Distributed hydrological forecasting method of confluence mode of confluence while demonstration

Abstract

The invention provides a distributed hydrological forecasting method in a confluence mode of confluence while performing confluence, which comprises the following steps of: analyzing and researching a confluence flow relationship of each grid in a drainage basin, and dividing all grids into a plurality of grades; establishing a hydrological forecasting model; the hydrological forecasting model comprises an underground runoff confluence model and a surface runoff confluence model of a combined runoff production model; the hydrological forecasting model is trained, and the hydrological forecasting model after training is completed is obtained; and carrying out outflow forecasting on each grid of the research basin by adopting the trained hydrological forecasting model. The distributed hydrological forecasting method of the confluence mode while performing confluence, provided by the invention, is a confluence method while performing confluence while performing confluence, can realize forecasting of outflow flow of each grid in a drainage basin, and has the advantages of high hydrological forecasting precision and high accuracy.

Description

Technical Field

[0001] The present invention relates to the technical field of hydrology and water resources, and in particular to a distributed hydrological forecasting method in a simultaneous flow-merging and confluence mode. Background Art

[0002] The VIC model (Variable Infiltration Capacity Model) is a semi-distributed hydrological model used to simulate hydrological processes and energy balance in the land surface. The VIC model can simulate large-scale hydrology without detailed terrain data by considering the spatial variation of surface cover and soil heterogeneity. The model can also fully simulate hydrological processes including infiltration, soil moisture dynamics, evaporation and plant transpiration, so that it can accurately respond to hydrological changes. This is of great significance to improving the accuracy of flood simulation.

[0003] When using the VIC model for flood simulation, the method of first performing and then merging is now mostly used. The method of first performing and then merging is: establish a hydrological routing model for the main and tributary rivers of the calculated river section, and obtain the flood outflow process of each river section through the hydrological routing model. Then, at the final outflow section, linearly superimpose the flood outflow processes of all river sections to obtain the outflow process. This method of first performing and then merging is suitable for basins where the main and tributary rivers have little mutual influence, and has certain limitations. In addition, the method of first performing and then merging uses a unit line for confluence, which can only obtain the flow process of the outlet section of the basin, and cannot obtain the flow process of the outlet of each river section. Since the flow processes at different river sections in the basin have their own characteristics, the method of first performing and then merging cannot meet the calculation of the flow process at any river section in the basin. How to achieve accurate calculation of the flow process at any river section in the basin is a key issue that needs to be solved at present. Summary of the invention

[0004] In view of the defects of the prior art, the present invention provides a distributed hydrological forecasting method with a simultaneous runoff and confluence mode, which can effectively solve the above problems.

[0005] The technical solution adopted by the present invention is as follows:

[0006] The present invention provides a distributed hydrological forecasting method in a simultaneous flow-merging mode, comprising the following steps:

[0007] Step S1, gridding the study basin, dividing the study basin into N grids;

[0008] Step S2, analyzing the confluence flow relationship of N grids, thereby dividing all grids into M levels, which are represented as: the first level R1, the second level R2, ..., the Mth level R M ;

[0009] Among them, there is no confluence relationship between the grids in each same level; each grid in the Mth level RM M is a grid that does not have an upstream grid in the research basin, and is a non-confluence grid; each grid from the (M - 1)th level RM - 1 M-1 to the 1st level R1 is a confluence grid, having at least one directly connected upstream grid, and, each grid from the (M - 1)th level RM - 1 M-1 to the 2nd level R2 has a unique directly connected downstream grid, and the 1st level R1 has only one grid, which is the final confluence grid of the other N - 1 grids;

[0010] Step S3, establish a hydrological forecasting model; the hydrological forecasting model includes a subsurface runoff confluence model and a surface runoff confluence model of a combined runoff generation model;

[0011] Step S4, determine the training period T; determine the initial values of the model parameters of the hydrological forecasting model;

[0012] Step S5, train the hydrological forecasting model to obtain the trained hydrological forecasting model;

[0013] Step S5.1, according to the calculation order of the grids from the Mth level RM M to the 1st level R1, adopt the side - calculating - side - combining confluence method, and through the hydrological forecasting model, sequentially calculate the subsurface runoff time series and the surface runoff time series of each grid at each moment t during the training period T, and add the subsurface runoff time series and the surface runoff time series of each grid at each moment t during the training period T at each moment t to obtain the outflow discharge time series of each grid at each moment t during the training period T;

[0014] Step S5.2, select a landmark grid in the research basin; by analyzing the outflow discharge time series of the landmark grid at each moment t during the training period T, evaluate whether the currently obtained hydrological forecasting model meets the accuracy requirements; if not, adjust the model parameters of the hydrological forecasting model and return to Step S5.1; if so, obtain the trained hydrological forecasting model;

[0015] Step S6, use the trained hydrological forecasting model to forecast the outflow discharge of each grid in the research basin.

[0016] Preferably, Step S2 is specifically:

[0017] Step S2.1, among the N grids in the research basin, find the grid with the largest cumulative confluence flow, and set its level as the 1st level R1;

[0018] Step S2.2, set the initial value of the level variable RANK to 1;

[0019] Step S2.3: In the study basin, locate the grid at the RANK level, called grid g; traverse the eight neighboring grids of grid g that have not been assigned a level. If the outflow of the traversed grid converges into grid g, set the level of the traversed grid to RANK + 1.

[0020] Step S2.4: If the levels of all N grids have been set, the grid level calculation ends; otherwise, set RANK = RANK + 1 and return to Step S2.3.

[0021] Preferably, the subsurface runoff concentration model of the combined runoff generation model includes the subsurface runoff concentration sub-model for non-concentrating grids shown in formula (1) and the subsurface runoff concentration sub-model for concentrating grids shown in formula (2):

[0022]

[0023] Where:

[0024] is the subsurface runoff of grid j at time t;

[0025] KGG is the subsurface runoff recession coefficient, which is a model parameter to be adjusted;

[0026] is the subsurface runoff of grid j at time t - 1, and its initial value is set to 0;

[0027] is the subsurface runoff generated by grid j itself at time t, which is predicted by the runoff generation model;

[0028] is the subsurface runoff inflow of grid j at time t, which is determined by formula (3):

[0029]

[0030] Where: represents the subsurface runoff generated by the upstream grid i directly connected to grid j itself at time t; upi represents the number of upstream grids i directly connected to grid j.

[0031] Preferably, the surface runoff concentration model includes the surface runoff concentration sub-model for non-concentrating grids shown in formula (4) and the surface runoff concentration sub-model for concentrating grids shown in formula (5):

[0032]

[0033] Where:

[0034] L is the grid length;

[0035] K is the conversion coefficient between flow rate and flow velocity; x is the flow rate proportion coefficient; is the conversion exponent between flow rate and flow velocity; K, x, are the model parameters to be adjusted;

[0036] is the surface runoff of grid j at time t;

[0037] is the surface runoff of grid j at time t - 1, and its initial value is set to 0;

[0038] Δt is the time interval between two adjacent times;

[0039] represents the surface runoff of the upstream grid i directly connected to grid j at time t; upi represents the number of upstream grids i directly connected to grid j;

[0040] represents the surface runoff of the upstream grid i directly connected to grid j at time t - 1.

[0041] Preferably, step S5 is specifically as follows:

[0042] Step S5.1, obtain the vegetation data, soil data, and meteorological data of each grid j in the study basin at each moment t during the training period T, and input them into the runoff generation model to obtain the time series of the base flow generated by each grid j at each moment t during the training period T t = 1, 2,..., T;

[0043] Step S5.2, determine the initial values of the model parameters of the hydrological forecasting model; the model parameters include: the base flow recession coefficient KGG, the conversion coefficient K between flow rate and flow velocity, the flow rate proportion coefficient x, the conversion exponent between flow rate and flow velocity

[0044] Step S5.3, perform calculations on each grid j in the Mth grade R M :

[0045] Adopt the base flow concentration sub - model of non - confluence grids shown in formula (1) to perform base flow calculations on each grid j in the Mth grade R M to obtain the time series of the base flow of each grid j in the Mth grade R M at each moment t during the training period T t = 1, 2,..., T;

[0046] Using the surface runoff concentration sub-model of the non-confluence grid shown in formula (4), perform surface runoff calculation for each grid j in the M-th level R M to obtain the surface runoff time series at each moment t of each grid j in the M-th level R M during the training period T, where t = 1, 2,..., T;

[0047] Using formula (6), add the subsurface runoff time series and the surface runoff time series at each moment t of each grid j in the M-th level R M during the training period T to obtain the outflow discharge time series at each moment t of each grid j, where and t = 1, 2,..., T; t = 1, 2,..., T;

[0048]

[0049] Step S5.4, perform calculations for each grid j in the (M - 1)-th level R M-1 as follows:

[0050] Using the subsurface runoff concentration sub-model of the confluence grid shown in formula (2), perform subsurface runoff calculation for each grid j in the (M - 1)-th level R M-1 to obtain the subsurface runoff time series at each moment t of each grid j in the (M - 1)-th level R M-1 during the training period T, where t = 1, 2,..., T;

[0051] Using the surface runoff concentration sub-model of the confluence grid shown in formula (5), perform surface runoff calculation for each grid j in the (M - 1)-th level R M-1 to obtain the surface runoff time series at each moment t of each grid j in the (M - 1)-th level R M-1 during the training period T, where t = 1, 2,..., T;

[0052] Using formula (6), add the subsurface runoff time series and the surface runoff time series at each moment t of each grid j in the (M - 1)-th level R M-1 during the training period T to obtain the outflow discharge time series at each moment t of each grid j, where and t = 1, 2,..., T; t = 1, 2,..., T;

[0053] Step S5.5, using the method of Step S5.4, successively perform calculations for each grid j in the (M - 2)-th level R M-2 and for each grid j in the (M - 3)-th level RM-3 Perform calculations for each grid j in it, and so on until the calculations for grid j in the first level R1 are completed;

[0054] Step S5.6, select landmark grids in the study basin; by analyzing the outflow discharge time series of the landmark grids at each moment t during the training period T, evaluate whether the currently obtained hydrological forecasting model meets the accuracy requirements; if not, adjust the model parameters of the hydrological forecasting model, including the groundwater runoff recession coefficient KGG, the conversion coefficient K between discharge and velocity, the discharge proportion coefficient x, and the conversion exponent between discharge and velocity Then return to step S5.3; if it meets the requirements, obtain the trained hydrological forecasting model.

[0055] Preferably, step S5.6 is specifically:

[0056] Among the N grids in the study basin, select the grid where the hydrological station is located as the landmark grid, denoted as landmark grid k;

[0057] Obtain the time series of observed outflow discharge values of landmark grid k at each moment t during the training period T Obtain the time series of observed outflow discharge values of each grid in the study basin at each moment t during the training period T, and take the average of the observed outflow discharge values of all grids at the same moment t to obtain the average time series of all observed outflow discharge values

[0058] Through this round of calculations, obtain the time series of outflow discharge of landmark grid k at each moment t during the training period T

[0060] Use formula (7) to obtain the Nash efficiency coefficient NSE:

[0061]

[0062] Evaluate whether the currently obtained hydrological forecasting model meets the accuracy requirements through the Nash efficiency coefficient NSE.

[0063] The distributed hydrological forecasting method with a confluence method of performing calculations while combining has the following advantages:

[0064] The present invention proposes a distributed hydrological forecasting method with a confluence method of performing calculations while combining, having the following characteristics: The present invention adopts the confluence method of performing calculations while combining, which can realize the forecasting of the outflow discharge of each grid in the basin, and has the advantages of high hydrological forecasting accuracy and high accuracy. Brief Description of the Drawings

[0065] Figure 1Flow chart of a distributed hydrological forecasting method with an edge-evolving and edge-merging confluence method provided by the present invention;

[0066] Figure 2 Schematic diagram of the grid calculation order of the Daning River Basin obtained in the embodiment of the present invention;

[0067] Figure 3 Runoff process graph at the basin outlet of the Daning River Basin during the training period obtained by using the traditional method of evolving first and then merging;

[0068] Figure 4 Runoff process graph at the basin outlet of the Daning River Basin during the training period obtained by using the edge-evolving and edge-merging method of the present invention;

[0069] Figure 5 Outflow discharge process graph of each grid during the training period T obtained by using the edge-evolving and edge-merging method of the present invention;

[0070] Figure 6 Outflow discharge process graph at the basin outlet during the verification period obtained by using the traditional method of evolving first and then merging;

[0071] Figure 7 Outflow discharge process graph at the basin outlet during the verification period obtained by using the edge-evolving and edge-merging method of the present invention. Specific implementation manner

[0072] In order to make the technical problems, technical solutions and beneficial effects solved by the present invention clearer, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0073] The present invention proposes a distributed hydrological forecasting method with an edge-evolving and edge-merging confluence method, which has the following characteristics: The present invention adopts the edge-evolving and edge-merging confluence method, which can realize the forecasting of the outflow discharge of each grid in the basin, and has the advantages of high hydrological forecasting accuracy and high accuracy; The present invention realizes the forecasting of the outflow discharge of each grid in the basin, which can effectively increase the hydrological data in the basin and has important reference significance for the management of basin water resources. The present invention has clear ideas, convenient operation and strong practicability.

[0074] Refer to Figure 1 , the present invention provides a distributed hydrological forecasting method with an edge-evolving and edge-merging confluence method, including the following steps:

[0075] Step S1, perform grid processing on the study basin and divide the study basin into N grids;

[0076] Specifically, the basin can be grid-processed based on the distribution of the basin and the spatial scale requirements to obtain a basin grid network.

[0077] Step S2: Analyze the confluence flow relationships of the N grids, and thus divide all the grids into M levels, which are respectively represented as: the first level R1, the second level R2, …, the Mth level R M ;

[0078] Among them, there is no confluence relationship between the grids in each same level; each grid in the Mth level R M is a grid that does not have an upstream grid in the research basin, and is a non-confluence grid; each grid from the (M - 1)th level R M-1 to the first level R1 is a confluence grid, which has at least one directly connected upstream grid, and moreover, each grid from the (M - 1)th level R M-1 to the second level R2 has a unique directly connected downstream grid, and the first level R1 has only one grid, which is the final confluence grid of the other N - 1 grids;

[0079] As a specific implementation manner, step S2 is specifically as follows:

[0080] Step S2.1: Among the N grids in the research basin, find the grid with the largest cumulative confluence flow, and set its level to the first level R1;

[0081] One way to obtain the grid with the largest cumulative confluence flow is: according to the topographic features of each grid in the research basin, estimate the grid with the largest cumulative confluence flow. Usually, the grid at the lowest topographic position is the grid with the largest cumulative confluence flow, and usually, it is the outlet grid of the entire basin.

[0082] Step S2.2: Set the initial value of the level variable RANK to 1;

[0083] Step S2.3: In the research basin, locate the grid of the RANKth level, called grid g; traverse the eight-neighborhood grids of grid g that have not been set with levels. If the outflow of the traversed grid converges into grid g, then set the level of the traversed grid to RANK + 1;

[0084] Step S2.4: If the levels of all N grids have been set, the grid level calculation ends; otherwise, let RANK = RANK + 1, and return to step S2.3.

[0085] In the present invention, the role of determining the level of each grid is: in the subsequent process, determine the calculation order of each grid according to the grid level, that is: perform calculations on each grid in the order from the Mth level R M to the first level R1.

[0086] Step S3: Establish a hydrological forecasting model; the hydrological forecasting model includes a subsurface runoff confluence model and a surface runoff confluence model of a combined runoff generation model;

[0087] For the subsurface runoff concentration model of the combined runoff generation model, a linear reservoir is used for concentration, specifically including the subsurface runoff concentration sub-model for non-concentrating grids shown in formula (1) and the subsurface runoff concentration sub-model for concentrating grids shown in formula (2):

[0088]

[0089] Where:

[0090] is the subsurface runoff at grid j at time t;

[0091] KGG is the subsurface runoff recession coefficient, which is a model parameter to be adjusted;

[0092] is the subsurface runoff at grid j at time t - 1, and its initial value is set to 0;

[0093] is the amount of subsurface runoff generated by grid j itself at time t, which is predicted by the runoff generation model;

[0094] is the subsurface runoff inflow at grid j at time t, which is determined by formula (3):

[0095]

[0096] Where: represents the amount of subsurface runoff generated by the upstream grid i directly connected to grid j at time t; upi represents the number of upstream grids i directly connected to grid j.

[0097] The surface runoff concentration model includes the surface runoff concentration sub-model for non-concentrating grids shown in formula (4) and the surface runoff concentration sub-model for concentrating grids shown in formula (5):

[0098]

[0099] Where:

[0100] L is the grid length;

[0101] K is the conversion coefficient between flow rate and flow velocity; x is the flow rate proportion coefficient; is the conversion exponent between flow rate and flow velocity; K, x, are model parameters to be adjusted;

[0102] is the surface runoff at grid j at time t;

[0103] is the surface runoff of grid j at time t-1, and its initial value is set to 0;

[0104] Δt is the time interval between two adjacent times;

[0105] represents the surface runoff of the upstream grid i directly connected to grid j at time t; upi represents the number of upstream grids i directly connected to grid j;

[0106] represents the surface runoff of the upstream grid i directly connected to grid j at time t-1.

[0107] Step S4, determine the training period T; determine the initial values of the model parameters of the hydrological forecasting model;

[0108] Step S5, train the hydrological forecasting model to obtain the trained hydrological forecasting model;

[0109] Step ①, according to the calculation order of the grids from the Mth level R M to the 1st level R1, adopt the side-calculation and side-confluence method, and through the hydrological forecasting model, calculate the groundwater runoff time series and surface runoff time series of each grid at each moment t in the training period T in turn. Add the groundwater runoff time series and surface runoff time series of each grid at each moment t in the training period T at each moment t to obtain the outflow discharge time series of each grid at each moment t in the training period T;

[0110] Step ②, select a landmark grid in the study basin; evaluate whether the currently obtained hydrological forecasting model meets the accuracy requirements by analyzing the outflow discharge time series of the landmark grid at each moment t in the training period T; if not, adjust the model parameters of the hydrological forecasting model and return to Step ①; if so, obtain the trained hydrological forecasting model;

[0111] Step S5 is specifically:

[0112] Step S5.1, obtain the vegetation data, soil data and meteorological data of each grid j in the study basin at each moment t in the training period T, input them into the runoff generation model, and obtain the time series of the groundwater runoff generated by itself of each grid j at each moment t in the training period T t = 1, 2,..., T;

[0113] In this step, the VIC runoff generation model can be used as the runoff generation model.

[0114] Step S5.2, determine the initial values of each model parameter of the hydrological forecasting model; the model parameters include: the groundwater runoff recession coefficient KGG, the conversion coefficient K between flow rate and flow velocity, the flow rate proportion coefficient x, and the conversion exponent between flow rate and flow velocity

[0115] Step S5.3, perform calculations for each grid j in the M-th level R M :

[0116] Adopt the groundwater runoff concentration sub-model of non-concentrating grids shown in formula (1) to perform groundwater runoff calculation for each grid j in the M-th level R M to obtain the groundwater runoff time series at each moment t during the training period T for each grid j in the M-th level R M t = 1, 2,..., T; t = 1, 2,..., T;

[0117] Adopt the surface runoff concentration sub-model of non-concentrating grids shown in formula (4) to perform surface runoff calculation for each grid j in the M-th level R M to obtain the surface runoff time series at each moment t during the training period T for each grid j in the M-th level R M t = 1, 2,..., T; t = 1, 2,..., T;

[0118] Adopt formula (6) to sum up the groundwater runoff time series M at each moment t during the training period T for each grid j in the M-th level R and the surface runoff time series at each moment t to obtain the outflow discharge time series at each moment t during the training period T for each grid j t = 1, 2,..., T;

[0119]

[0120] Step S5.4, perform calculations for each grid j in the (M - 1)-th level R M-1 :

[0121] Adopt the groundwater runoff concentration sub-model of concentrating grids shown in formula (2) to perform groundwater runoff calculation for each grid j in the (M - 1)-th level R M-1 to obtain the groundwater runoff time series at each moment t during the training period T for each grid j in the (M - 1)-th level R M-1 t = 1, 2,..., T; t = 1, 2,..., T;

[0122] Adopt the surface runoff concentration sub-model of concentrating grids shown in formula (5) to perform surface runoff calculation for each grid j in the (M - 1)-th level R M-1Perform surface runoff calculation for each grid j in [the relevant area], and obtain the runoff at the (M - 1)th level R M-1 The surface runoff time series of each grid j in the training period T at each moment t in [the relevant area] t = 1, 2,..., T;

[0123] Using formula (6), for the runoff at the (M - 1)th level R M-1 The groundwater runoff time series of each grid j in the training period T at each moment t in [the relevant area] And the surface runoff time series Perform addition at each moment t to obtain the outflow discharge time series of each grid j in the training period T at each moment t t = 1, 2,..., T;

[0124] Step S5.5, in the same way as in step S5.4, perform calculations on each grid j in the (M - 2)th level R M-2 in [the relevant area], perform calculations on each grid j in the (M - 3)th level R M-3 in [the relevant area], and so on, until the calculations on the grid j in the first level R1 are completed;

[0125] Step S5.6, select a landmark grid in the study basin; by analyzing the outflow discharge time series of the landmark grid at each moment t in the training period T, evaluate whether the current obtained hydrological forecasting model meets the accuracy requirements; if not, adjust the model parameters of the hydrological forecasting model, including the groundwater runoff recession coefficient KGG, the conversion coefficient K between discharge and velocity, the discharge proportion coefficient x, and the conversion exponent between discharge and velocity Then return to step S5.3; if it meets the requirements, obtain the trained hydrological forecasting model.

[0126] Step S5.6 is specifically as follows:

[0127] Among the N grids in the study basin, select the grid where the hydrological station is located as the landmark grid, denoted as landmark grid k;

[0128] Obtain the observed value time series of the outflow discharge of landmark grid k at each moment t in the training period T Obtain the observed value time series of the outflow discharge of each grid in the study basin at each moment t in the training period T, and take the average of the observed values of the outflow discharge of all grids at the same moment t to obtain the average value time series of all the observed values of the outflow discharge

[0129] Through this round of calculations, obtain the outflow discharge time series of landmark grid k at each moment t in the training period T

[0131] Using formula (7), obtain the Nash efficiency coefficient NSE:

[0132]

[0133] Evaluate whether the obtained hydrological forecasting model meets the accuracy requirements through the Nash efficiency coefficient NSE. The larger the value of the Nash efficiency coefficient NSE, the higher the accuracy of the hydrological forecasting model.

[0134] Step S6: Use the trained hydrological forecasting model to forecast the outflow discharge for each grid in the study basin.

[0135] Specifically, through training the hydrological forecasting model, the trained hydrological forecasting model is obtained, and its model parameters, namely the groundwater runoff recession coefficient KGG, the conversion coefficient K between discharge and velocity, the discharge proportion coefficient x, and the conversion exponent between discharge and velocity have all been determined and will no longer change; then, use the trained hydrological forecasting model to forecast the outflow discharge for each grid in the study basin.

[0136] Specifically, when conducting hydrological forecasting for the study basin, first obtain the vegetation data, soil data, and meteorological data of each grid during the forecasting period, and then input them into the VIC runoff generation model to obtain the time series of the groundwater runoff generated by each grid j at each moment t during the forecasting period. Then, use the groundwater runoff concentration model and the surface runoff concentration model of the trained combined runoff generation model, and according to the calculation order determined above, conduct combined concentration while calculating for each grid, and the outflow discharge of each grid at each moment t during the forecasting period can be obtained.

[0137] The following takes the Daning River Basin as an example to introduce an embodiment:

[0138] Step 1: Take the Daning River Basin as the study basin and divide the study basin into N grids;

[0139] Step 2: Based on ArcGIS, process the DEM data of the basin, divide all grids into several levels according to the level division method described above, and then determine the concentration calculation order according to the levels of each grid; as Figure 2 shown, it is a schematic diagram of the grid calculation order of the Daning River Basin.

[0140] Step 3: Determine the training period T from May 1, 2014 to October 31, 2014, and use the hydrological forecasting model and the training method of the hydrological forecasting model described above in the present invention to obtain a trained hydrological forecasting model that meets the accuracy requirements.

[0141] Using the trained hydrological forecasting model of the present invention, the following can be obtained as Figure 4The runoff process at the basin outlet during the training period T is shown. As a comparison, the traditional method of first routing and then combining is adopted to obtain the runoff process at the basin outlet of the Daning River Basin from May 1, 2014 to October 31, 2014 as shown in Figure 3 .

[0142] Furthermore, by using the hydrological forecasting model trained according to the present invention, the outflow discharge process of each grid during the training period T can also be obtained. For example, taking September 2, 2014 as an example, the outflow discharge process of each grid in the basin as shown in Figure 5 is obtained.

[0143] Step 4: Select a suitable verification period to verify the hydrological forecasting model trained according to the present invention.

[0144] Since the training period is the flood season in a year, the verification period is also selected as the flood season (May - October) for verification.

[0145] In this embodiment, May 2016 - October 2016 is selected as the verification period. As shown in Figure 6 , the runoff process diagram at the basin outlet during the verification period obtained by using the traditional method of first routing and then combining is shown. As shown in Figure 7 , the runoff process diagram at the basin outlet during the verification period obtained by using the method of routing while combining according to the present invention is shown.

[0146] Furthermore, calculate the Nash efficiency coefficients of the hydrological forecasting model trained according to the present invention and the traditional method of first routing and then combining for confluence respectively.

[0147] By using the method of routing while combining according to the present invention, its Nash efficiency coefficient is 0.926 during the training period and 0.805 during the verification period.

[0148] By using the traditional method of first routing and then combining for confluence, its Nash efficiency coefficient is 0.904 during the training period and 0.739 during the verification period.

[0149] From the results of the Nash efficiency coefficients, it can be seen that by using the method of routing while combining according to the present invention, its Nash efficiency coefficient is higher, the simulation result is better and the flow process of each grid can be obtained. Therefore, the method of routing while combining according to the present invention is suitable for the Daning River Basin.

[0150] The present invention provides a distributed hydrological forecasting method with a method of routing while combining, having the following characteristics:

[0151] (1) In the present invention, only one hydrological routing model is established for all grids within the basin, which is the hydrological forecasting model in the hydrological forecasting stage when the training is completed. By using this hydrological routing model, the hydrological routing in the training stage and the hydrological forecasting in the forecasting stage can be quickly realized. Since the present invention does not require establishing hydrological routing models for each river reach and tributary separately, the complexity of the system can be simplified, and the efficiency of hydrological routing in the training stage and hydrological forecasting in the hydrological forecasting stage can be improved.

[0152] (2) In the present invention, the routing order is determined according to the grades of each grid within the basin. According to the routing order, the subsurface runoff routing and surface runoff routing are performed on each grid in turn, and then the subsurface runoff routing result and surface runoff routing result of each grid are superimposed, so that the outflow discharge of each grid can be obtained. Moreover, when performing routing on each grid according to the routing order, the subsurface runoff routing result and surface runoff routing result of the upstream grid are used as the input of the hydrological routing model of its directly connected downstream grid, and then the subsurface runoff routing and surface runoff routing are performed on the downstream grid. Therefore, the present invention realizes the convergent routing of performing routing and convergence simultaneously for each grid within the basin, and can improve the accuracy of the convergent routing of the outflow discharge of each grid.

[0153] (3) In the present invention, in the hydrological forecasting stage, the forecasting of the outflow discharge of each grid in the basin can be realized, and the accuracy of hydrological forecasting is improved.

[0154] (4) The hydrological forecasting model of the present invention comprehensively considers the subsurface runoff and surface runoff affected by the runoff generation model, thereby improving the accuracy of the outflow discharge of each grid in the basin.

[0155] The above is only the preferred embodiment of the present invention. It should be noted that for those of ordinary skill in the art, without departing from the principle of the present invention, several improvements and refinements can be made, and these improvements and refinements should also fall within the protection scope of the present invention.

Claims

1. A distributed hydrological forecasting method with a confluence method of performing while evolving, characterized in that, Including the following steps: Step S1: Perform grid processing on the study basin, dividing the study basin into N grids; Step S2, analyze the confluence flow relationship of N grids, so as to divide all grids into M levels, which are respectively represented as: the first level R1, the second level R2, …, the Mth level R M ; Among them, there is no confluence relationship between the grids in each same level; each grid in the Mth level RM M is a grid that does not have an upstream grid in the research basin and is a non-confluent grid; each grid from the (M - 1)th level RM M-1 to the 1st level R1 is a confluent grid, having at least one directly connected upstream grid, and, each grid from the (M - 1)th level RM M-1 to the 2nd level R2 has a unique directly connected downstream grid, and the 1st level R1 has only one grid, which is the final inflow grid for the other N - 1 grids; Step S3: Establish a hydrological forecasting model; the hydrological forecasting model includes a groundwater runoff concentration model and a surface runoff concentration model of the combined runoff generation model; Step S4: Determine the training period T; determine the initial values of the model parameters of the hydrological forecasting model; Step S5: Train the hydrological forecasting model to obtain the trained hydrological forecasting model; Step S5.1, according to the calculation order from the Mth level R M to the 1st level R1 in the grid, adopting the side-calculation and side-confluence confluence method, through the hydrological forecasting model, calculate in sequence the subsurface runoff time series and surface runoff time series at each moment t in the training period T for each grid, add the subsurface runoff time series and surface runoff time series at each moment t in the training period T for each grid at each moment t, and obtain the outflow discharge time series at each moment t in the training period T for each grid; Step S5.2: Select a landmark grid in the study basin; by analyzing the outflow discharge time series of the landmark grid at each moment t during the training period T, evaluate whether the currently obtained hydrological forecasting model meets the accuracy requirements; if not, adjust the model parameters of the hydrological forecasting model and return to Step S5.1; if so, obtain the trained hydrological forecasting model; Step S6: Use the trained hydrological forecasting model to forecast the outflow discharge of each grid in the study basin.

2. The distributed hydrological forecasting method with a confluence method of performing evolution and confluence simultaneously according to claim 1, wherein Step S2 is specifically as follows: Step S2.1: In the N grids of the study basin, find the grid with the largest cumulative runoff, and set its level to the first level R1; Step S2.2: Set the initial value of the level variable RANK to 1; Step S2.3: In the study basin, locate the grid of the RANK level, called grid g; traverse the eight-neighborhood grids of grid g that have not been set with levels. If the outflow of the traversed grid converges into grid g, set the level of the traversed grid to RANK + 1; Step S2.4: If the levels of all N grids have been set, the grid level calculation ends; otherwise, let RANK = RANK + 1 and return to Step S2.

3.

3. The distributed hydrological forecasting method with a side-play and side-merging confluence method according to claim 1, characterized in that, The groundwater runoff concentration model of the combined runoff generation model includes the groundwater runoff concentration sub-model of non-concentration grids shown in formula (1) and the groundwater runoff concentration sub-model of concentration grids shown in formula (2): Where: is the subsurface runoff of grid j at time t; KGG is the groundwater runoff recession coefficient, which is a model parameter to be adjusted; is the subsurface runoff of grid j at time t-1, and its initial value is set to 0; is the subsurface runoff generated by grid j itself at time t, which is predicted by the runoff generation model; is the subsurface runoff inflow for grid j at time t, which is determined by Equation (3): Wherein: represents the groundwater runoff generated by the upstream grid i directly connected to grid j at time t itself; upi represents the number of upstream grids i directly connected to grid j.

4. The distributed hydrological forecasting method with a side-playing and side-merging confluence mode according to claim 3, characterized in that, The above surface runoff concentration model includes the surface runoff concentration sub-model for non-concentrating grids shown in formula (4) and the surface runoff concentration sub-model for concentrating grids shown in formula (5): Where: L is the grid length; K is the conversion coefficient between flow rate and flow velocity; x is the flow rate proportion coefficient; is the conversion exponent between flow rate and flow velocity; K, x, are the model parameters to be adjusted; is the surface runoff of grid j at time t; is the surface runoff of grid j at time t - 1, and its initial value is set to 0; Δt is the time interval between two adjacent moments; represent the surface runoff of the upstream grid i directly connected to grid j at time t; upi represents the number of upstream grids i directly connected to grid j; Represents the surface runoff of the upstream grid i directly connected to grid j at time t-1.

5. The distributed hydrological forecasting method with a confluence method of performing evolution and confluence simultaneously according to claim 4, characterized in that, Step S5 is specifically as follows: Step S5.1: Obtain the vegetation data, soil data, and meteorological data at each moment t of each grid j in the study basin during the training period T, and input them into the runoff generation model to obtain the time series of the baseflow generated by each grid j at each moment t during the training period T t = 1, 2,..., T; Step S5.2, determine the initial values of the model parameters of the hydrological forecasting model; the model parameters include: the underground runoff recession coefficient KGG, the conversion coefficient K between flow rate and flow velocity, the flow rate proportion coefficient x, and the conversion exponent between flow rate and flow velocity Step S5.3, perform calculations on each grid j in the Mth level R M : Adopt the subsurface runoff concentration sub-model of non-confluence grid shown in formula (1) to conduct subsurface runoff routing for each grid j in the Mth level R M to obtain the subsurface runoff time series of each grid j in the Mth level R M at each moment t during the training period T where t = 1, 2,..., T; Using the surface runoff concentration sub-model of non-confluence grid shown in formula (4), perform surface runoff calculation for each grid j in the M-th level R M to obtain the surface runoff time series of each grid j in the M-th level R M at each moment t during the training period T t = 1, 2,..., T; Using formula (6), for the $R$ of the $M$-th level M the time series of subsurface runoff at each moment $t$ during the training period $T$ for each grid $j$ and the time series of surface runoff are added at each moment $t$ to obtain the time series of outflow discharge at each moment $t$ during the training period $T$ for each grid $j$ $t = 1, 2, \ldots, T$; Step S5.4, perform calculations on each grid j in the (M - 1)-th level R M-1 : Adopt the subsurface runoff concentration submodel of the current collection grid shown in formula (2) to perform subsurface runoff routing for each grid j in the (M - 1)-th level R M-1 to obtain the subsurface runoff time series of each grid j in the (M - 1)-th level R M-1 at each moment t during the training period T where t = 1, 2,..., T; Using the surface runoff concentration sub-model of the busbar grid shown in formula (5), perform surface runoff calculation on each grid j in the M-1th level R M-1 to obtain the surface runoff time series of each grid j in the M-1th level R M-1 at each moment t during the training period T t = 1, 2,..., T; Using formula (6), for the R at the (M - 1)th level M-1 the time series of baseflow at each grid j in the training period T at each moment t and the time series of surface runoff are added at each moment t to obtain the time series of outflow discharge at each grid j in the training period T at each moment t t = 1, 2,..., T; Step S5.5: In the same way as in Step S5.4, perform calculations on each grid j in the M-2nd level R M-2 , perform calculations on each grid j in the M-3rd level R M-3 , and so on, until the calculations on the grid j in the 1st level R1 are completed; Step S5.6, select landmark grids in the study basin; by analyzing the outflow discharge time series of the landmark grids at each moment t during the training period T, evaluate whether the currently obtained hydrological forecasting model meets the accuracy requirements; if not, adjust the model parameters of the hydrological forecasting model, including the baseflow recession coefficient KGG, the conversion coefficient K between discharge and velocity, the discharge proportion coefficient x, and the conversion exponent between discharge and velocity Then return to step S5.3; if it meets the requirements, obtain the trained hydrological forecasting model.

6. A distributed hydrological forecasting method with a side-play and side-merging confluence method according to claim 5, characterized in that, Step S5.6 is specifically as follows: In the N grids of the study basin, select the grid where the hydrological station is located as the landmark grid, denoted as landmark grid k; Obtain the time series of the outflow discharge observations for each moment t of the landmark grid k during the training period T Obtain the time series of the outflow discharge observations for each moment t of each grid in the study basin during the training period T, and take the average of the outflow discharge observations of all grids at the same moment t to obtain the average time series of all outflow discharge observations Through this round of calculations, the outflow discharge time series of the landmark grid k at each moment t during the training period T is obtained Use formula (7) to obtain the Nash efficiency coefficient NSE; Evaluate whether the currently obtained hydrological forecasting model meets the accuracy requirements through the Nash efficiency coefficient NSE.

Images