A parameter calibration method for WRF-Hydro model

Abstract

The present invention discloses a parameter calibration method for a WRF-Hydro model, comprising the following main steps: selecting parameters to be calibrated in the WRF-Hydro model and determining a calibration order; selecting indicators for evaluating simulation results; dividing flood events into rate-period events and test-period events; determining the value range of each parameter; calculating the error vector or error matrix of each indicator; calculating the error vector and standardizing it; repeating the steps until the standard error vectors of all indicators are calculated; obtaining the optimal estimate of the current parameter according to the decision vector; and repeating the steps until the optimal estimate of all parameters is obtained. The method of the present invention provides a quantitative multi-indicator parameter calibration method, which can comprehensively, objectively and quantitatively determine the parameters to be calibrated of the WRF-Hydro model.

Description

Technical Field

[0001] The invention belongs to the technical field of hydrology and water conservancy, and in particular relates to a parameter calibration method of a WRF-Hydro model. Background Art

[0002] The WRF-Hydro model is a new generation of distributed hydrometeorological model system. The physics-based WRF-Hydro model can be coupled with the WRF model in one or two directions. In recent years, the model has been widely used around the world and has shown good potential in hydrometeorological simulation, flood forecasting, water resources management and evaluation. However, the WRF-Hydro model involves a large number of parameters, some of which are preset parameters and the other are parameters to be calibrated. Preset parameters refer to parameter values ​​that are predefined and do not change with the application of the model in different regions; parameters to be calibrated need to be determined based on basin geographic information and observation data. In addition, because the explicit difference method is used in the calculation of slope and river confluence, the calculation step length of the WRF-Hydro model is short and the calculation cost is high. The large number of parameters to be calibrated and the high calculation cost make it impossible to use traditional calibration methods to determine the parameters of the WRF-Hydro model. Therefore, some scholars have proposed the use of the manual stepwise approximation (MSA) method to calibrate the main parameters of the model, and this method has been widely used in related studies. However, there is a lack of a systematic quantitative decision-making basis when using the MSA method to calibrate the model. For example, in some studies, the parameters of the WRF-Hydro model are obtained by comprehensively analyzing the performance of multiple indicators. Summary of the invention

[0003] Purpose of the invention: In view of the above problems, the present invention proposes a parameter calibration method for the WRF-Hydro model.

[0004] Technical solution: To achieve the purpose of the present invention, the technical solution adopted by the present invention is: a parameter calibration method of the WRF-Hydro model, which specifically includes the following steps:

[0005] Step 1: Select the parameters to be calibrated for the WRF-Hydro model, and determine the calibration order of the parameters to be calibrated according to certain standards, and obtain the parameter vector θ = (θ1, θ2, …, θ i ,…,θ I ) T , 1≤i≤I; where θ i represents the i-th parameter to be calibrated, and I represents the total number of parameters to be calibrated;

[0006] Step 2: Determine the evaluation index for the simulation results of the WRF-Hydro model, and arbitrarily sort the evaluation indexes to obtain the evaluation index vector η = (η1, η2, …, η j ,…,η J ) T , 1≤j≤J; where η j represents the jth evaluation index, and J represents the total number of evaluation indexes;

[0007] Step 3: Divide flood events into rate-period events and test-period events, sort flood events within the rate-period by time, and obtain the rate-period flood event vector ζ = (ζ1, ζ2, …, ζ k ,…,ζ K ) T , 1≤k≤K, K∈N + ; Among them, ζ k represents the kth rate-regular flood event, K represents the total number of rate-regular flood events; N + represents a positive integer;

[0008] Step 4: According to the calibration order of the parameters to be calibrated, the optimal estimated value is determined for each parameter to be calibrated in turn using the evaluation index vector and the rate period flood event vector, thereby obtaining the optimal estimated value vector of all parameters to be calibrated of the WRF-Hydro model. in, represents the optimal estimate of the i-th parameter to be calibrated.

[0009] Furthermore, in step 3, if the rate period contains only one flood event or a preset time continuous runoff process, the total number of rate period flood events K=1, and the rate period flood event vector ζ=(ζ1); otherwise, the total number of rate period flood events K>1, and each flood event or preset long-term continuous runoff process is regarded as a rate period flood event to form a rate period flood event vector ζ=(ζ1,ζ2,…,ζ k ,…,ζ K ) T , 1<k≤K.

[0010] Furthermore, in step 4, the method for determining the optimal estimated value for the i-th parameter to be calibrated is as follows:

[0011] Step 4.1, determine the value range of the i-th parameter to be calibrated, and obtain the value vector of the i-th parameter to be calibrated Among them, ξ l represents the lth value of the i-th parameter to be calibrated; L i represents the total number of values ​​of the i-th parameter to be calibrated;

[0012] Step 4.2, determine whether the total number of regular flood events K is greater than 1;

[0013] If it is established, the utilization rate periodic flood event vector and the value vector of the i-th parameter to be calibrated are used to calculate the index matrix and error matrix of the i-th parameter to be calibrated under each evaluation index, and the error vector is calculated based on the error matrix;

[0014] If it is not true, then the index vector and error vector of the i-th parameter to be calibrated are calculated for each evaluation index of the i-th parameter to be calibrated by using the periodic flood event vector and the value vector of the i-th parameter to be calibrated;

[0015] Step 4.3, standardize the error vector obtained in step 4.2 using the deviation standardization method;

[0016] Step 4.4, using the normalized error vector of the i-th parameter to be calibrated under each evaluation index, calculate the decision vector of the i-th parameter to be calibrated;

[0017] The value of the parameter to be calibrated corresponding to the minimum element in the decision vector is the optimal estimated value of the parameter to be calibrated.

[0018] Furthermore, in step 4.2, if the total number of flood events K in the rate period is greater than 1, the index matrix of the i-th parameter to be calibrated under the j-th evaluation index is M=M Li×K , the element is m lk ; Among them, m lk It represents the result value of the jth evaluation index under the kth rate-determined flood event when the i-th parameter to be calibrated takes the l-th value;

[0019] The error matrix M of the i-th parameter to be calibrated under the j-th evaluation index error for:

[0020] M error =MM ideal , K>1

[0021] In the formula, M represents the index matrix of the i-th parameter to be calibrated under the j-th evaluation index, M ideal represents the ideal index matrix of the i-th parameter to be calibrated under the j-th evaluation index; the ideal index matrix is ​​when the simulation results of the WRF-Hydro model are completely consistent with the observation results, the index matrix M at this time is the ideal index vector M ideal ; The simulation results are obtained by WRF-Hydro model simulation, and the observation results are obtained from the hydrological station data;

[0022] The error vector p of the i-th parameter to be calibrated under the j-th evaluation indexj for:

[0023]

[0024] In the formula, m error,lk is the element in the error matrix of the i-th parameter to be calibrated under the j-th evaluation index, specifically indicating the error result value of the j-th evaluation index under the k-th rate-determined flood event when the i-th parameter to be calibrated takes the l-th value; p l It indicates the error result value of the jth evaluation index when the i-th parameter to be calibrated takes the l-th value.

[0025] Furthermore, in step 4.2, if the total number of flood events in the rate period K is equal to 1, the index vector of the i-th parameter to be calibrated under the j-th evaluation index is m=m Li×1 , the element is m l ; Among them, m l It indicates the result value of the jth evaluation index when the i-th parameter to be calibrated takes the l-th value;

[0026] The error vector p of the i-th parameter to be calibrated under the j-th evaluation index j for:

[0027] p j =m error =mm ideal ,m error,l ∈m error ,K=1

[0028] In the formula, m error represents the error vector of the i-th parameter to be calibrated under the j-th evaluation index; m ideal represents the ideal index vector of the i-th parameter to be calibrated under the j-th evaluation index; the ideal index vector is the index vector m when the simulation results of the WRF-Hydro model are completely consistent with the observation results. error ; The simulation results are obtained by WRF-Hydro model simulation, and the observation results are obtained from the hydrological station data; m error,l It is the element in the error vector of the i-th parameter to be calibrated under the j-th evaluation index, specifically indicating the error result value of the j-th evaluation index when the i-th calibrated parameter takes the l-th value.

[0029] Furthermore, the error vector obtained in step 4.2 is normalized by using the deviation normalization method in step 4.3, and the method is as follows:

[0030]

[0031] In the formula, q lrepresents the standardized error result value of the jth evaluation index when the i-th parameter to be calibrated takes the l-th value; p l represents the error result value of the jth evaluation index when the i-th parameter to be calibrated takes the l-th value; p t It represents the error result value of the jth evaluation index when the i-th parameter to be calibrated takes the t-th value; q j Represents the standardized error vector of the jth evaluation indicator.

[0032] Furthermore, in step 4.4, the decision vector of the i-th parameter to be calibrated is calculated by using the standardized error vector of the i-th parameter to be calibrated under each evaluation index, and the method is as follows:

[0033] f i =f(q1,q2,…,q j ,…,q J ),j∈J

[0034] In the formula, f i represents the decision vector of the i-th parameter to be calibrated; f() represents the objective function; q j Represents the standardized error vector of the jth evaluation indicator.

[0035] Beneficial effects: Compared with the prior art, the technical solution of the present invention has the following beneficial technical effects:

[0036] The parameter calibration method of the WRF-Hydro model described in the present invention first selects the parameters to be calibrated in the WRF-Hydro model and determines the calibration order; selects the index for evaluating the simulation results; divides the flood events into the rate period events and the test period events; determines the value range of each parameter; calculates the index error vector or index error matrix of each index; calculates the error vector and standardizes it; repeats until the standard error vector calculation of all indexes is completed; obtains the optimal estimate of the current parameter according to the decision vector; thereby obtains the optimal estimate of all parameters. The method of the present invention provides a quantitative multi-index parameter calibration method, which can comprehensively, objectively and quantitatively determine the parameters to be calibrated of the WRF-Hydro model. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] Figure 1 The present invention is a block diagram of a parameter calibration method of a WRF-Hydro model according to an embodiment of the present invention. DETAILED DESCRIPTION

[0038] The technical solution of the present invention is further described below in conjunction with the accompanying drawings and embodiments.

[0039] The parameter calibration method of the WRF-Hydro model described in the present invention is based on Figure 1, specifically including the following steps:

[0040] S1: Select the parameters to be calibrated in the WRF-Hydro model and determine the calibration order, which specifically includes the following steps:

[0041] 1) All parameters use default values ​​(or preset values).

[0042] 2) Select several parameters to be calibrated for the WRF-Hydro model. For example, select ZSOILFAC (soil thickness multiplier), REFKDT (permeability parameter), LKSATFAC (lateral saturated hydraulic conductivity multiplier), GWEXP (groundwater reservoir outflow index), OVROUGHRTFAC (surface roughness multiplier), and MANNFAC (river channel Manning roughness multiplier) as parameters to be calibrated for the WRF-Hydro model.

[0043] 3) According to the runoff principle, the parameter calibration order is determined to form the parameter vector θ. For example, ZSOILFAC and REFKDT mainly control the runoff calculation, and the two are calibrated first (θ1 and θ2 respectively); LKSATFAC controls the lateral movement of soil in the soil layer, and this movement affects both surface runoff and underground runoff, so this parameter needs to be used as θ3; GWEXP and OVROUGHRTFAC control the calculation of underground runoff (base flow) and overland flow respectively, so the two are θ4 and θ5 respectively; finally, MANNFAC controls the calculation of river flow, so this parameter is the last parameter to be calibrated θ6.

[0044] S2: Select the indicators for evaluating the simulation results, which specifically include the following steps:

[0045] 1) Select appropriate evaluation indicators to evaluate the simulation results of the WRF-Hydro model according to the needs. For example, select the relative error of flood volume (PB fv ), peak relative error (PB pf ), peak time error (AB pt ), root mean square error (RMSE) and Nash coefficient (NSE) as evaluation indicators;

[0046] 2) Determine the parameters according to a certain order to form the evaluation index vector η. The order here can be set arbitrarily. For example, PB fv PB pf , AB pt , RMSE and NSE are defined as η1, η2, η3, η4 and η5 respectively.

[0047] S3: Divide flood events into rate-period events and test-period events, which specifically includes the following steps:

[0048] 1) Divide flood events into rate-period events and test-period events according to relevant standards or requirements;

[0049] 2) All events within the rate period are organized into a flood event vector ζ according to the time order; if the rate period contains only one flood event or a long-term continuous runoff process (K = 1), then this process constitutes ζ1; if the rate period contains K flood events (K>1), then these K floods constitute ζ1, ζ2, ..., ζ K .

[0050] S4: Determine the value range of each parameter, which specifically includes the following steps:

[0051] 1) For parameter θ i , determine its value range. For example, the values ​​of MANNFAC are 0.2, 0.4, 0.6, 0.8, 1.0, 1.2, 1.4, 1.6, 1.8, 2.0.

[0052] 2) These values ​​constitute the parameter θ i The value vector ξ θi ; For example, MANNFAC is defined as ξ1, ξ2, …, ξ Li .

[0053] S5: Calculate the indicator error vector or indicator error matrix of each indicator, specifically including the following steps:

[0054] 1) For the index η j , when K = 1, the index η j As the vector ξ changes, the calculated values ​​of these indicators form a vector, called the indicator vector (m Li×1 ); when the rate period contains multiple floods (K>1), η j As the vectors ξ and ζ change, the calculated values ​​of these indicators form a matrix, called the indicator matrix (M Li×K ); the elements of m or M are respectively m l or m lk (1≤k≤K,1≤l≤L i , and k,l∈N + );

[0055] 2) For the index η j , so that the simulation results and the observed results are completely consistent, and m or M is calculated to become an ideal indicator vector (m ideal ) or matrix (M ideal ); for example, PB fv M ideal is 0 Li×K , NSE's M ideal is 1 Li×K ;

[0056] 3) For the index η j , the corresponding indicator error vector (m error ) or indicator error matrix (M error ) is calculated by the following formula:

[0057]

[0058] S6: Calculate the error vector and normalize it, specifically including the following steps:

[0059] 1) For the index η j , when K=1, m error It is further defined as the error vector (p j ); when K>1, calculate M error The sum of squares of the lth row (p l ), these p l The vector p j :

[0060]

[0061] 2) Use the deviation standardization method (also called min-max standardization method) to convert p j Treated as a standard error vector (q j ):

[0062]

[0063] S7: Repeat steps S5 to S6 until the standard error vectors of all indicators are calculated.

[0064] S8: Obtaining the optimal estimated value of the current parameter according to the decision vector, specifically including the following steps:

[0065] 1) For parameter θ i , calculate the parameter θ through a suitable objective function i The decision vector (f i ):

[0066] f i =f(q1,q2,…,q j ,…,q J ),j∈J

[0067] 2) For parameter θ i , according to the decision vector f i The smallest element determines the parameter θ i The optimal value of ; in particular, when the parameter θ is i When different values ​​are taken but the simulation results are the same (i.e. f i does not exist), parameter θ i The default value must be used:

[0068] S9: Repeat steps 4 to 8 until all parameters are processed:

[0069]

[0070] For example, θ in S1 = (θ1, θ2, θ3, θ4, θ5, θ6) T The optimal estimate (corresponding to ZSOILFAC, REFKDT, LKSATFAC, GWEXP, OVROUGHRTFAC, and MANNFAC, respectively) is

[0071] Furthermore, the calculation method of the parameters of the WRF-Hydro model is determined, specifically including:

[0072] According to the calibration order of the parameters to be calibrated, the optimal estimated value is determined for each parameter to be calibrated in turn using the evaluation index vector and the rate period flood event vector, thereby obtaining the optimal estimated value vector of all parameters to be calibrated of the WRF-Hydro model. in, represents the optimal estimate of the i-th parameter to be calibrated.

[0073] Furthermore, the method for determining the optimal estimate for the i-th parameter to be calibrated is as follows:

[0074] Step 4.1, determine the value range of the i-th parameter to be calibrated, and obtain the value vector of the i-th parameter to be calibrated Among them, ξ l represents the lth value of the i-th parameter to be calibrated; L i represents the total number of values ​​of the i-th parameter to be calibrated;

[0075] Step 4.2, determine whether the total number of regular flood events K is greater than 1;

[0076] If it is established, the utilization rate periodic flood event vector and the value vector of the i-th parameter to be calibrated are used to calculate the index matrix and error matrix of the i-th parameter to be calibrated under each evaluation index, and the error vector is calculated based on the error matrix;

[0077] If it is not true, then the index vector and error vector of the i-th parameter to be calibrated are calculated for each evaluation index of the i-th parameter to be calibrated by using the periodic flood event vector and the value vector of the i-th parameter to be calibrated;

[0078] Step 4.3, standardize the error vector obtained in step 4.2 using the deviation standardization method;

[0079] Step 4.4, using the normalized error vector of the i-th parameter to be calibrated under each evaluation index, calculate the decision vector of the i-th parameter to be calibrated;

[0080] The value of the parameter to be calibrated corresponding to the minimum element in the decision vector is the optimal estimated value of the parameter to be calibrated.

[0081] Furthermore, in step 4.2, if the total number of flood events K in the rate period is greater than 1, the index matrix of the i-th parameter to be calibrated under the j-th evaluation index is M=M Li×K , the element is m lk ; Among them, m lk It represents the result value of the jth evaluation index under the kth rate-determined flood event when the i-th parameter to be calibrated takes the l-th value;

[0082] The error matrix M of the i-th parameter to be calibrated under the j-th evaluation index error for:

[0083] M error =MM ideal , K>1

[0084] In the formula, M represents the index matrix of the i-th parameter to be calibrated under the j-th evaluation index, M ideal represents the ideal index matrix of the i-th parameter to be calibrated under the j-th evaluation index;

[0085] The error vector p of the i-th parameter to be calibrated under the j-th evaluation index j for:

[0086]

[0087] In the formula, m error,lk is the element in the error matrix of the i-th parameter to be calibrated under the j-th evaluation index, specifically indicating the error result value of the j-th evaluation index under the k-th rate-determined flood event when the i-th parameter to be calibrated takes the l-th value; p l It indicates the error result value of the jth evaluation index when the i-th parameter to be calibrated takes the l-th value.

[0088] Furthermore, in step 4.2, if the total number of flood events in the rate period K is equal to 1, the index vector of the i-th parameter to be calibrated under the j-th evaluation index is m=m Li×1 , the element is m l ; Among them, m l It indicates the result value of the jth evaluation index when the i-th parameter to be calibrated takes the l-th value;

[0089] The error vector p of the i-th parameter to be calibrated under the j-th evaluation index jfor:

[0090] p j =m error =mm ideal ,m error,l ∈m error ,K=1

[0091] In the formula, m error represents the error vector of the i-th parameter to be calibrated under the j-th evaluation index; m ideal represents the ideal index vector of the i-th parameter to be calibrated under the j-th evaluation index; m error,l It is the element in the error vector of the i-th parameter to be calibrated under the j-th evaluation index, specifically indicating the error result value of the j-th evaluation index when the i-th calibrated parameter takes the l-th value.

[0092] Furthermore, the error vector obtained in step 4.2 is standardized by using the deviation standardization method in step 4.3, and the method is as follows:

[0093]

[0094] In the formula, q l represents the standard error value of the jth evaluation index when the i-th parameter to be calibrated takes the l-th value; p l represents the error result value of the jth evaluation index when the i-th parameter to be calibrated takes the l-th value; p t It represents the error result value of the jth evaluation index when the i-th parameter to be calibrated takes the t-th value; q j Represents the error standard vector of the jth evaluation indicator.

[0095] Furthermore, in step 4.4, the decision vector of the i-th parameter to be calibrated is calculated by using the standardized error vector of the i-th parameter to be calibrated under each evaluation index, and the method is as follows:

[0096] f i =f(q1,q2,…,q j ,…,q J ),j∈J

[0097] In the formula, f i represents the decision vector of the i-th parameter to be calibrated; f() represents the objective function; q j Represents the error standard vector of the jth evaluation indicator.

[0098] The above are only preferred embodiments of the present invention, and the present invention is not limited to the contents of the embodiments. For those skilled in the art, various changes and modifications can be made within the scope of the technical solution of the present invention, and any changes and modifications made are within the protection scope of the present invention.

Claims

1. A parameter calibration method for the WRF-Hydro model, characterized in that: The specific steps include: Step 1: Select the parameters to be calibrated for the WRF-Hydro model and determine the calibration order of the parameters to be calibrated to obtain the parameter vector θ = (θ1, θ2, …, θ i ,…,θ I ) T , 1≤i≤I; where θ i represents the i-th parameter to be calibrated, and I represents the total number of parameters to be calibrated; soil thickness multiplier, infiltration parameter, lateral saturated hydraulic conductivity multiplier, groundwater reservoir outflow index, surface roughness multiplier, and river channel Manning roughness multiplier are selected as parameters to be calibrated for the WRF-Hydro model; Step 2: Determine the evaluation index for the simulation results of the WRF-Hydro model, and arbitrarily sort the evaluation indexes to obtain the evaluation index vector η = (η1, η2, …, η j ,…,η J ) T , 1≤j≤J; where η j represents the jth evaluation index, J represents the total number of evaluation indexes; the relative error of flood volume, the relative error of flood peak, the error of peak appearance time, the root mean square error and the Nash coefficient are selected as evaluation indexes; Step 3: Divide flood events into rate-period events and test-period events, sort flood events within the rate-period by time, and obtain the rate-period flood event vector ζ = (ζ1, ζ2, …, ζ k ,…,ζ K ) T , 1≤k≤K, K∈N + ; Among them, ζ k represents the kth rate-regular flood event, K represents the total number of rate-regular flood events; N + represents a positive integer; Step 4: According to the calibration order of the parameters to be calibrated, the optimal estimated value is determined for each parameter to be calibrated in turn using the evaluation index vector and the rate period flood event vector, thereby obtaining the optimal estimated value vector of all parameters to be calibrated of the WRF-Hydro model. in, represents the optimal estimate of the i-th parameter to be calibrated; In step 3, if the rate period contains only one flood event or a continuous runoff process of a preset time, the total number of flood events in the rate period K=1, and the rate period flood event vector ζ=(ζ1); otherwise, the total number of flood events in the rate period K>1, and each flood event or a continuous runoff process of a preset time is regarded as a rate period flood event to form a rate period flood event vector ζ=(ζ1,ζ2,…,ζ k ,…,ζ K ) T , 1<k≤K.

2. The parameter calibration method of a WRF-Hydro model according to claim 1, characterized in that: In step 4, the method for determining the optimal estimated value for the i-th parameter to be calibrated is as follows: Step 4.1, determine the value range of the i-th parameter to be calibrated, and obtain the value vector of the i-th parameter to be calibrated Among them, ξ l represents the lth value of the i-th parameter to be calibrated; L i represents the total number of values ​​of the i-th parameter to be calibrated; Step 4.2, determine whether the total number of regular flood events K is greater than 1; If it is established, the utilization rate periodic flood event vector and the value vector of the i-th parameter to be calibrated are used to calculate the index matrix and error matrix of the i-th parameter to be calibrated under each evaluation index, and the error vector is calculated based on the error matrix; If it is not true, then the index vector and error vector of the i-th parameter to be calibrated are calculated for each evaluation index of the i-th parameter to be calibrated by using the periodic flood event vector and the value vector of the i-th parameter to be calibrated; Step 4.3, standardize the error vector obtained in step 4.2 using the deviation standardization method; Step 4.4, using the normalized error vector of the i-th parameter to be calibrated under each evaluation index, calculate the decision vector of the i-th parameter to be calibrated; The value of the parameter to be calibrated corresponding to the minimum element in the decision vector is the optimal estimated value of the parameter to be calibrated; when the decision vector of the i-th parameter to be calibrated does not exist, the i-th parameter to be calibrated takes the default value.

3. The parameter calibration method of a WRF-Hydro model according to claim 2, characterized in that: In step 4.2, if the total number of flood events K in the calibration period is greater than 1, the index matrix of the i-th parameter to be calibrated under the j-th evaluation index is M=M Li×K , the element is m lk ; Among them, m lk It represents the result value of the jth evaluation index under the kth rate-determined flood event when the i-th parameter to be calibrated takes the l-th value; The error matrix M of the i-th parameter to be calibrated under the j-th evaluation index error for: M error =M-M ideal ,K>1 In the formula, M represents the index matrix of the i-th parameter to be calibrated under the j-th evaluation index, M ideal represents the ideal index matrix of the i-th parameter to be calibrated under the j-th evaluation index; The error vector p of the i-th parameter to be calibrated under the j-th evaluation index j for: In the formula, m error,lk is the element in the error matrix of the i-th parameter to be calibrated under the j-th evaluation index, specifically indicating the error result value of the j-th evaluation index under the k-th rate-determined flood event when the i-th parameter to be calibrated takes the l-th value; p l It indicates the error result value of the jth evaluation index when the i-th parameter to be calibrated takes the l-th value.

4. The parameter calibration method of a WRF-Hydro model according to claim 2, characterized in that: In step 4.2, if the total number of flood events K in the calibration period is equal to 1, the index vector of the i-th parameter to be calibrated under the j-th evaluation index is m=m Li×1 , the element is m l ; Among them, m l It indicates the result value of the jth evaluation index when the i-th parameter to be calibrated takes the l-th value; The error vector p of the i-th parameter to be calibrated under the j-th evaluation index j for: p j =m error =m-m ideal ,m error,l ∈m error ,K=1 In the formula, m error represents the error vector of the i-th parameter to be calibrated under the j-th evaluation index; m ideal represents the ideal index vector of the i-th parameter to be calibrated under the j-th evaluation index; m error,l It is the element in the error vector of the i-th parameter to be calibrated under the j-th evaluation index, specifically indicating the error result value of the j-th evaluation index when the i-th calibrated parameter takes the l-th value.

5. The parameter calibration method of a WRF-Hydro model according to claim 2, characterized in that: Step 4.3 uses the deviation standardization method to standardize the error vector obtained in step 4.2, and the method is as follows: In the formula, q l represents the standardized error result value of the jth evaluation index when the i-th parameter to be calibrated takes the l-th value; p l represents the error result value of the jth evaluation index when the i-th parameter to be calibrated takes the l-th value; p t It represents the error result value of the jth evaluation index when the i-th parameter to be calibrated takes the t-th value; q j Represents the standardized error vector of the jth evaluation indicator.

6. A parameter calibration method for a WRF-Hydro model according to claim 2, characterized in that: Step 4.4 uses the standardized error vector of the i-th parameter to be calibrated under each evaluation index to calculate the decision vector of the i-th parameter to be calibrated, and the method is as follows: f i =f(q1,q2,…,q j ,…,q J ),j∈J In the formula, f i represents the decision vector of the i-th parameter to be calibrated; f() represents the objective function; q j Represents the standardized error vector of the jth evaluation indicator.

Images