainfall is corrected using the solved surface average rainfall error: P' = P + ΔP = [p ₁ + Δp ₁ , p ₂ + Δp ₂ , …, p _m + Δp _m ], where P' is the corrected surface average rainfall; the corrected surface average rainfall is used as the input of the Xin'anjiang model, and the model is recalculated to obtain a new simulated forecast runoff process Qu. The obtained rainfall error ΔP, the corrected surface average rainfall P' and the new simulated forecast runoff Qu are shown in Table 3.

Table 3 Rainfall error ΔP, corrected surface average rainfall P’ and new simulated runoff forecast Qu

In this embodiment, the deterministic coefficient NSE and the relative error of the flood peak, which are commonly used indicators in flood forecasting, are used to evaluate the correction effect of the method. The measured runoff sequence is Qo = [qo ₁ ,qo ₂ ,…,qo _n ], and the corresponding simulated runoff sequence is Qc = [qc ₁ ,qc ₂ ,…,qc _n ], and the deterministic coefficient NSE is calculated as follows:

In the formula is the mean of the measured runoff.

The peak value of the measured runoff sequence is _qop , the peak value of the simulated runoff sequence is _qcp , and the relative error of the peak value is calculated as follows:

The calculation results of the coefficient of certainty and relative error of flood peak in this embodiment are shown in Table 4.

Table 4 Evaluation results of the coefficient of certainty and relative error of flood peak before and after correction

According to the results in Table 4, it can be seen that after the correction method proposed in the present invention is used to correct the precipitation error, the runoff simulation forecast accuracy is significantly improved: the certainty coefficient is increased from 0.95 to 0.98, and the relative error of the flood peak is reduced from -14.49% to -4.71%.

Based on the same technical concept as the method embodiment, another embodiment of the present invention further provides a real-time flood forecast rainfall error correction device considering the influence of time lag, comprising:

The average confluence time calculation module is used to collect the measured rainfall data and outlet section runoff data of different flood events in the basin history, calculate the average rainfall, count the time difference between the peak time of the average rainfall in different events and the peak time of the corresponding runoff process, and calculate the average value of the time difference to obtain the average confluence time step δ of the basin;

a runoff error sequence calculation module, for using the surface average rainfall sequence P = [p ₁ ,p ₂ ,…, _pm ] calculated based on the real-time monitoring rainfall data of the watershed as the input of the watershed hydrological model, _wherepm represents the mth surface average rainfall, and using the watershed hydrological model to obtain the simulated runoff sequence Qc = [qc ₁ ,qc ₂ ,…,qc _n ], whereqc _n represents the nth simulated runoff, n>m, and then calculating the runoff error sequence ΔQ = Qo-Qc = [qo ₁ -qc ₁ ,qo ₂ -qc ₂ ,…,qo _m -qc _m ] = [ _Δq 1 ,Δq ₂ ,…,Δq _m ] based on the real-time monitored measured runoff sequence Qo = [qo ₁ ,qo ₂ ,…,qo _m ];

The runoff error sequence prediction and update module is used to analyze the runoff error sequence using the moving autoregressive average model ARMA to obtain the predicted runoff error sequence [Δq _m+1 ,Δq _m+2 ,…,Δq _m+δ ], and integrate the original runoff error sequence with the predicted sequence to obtain a new runoff error sequence ΔQ'=[Δq _1+δ ,Δq _2+δ ,…,Δq _m+δ ];

The hysteresis sensitivity calculation module is used to apply disturbance to each rainfall in the surface average rainfall sequence P, together with the original surface average rainfall sequence, as the input of the basin hydrological model, calculate and output the corresponding simulated runoff, calculate the hysteresis sensitivity of the simulated runoff sequence to the surface average rainfall, and obtain the hysteresis sensitivity matrix S;

The rainfall error correction module is used to calculate the surface average rainfall error sequence ΔP = [Δp ₁ , Δp ₂ , …, Δp _m ] based on the new runoff error sequence ΔQ' and the lag sensitivity matrix S according to the least squares principle, correct the surface average rainfall sequence P according to ΔP, and input the corrected surface average rainfall sequence into the basin hydrological model to re-carry out runoff simulation forecast.

It should be understood that the real-time flood forecast rainfall error correction device taking into account the influence of time lag in the embodiment of the present invention can implement all the technical solutions in the above-mentioned method embodiment, and the functions of its various functional modules can be specifically implemented according to the methods in the above-mentioned method embodiments. The specific implementation process can refer to the relevant description in the above-mentioned embodiments, which will not be repeated here.

The present invention also provides a computer device, comprising: one or more processors; a memory; and one or more programs, wherein the one or more programs are stored in the memory and are configured to be executed by the one or more processors, and when the programs are executed by the processors, the steps of the real-time flood forecast rainfall error correction method considering the effect of time lag as described above are implemented.

The present invention also provides a computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the steps of the real-time flood forecast rainfall error correction method considering the time lag effect as described above are implemented.

It will be appreciated by those skilled in the art that embodiments of the present invention may be provided as methods, devices (systems), computer equipment or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program codes.

The present invention is described with reference to a flow chart of a method according to an embodiment of the present invention. It should be understood that each process in the flow chart and a combination of processes in the flow chart can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device produce a device for implementing the functions specified in one or more processes in the flow chart.

These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer-readable memory produce a product including an instruction device that implements the functions specified in one or more processes of the flowchart.

These computer program instructions may also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, whereby the instructions executed on the computer or other programmable device provide steps for implementing the functions specified in one or more processes in the flowchart.

Claims (9)

1. The real-time flood forecast rainfall error correction method considering time lag influence is characterized by comprising the following steps of:

(1) Collecting actual measurement rainfall data and outlet section runoff data of different flood fields of a basin history, calculating the average rainfall of the surface, counting the time difference between the average rainfall peak time of different fields and the flood peak time of the corresponding runoff process, and calculating the average value of the time difference to obtain the average confluence time step delta of the basin;

(2) Taking a surface average rainfall sequence P= [ P ₁,p₂,…,p_m ] calculated based on the basin real-time monitoring rainfall data as input of a basin hydrological model, wherein P _m represents the m-th surface average rainfall, obtaining a simulated runoff sequence Qc= [ Qc ₁,qc₂,…,qc_n],qc_n represents the n-th simulated runoff by using the basin hydrological model, and calculating a runoff error sequence based on the real-time monitoring actual measurement runoff sequence Qo= [ Qo ₁,qo₂,…,qo_m ] ΔQ＝Qo-Qc＝[qo₁-qc₁,qo₂-qc₂,…,qo_m-qc_m]＝[Δq₁,Δq₂,…,Δq_m];

(3) Analyzing the runoff error sequence by adopting a moving autoregressive average model ARMA to obtain a predicted runoff error sequence [ delta Q _m+1,Δq_m+2,…,Δq_m+δ ], and integrating the original runoff error sequence and the predicted sequence to obtain a new runoff error sequence delta Q' = [ delta Q _1+δ,Δq_2+δ,…,Δq_m+δ ];

(4) Disturbance is applied to each rainfall in the surface average rainfall sequence P, the disturbance is used as input of a watershed hydrological model together with the original surface average rainfall sequence, corresponding simulated runoffs are calculated and output, and hysteresis sensitivity of the simulated runoff sequence to the surface average rainfall is calculated to obtain a hysteresis sensitivity matrix S;

(5) Based on a new runoff error sequence delta Q' and a hysteresis sensitivity matrix S, calculating a surface average rainfall error sequence delta P= [ delta P ₁,Δp₂,…,Δp_m ] according to a least square principle, correcting a surface average rainfall sequence P according to the delta P, and inputting the corrected surface average rainfall sequence into a watershed hydrological model to carry out runoff simulation forecast again;

wherein, the step (4) specifically comprises:

applying a disturbance term dp ₁ to P ₁ in the opposite average rainfall sequence P= [ P ₁,p₂,…,p_m ] to obtain a sequence P+dp ₁＝[p₁+dp₁,p₂,…,p_m, and respectively taking the two rainfall sequences as input of a watershed hydrologic model to obtain simulated runoff Qc (P) = [ Qc ₁,qc₂,…,qc_n ] and The lag sensitivity of the output runoff sequence to the surface average rainfall p ₁ is:

Applying a disturbance term dp ₂ to P ₂ in the opposite average rainfall sequence P= [ P ₁,p₂,…,p_m ] to obtain a sequence P+dp ₂＝[p₁,p₂+dp₂,…,p_m, and respectively taking the two rainfall sequences as input of a watershed hydrologic model to obtain simulated calculation runoff Qc (P) = [ Qc ₁,qc₂,…,qc_n ] and The lag sensitivity of the output runoff sequence to the surface average rainfall p ₂ is:

repeating the steps until the calculation of the hysteresis sensitivity of the output runoff sequence to all the values in the surface average rainfall sequence is completed, and finally obtaining a hysteresis sensitivity matrix S:

2. The method according to claim 1, wherein in the step (1), the average rainfall on the surface is calculated by using a Thiessen polygon method.

3. The method of claim 1, wherein in step (2), the basin hydrologic model uses a new enjiang model.

4. The method of claim 1, wherein in step (3), the ARMA model is expressed as:

Wherein: j is the autoregressive order, k is the moving average order, phi _i is the autoregressive coefficient, For the moving average coefficients, ε _t and ε _t-i are white noise satisfying Gaussian distribution at time t and time t-i, and Δq _t-i is the t-i th runoff error.

5. The method of claim 4, wherein the autoregressive order j and the moving average order are determined by AIC criteria, expressed as:

AIC＝Nln(MSE)+2K

wherein: n is the actual number of samples, K is the number of model parameters, and MSE is the mean square error.

6. The method according to claim 1, wherein in the step (5), the precipitation error sequence Δp is calculated according to a least squares principle, and the expression is:

ΔP＝(S^TS)^-1S^TΔQ'

Correcting the original surface average rainfall sequence by using the solved surface average rainfall error sequence:

P'＝P+ΔP＝[p₁+Δp₁,p₂+Δp₂,…,p_m+Δp_m]

P' is the modified face average rainfall sequence.

7. A real-time flood forecast rainfall error correction device taking time lag effects into consideration, comprising:

The average converging time calculation module is used for collecting actual measurement rainfall data and outlet section runoff data of different flood scenes of the basin history, calculating the average rainfall on the surface, counting time differences between the average rainfall peak time of the different scenes and the flood peak time of the corresponding runoff process, and calculating the average value of the time differences to obtain the average converging time step delta of the basin;

The runoff error sequence calculation module is used for taking a surface average rainfall sequence P= [ P ₁,p₂,…,p_m ] calculated based on the real-time monitoring rainfall data of the river basin as input of a hydrologic model of the river basin, P _m represents the m-th surface average rainfall, obtaining a simulated runoff sequence Qc= [ Qc ₁,qc₂,…,qc_n],qc_n represents the n-th simulated runoff by using the hydrologic model of the river basin, n > m, and calculating the runoff error sequence based on the real-time monitoring actual measurement runoff sequence Qo= [ Qo ₁,qo₂,…,qo_m ] ΔQ＝Qo-Qc＝[qo₁-qc₁,qo₂-qc₂,…,qo_m-qc_m]＝[Δq₁,Δq₂,…,Δq_m];

The runoff error sequence prediction updating module is used for analyzing the runoff error sequence by adopting a moving autoregressive average model ARMA to obtain a predicted runoff error sequence [ delta Q _m+1,Δq_m+2,…,Δq_m+δ ], and integrating the original runoff error sequence and the predicted sequence to obtain a new runoff error sequence delta Q' = [ delta Q _1+δ,Δq_2+δ,…,Δq_m+δ ];

The hysteresis sensitivity calculation module is used for applying disturbance to each rainfall in the surface average rainfall sequence P, taking the disturbance together with the original surface average rainfall sequence as the input of a watershed hydrological model, calculating and outputting corresponding simulated runoffs, and calculating the hysteresis sensitivity of the simulated runoff sequence to the surface average rainfall to obtain a hysteresis sensitivity matrix S;

The rainfall error correction module is used for calculating a face average rainfall error sequence delta P= [ delta P ₁,Δp₂,…,Δp_m ] according to the least square principle based on a new runoff error sequence delta Q' and a hysteresis sensitivity matrix S, correcting a face average rainfall sequence P according to the delta P, and inputting the corrected face average rainfall sequence into a watershed hydrological model to carry out runoff simulation forecast again;

The hysteresis sensitivity calculation module specifically includes:

applying a disturbance term dp ₁ to P ₁ in the opposite average rainfall sequence P= [ P ₁,p₂,…,p_m ] to obtain a sequence P+dp ₁＝[p₁+dp₁,p₂,…,p_m, and respectively taking the two rainfall sequences as input of a watershed hydrologic model to obtain simulated runoff Qc (P) = [ Qc ₁,qc₂,…,qc_n ] and The lag sensitivity of the output runoff sequence to the surface average rainfall p ₁ is:

Applying a disturbance term dp ₂ to P ₂ in the opposite average rainfall sequence P= [ P ₁,p₂,…,p_m ] to obtain a sequence P+dp ₂＝[p₁,p₂+dp₂,…,p_m, and respectively taking the two rainfall sequences as input of a watershed hydrologic model to obtain simulated calculation runoff Qc (P) = [ Qc ₁,qc₂,…,qc_n ] and The lag sensitivity of the output runoff sequence to the surface average rainfall p ₂ is:

repeating the steps until the calculation of the hysteresis sensitivity of the output runoff sequence to all the values in the surface average rainfall sequence is completed, and finally obtaining a hysteresis sensitivity matrix S:

8. a computer device, comprising:

One or more processors;

A memory; and

One or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, which when executed by the processors implement the steps of the real-time flood forecast rainfall error correction method taking into account the lag time effect of any of claims 1-6.

9. A computer readable storage medium having stored thereon a computer program, wherein the computer program when executed by a processor implements the steps of the real-time flood forecast rainfall error correction method taking into account the lag time effect as claimed in any of claims 1-6.