Air-ground reservoir water four-dimensional coupling medium and long term runoff chain prediction method

Abstract

The application provides a kind of air-land reservoir water four-dimensional coupling medium and long term runoff chain prediction method, according to the research area hydrology meteorological characteristics clear control section and control reservoir, constructs rainfall-runoff-confluence-evolution chain prediction framework, and is decomposed into meteorological rainfall, land runoff, river confluence and runoff evolution 4 modules;Then, for the time scale to be calculated, select the calculation model suitable for each module, through the mixed downscaling model, land runoff calculation formula, river confluence calculation method and runoff evolution equation considering the influence of reservoir regulation, rainfall, runoff, confluence and evolution calculation are completed respectively, and different control section runoff process is obtained;Finally, the runoff process calculation of different time scales is carried out in turn, and is connected as the medium and long term runoff calculation result of air-land reservoir water four-dimensional coupling of the research area. The method can effectively improve the medium and long term runoff prediction accuracy, and provide important technical support for the optimal allocation and efficient use of water resources in the basin.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of hydrology and water resources prediction, and particularly relates to a four-dimensional coupling of air, land, reservoir and water medium and long-term runoff chain prediction method. BACKGROUND

[0002] Medium and long-term runoff prediction is an important supporting tool in the fields of water resources management, flood control and drought resistance, energy planning, and ecological restoration. With the intensification of global climate change and human activities, the contradiction between water supply and demand is becoming increasingly prominent, and the accuracy and reliability of medium and long-term runoff prediction are particularly important. Accurate medium and long-term runoff prediction can not only provide scientific basis for reservoir regulation, agricultural irrigation, power generation planning, etc., but also provide strong support for regional water resources allocation and flood control and disaster reduction decision-making. Especially in the context of frequent extreme weather events, the uncertainty analysis and risk assessment of medium and long-term runoff prediction have important practical significance.

[0003] However, there is still a lot of room for improvement in the accuracy and reliability of current medium and long-term runoff prediction. Traditional prediction methods based on statistics and hydrological models often show great limitations in the face of complex hydrological processes and nonlinear changes. On the one hand, traditional methods lack sufficient description of the physical mechanism of hydrological processes, making it difficult to capture the long-term impact of climate change and human activities on runoff; on the other hand, with the introduction of machine learning technology, although the prediction accuracy has been improved to some extent, the stability and applicability in complex hydrological environment still need to be verified. In addition, the existing prediction methods have weak fusion ability for multi-source data, making it difficult to fully utilize meteorological, remote sensing and hydrological multi-dimensional data, further restricting the improvement of prediction accuracy.

[0004] In summary, medium and long-term runoff prediction plays an irreplaceable role in water resources management and social and economic development, but its lack of accuracy and method limitations are still the difficulties of current research. By improving the prediction method, optimizing the model structure, improving the efficiency of data utilization, and strengthening the research on the impact of climate change, it is expected to significantly improve the accuracy and reliability of medium and long-term runoff prediction, and provide more solid technical support for sustainable management of water resources. SUMMARY

[0005] The present application aims to overcome the shortcomings of the prior art and provides a four-dimensional coupling of air, land, reservoir and water medium and long-term runoff chain prediction method. Starting from the four links of rainfall, runoff generation, confluence and evolution, the present application provides a four-dimensional coupling of air, land, reservoir and water medium and long-term runoff chain prediction method. According to the characteristics of rainfall prediction, runoff calculation, confluence calculation and runoff evolution process, the present application proposes calculation model selection criteria and result coupling method, and realizes the improvement of medium and long-term runoff prediction accuracy.

[0006] To achieve the above-mentioned purpose, the present application adopts the following technical solutions:

[0007] The application provides a four-dimensional coupling medium and long-term runoff chain prediction method for air-land reservoir water, comprising the following steps:

[0008] S1, selecting a research area, determining a control section, a control reservoir, collecting and arranging long sequence historical data of meteorology and hydrology,

[0009] Meteorological data of the control section is recorded as A f ={a1,a2,a3,...,a i ,...,a N},

[0010] Hydrological data of the control section is recorded as H f ={h1,h2,h3,...,h i ,...,h N};

[0011] Meteorological data of the control reservoir is recorded as A r ={b1,b2,b3,...,b j ,...,b M};

[0012] Hydrological data of the control reservoir is recorded as H r ={l1,l2,l3,...,l j ,...,l M};

[0013] Wherein, A f represents meteorological data of the control section; a i represents meteorological data of the control section i; H f represents hydrological data of the control section; h i represents hydrological data of the control section i; N represents the number of control sections; A r represents meteorological data of the control reservoir; b j represents meteorological data of the control reservoir j; H r represents hydrological data of the control reservoir; l j represents hydrological data of the control reservoir j; and M represents the number of control reservoirs.

[0014] S2, a rainfall runoff evolution chain prediction framework is established, and the series calculation of a meteorological rainfall module R(r1, r2,..., r x ), a land runoff module Q(R, S, Pa, M), a river confluence module M(Q, L, G, X) and a runoff evolution module E(M, L, G, O, Y) is carried out.

[0015] Wherein, R(r1, r2,..., r x ) represents a rainfall calculation equation; r x represents different factors affecting the rainfall value.

[0016] Q(R,S,Pa,M) represents the runoff calculation equation, R is the rainfall value; S is the watershed catchment area; Pa is the previous rainfall; M is the runoff mode, including full storage runoff, excess infiltration runoff;

[0017] M(Q,L,G,X) represents the confluence calculation equation; Q is the runoff value; L is the confluence river length; G is the confluence river ratio; X is the factor affecting confluence;

[0018] E(M,L,G,O,Y) is the evolution calculation equation; M is the confluence value; O is the storage influence of the evolution process reservoir; Y is the factor affecting evolution;

[0019] S3, in the meteorological rainfall module, a mixed downscaling model R' is constructed, and long-term rainfall prediction in the research area is carried out;

[0020] S4, in the land surface runoff module, taking rainfall prediction as input, using hydrological model, runoff calculation of different subareas corresponding to control section in the research area is carried out, assuming S m is the mth subarea of the research area, and the runoff is Q m = Γ(S m ), wherein Γ(S m ) is the runoff calculation formula of the mth subarea;

[0021] S5, in the river confluence module, combined with the equal flow time line method, the time period unit line method and the instantaneous unit line method, the river confluence calculation is carried out, and the flow process of different control sections is obtained;

[0022] S6, in the runoff evolution module, the river evolution calculation is carried out by using the Muskingum method, the synthetic flow method and the hydrodynamics method, and the runoff process of different control sections after river evolution is obtained;

[0023] S7, from upstream to downstream, each node is calculated in the same time scale, until the runoff process of each control section in the time scale is output;

[0024] S8, for the research area, the runoff process calculation in different time scales of ten, month, season and year is completed.

[0025] Further, in the S1, the method for determining the control section and the control reservoir is that, for the flood protection object situation in the research area, the section to be predicted and the reservoir with total reservoir capacity in the basin, the reservoir with adjusting capacity is selected as the control section and the control reservoir, so as to ensure the flood control safety of the basin. 3 or the reservoir with total reservoir capacity ranking occupying the top 5-10% of the total reservoirs in the basin as the control section and the control reservoir.

[0026] Further, in the S1, the meteorological data includes rainfall, air temperature, wind speed, and radiation.

[0027] The hydrological data includes flow, water level, evaporation, reservoir water level, reservoir inflow, and reservoir outflow.

[0028] Further, in the S3, the medium term is calculated in terms of ten days, and the daily rainfall of the area above each control section is output; the long term is calculated in terms of months, seasons, and years, and the monthly rainfall of the area above each control section is output.

[0029] Further, in the S3, the hybrid downscaling model refers to a medium and long term rainfall prediction model constructed by coupling dynamic downscaling, statistical downscaling, and artificial intelligence methods, and specifically includes:

[0030] S301, obtaining output data of different global climate models For the rainfall output data of the kth climate model, K is the type of climate model participating in the calculation, and the measured data of the research area are compared and analyzed to evaluate the prediction accuracy P of different climate models k , and the availability of different climate models is sorted from high to low according to the prediction accuracy, wherein wherein, P k is the prediction accuracy of the kth climate model; is the prediction accuracy of the control section in the kth climate model; is the prediction accuracy of the control reservoir in the kth climate model;

[0031] S302, according to the evaluation results, combined with the characteristics of the research area, select one or more climate models with high availability, set the initial conditions and boundary conditions respectively, and perform simulation operation to obtain a regional climate model with good prediction effect for the research area;

[0032] If one climate model is selected, the calculation result of the regional climate model is equal to the output data of the global climate model

[0033] If multiple climate models are selected, the calculation result of the regional climate model is equal to the arithmetic mean of the multi-model calculation results, and the calculation formula is:

[0034]

[0035] S303, analyze the simulation results, evaluate the error in the global climate model, use deep learning artificial intelligence algorithm to fit the error distribution law, and obtain the error correction calculation formula e(x), wherein e is the error correction data; x is a matrix composed of factors affecting the error distribution law;

[0036] S304, use the regional climate model to study the long-term rainfall prediction in the research area, and based on the error correction calculation formula, revise the prediction results as the final rainfall prediction results, that is, the final rainfall prediction results

[0037] Further, in the S4, the hydrological model includes Xin'anjiang model, rainfall runoff correlation diagram, SAC model, SWAT model, TOPMODEL model and hydrological model suitable for runoff calculation on monthly scale and above;

[0038] The SAC model is the Sacramento Hydrology Model, which is a hydrological model used to simulate the hydrological process of a watershed. It is mainly used to predict rainfall runoff, and converts rainfall into surface runoff and groundwater runoff by simulating the change of soil moisture;

[0039] SWAT model is Soil and Water Assessment Tool, which is a distributed watershed hydrological model based on GIS. It uses spatial information provided by remote sensing and geographic information system to simulate various hydrological physical and chemical processes;

[0040] TOPMODEL model is TOPOgraphy-based MODel of Landscape Evolution, which is a distributed hydrological model based on terrain. It is mainly used to simulate surface runoff, water movement and watershed hydrological process. The above three models are well-known and commonly used models in the field of hydrological prediction;

[0041] For medium-term runoff calculation within ten days, according to the climate division and runoff characteristics of the research area, Xin'anjiang model, rainfall runoff correlation diagram, SAC model, SWAT model and TOPMODEL model are used for calculation, and daily runoff prediction is output;

[0042] For long-term runoff calculation on monthly scale and above, SWAT model, TOPMODEL model and hydrological model suitable for runoff calculation on monthly scale and above are used for calculation, and monthly runoff prediction is output.

[0043] Further, in the S5, when carrying out river confluence calculation, if the rainfall surface distribution in the research area is uneven, the equal flow time method is used;

[0044] If the spatial and temporal distribution of rainfall is uniform and the net rainfall duration is short within the calculation period, the time interval unit line method is adopted.

[0045] If there is a significant runoff change process in the forecast within the calculation period, and the net rainfall intensity is uniformly distributed within the unit time interval, the instantaneous unit line method is adopted.

[0046] Further, in the S6, when the river channel evolution calculation is performed, if the river channel is straight and the flow changes linearly along the river, the Muskingum method is adopted.

[0047] If the river channel evolution length is short and the tributaries are many, the synthetic flow method is adopted.

[0048] If the river channel terrain is complex and the flow changes in various forms along the river, the hydrodynamic method is adopted.

[0049] Further, it comprises at least one processor and a memory connected in communication with the at least one processor, wherein,

[0050] The memory stores instructions executed by the processor, and the instructions are executed by the processor to implement a four-dimensional coupling of air-land-reservoir-water long-term runoff chain prediction method.

[0051] The beneficial effects of the present application are: first, according to the hydrological and meteorological characteristics of the research area, the control section and the control reservoir (group) are determined, the rainfall, runoff, confluence and evolution chain prediction framework is constructed, and the rainfall, runoff, confluence and evolution are divided into four modules of meteorological rainfall, land runoff, river confluence and runoff evolution; then, for the time scale to be calculated, the calculation model suitable for each module is selected, and through the mixed downscaling model, the land runoff calculation formula, the river confluence calculation method and the runoff evolution equation considering the influence of reservoir (group) dispatching, rainfall, runoff, confluence and evolution calculation are completed respectively, and the runoff process of different control sections is obtained; finally, the runoff process calculation of different time scales is carried out in sequence, and is connected as the four-dimensional coupling of air-land-reservoir-water long-term runoff calculation result of the research area. Through this technology, the accuracy of long-term runoff prediction can be effectively improved, which provides important technical support for the optimal allocation and efficient use of water resources in the basin. BRIEF DESCRIPTION OF DRAWINGS

[0052] Figure 1 It is a flow chart of a four-dimensional coupling of air-land-reservoir-water long-term runoff chain prediction method;

[0053] Figure 2 It is a system structure schematic diagram of the Hanjiang River upstream basin in the embodiment, comprising a Hanjiang River upstream water system distribution map and a Hanjiang River upstream control reservoir schematic diagram;

[0054] Figure 3 It is a rainfall zoning schematic diagram of the Hanjiang River upstream basin system in the embodiment;

[0055] Figure 4 Fig. 8 is a schematic diagram of the runoff process of the Yangxian station section in the embodiment;

[0056] Figure 5 Fig. 9 is a schematic diagram of the runoff process of the Ankang station section in the embodiment;

[0057] Figure 6 Fig. 10 is a schematic diagram of the reservoir inflow and outflow process of the Shiquan reservoir section in the embodiment;

[0058] Figure 7 Fig. 11 is a schematic diagram of the reservoir inflow and outflow process of the Ankang reservoir section in the embodiment;

[0059] Figure 8 Fig. 12 is a schematic diagram of the reservoir inflow and outflow process of the Danjiangkou reservoir section in the embodiment. DETAILED DESCRIPTION

[0060] In order to make the objects, technical solutions and advantages of the present application clearer, further detailed description will be made to the present application with reference to the drawings. It should be understood that the specific embodiments described herein are only used to explain the present application, and are not used to limit the present application.

[0061] Please refer to Figure 1 A four-dimensional coupling medium and long-term runoff chain prediction method of air-land-reservoir-water, comprising:

[0062] S1, selecting a research area, determining control sections, control reservoirs, collecting and arranging long sequence historical data of meteorology and hydrology,

[0063] The meteorological data of the control section is denoted as A f ={a1, a2, a3,..., a i ,...,a N},

[0064] The hydrological data of the control section is denoted as H f ={h1, h2, h3,..., h i ,...,h N};

[0065] The meteorological data of the control reservoir is denoted as A r ={b1, b2, b3,..., b j ,...,b M};

[0066] The hydrological data of the control reservoir is denoted as H r ={l1, l2, l3,..., l j ,...,l M};

[0067] Wherein, A f represents the meteorological data of the control section; a iMeteorological data of control section i; H f Hydrological data of control section; h i Hydrological data of control section i; N represents the number of control sections; A r Meteorological data of control reservoir; b j Meteorological data of control reservoir j; H r Hydrological data of control reservoir; l j Hydrological data of control reservoir j; M represents the number of control reservoirs;

[0068] S2, establish rainfall runoff concentration evolution chain prediction framework, carry out meteorological rainfall module R(r1, r2,..., r x ), land surface runoff module Q(R, S, Pa, M), river channel concentration module M(Q, L, G, X) and runoff evolution module E(M, L, G, O, Y) in series calculation;

[0069] Where, R(r1, r2,..., r x ) represents the rainfall calculation equation; r x Indicates different factors affecting rainfall value;

[0070] Q(R, S, Pa, M) represents the runoff calculation equation, R is the rainfall value; S is the watershed catchment area; Pa is the previous influence rainfall; M is the runoff mode, including full storage runoff, excess infiltration runoff;

[0071] M(Q, L, G, X) represents the concentration calculation equation; Q is the runoff value; L is the length of the concentration river channel; G is the concentration river channel ratio; X is the factor affecting concentration;

[0072] E(M, L, G, O, Y) is the evolution calculation equation; M is the concentration value; O is the storage influence of the evolution process reservoir; Y is the factor affecting evolution;

[0073] S3, in the meteorological rainfall module, build a mixed downscaling model R', carry out long-term rainfall prediction in the research area;

[0074] S4, in the land surface runoff module, take rainfall prediction as input, use hydrological model, carry out runoff calculation of different partitions corresponding to control section in the research area, assuming S m is the mth partition of the research area, and the runoff is Q m = Γ(S m ), wherein Γ(S m ) is the runoff calculation formula of the mth partition;

[0075] S5, in the river channel concentration module, combined with the isochrone method, time interval unit line method and instantaneous unit line method, carry out river channel concentration calculation, and obtain the flow process of different control sections;

[0076] S6. In the runoff evolution module, the Muskingum, composite flow, and hydrodynamic methods are used to calculate the river evolution and obtain the flow process at different control sections after the river evolution.

[0077] S7. From upstream to downstream, each node is calculated and evolved sequentially at the same time scale until the runoff process at each control section at the time scale is output.

[0078] S8. For the study area, complete the runoff process calculations at different time scales of ten days, month, season, and year.

[0079] In S1, the method for determining control sections and key reservoirs is based on the flood protection targets within the study area, the sections to be predicted, and reservoirs with large total storage capacity and regulation capabilities within the basin. This is done with the premise of ensuring flood control safety in the basin, selecting sections where important flood protection targets are located, sections of interest, and reservoirs with flood control capacity greater than or equal to 100 million m³. 3 Alternatively, reservoirs whose total storage capacity ranks among the top 5-10% of all reservoirs in the basin can be designated as control sections and control reservoirs.

[0080] In S1, the meteorological data includes: rainfall, temperature, wind speed, and radiation;

[0081] Hydrological data include: flow rate, water level, evaporation, reservoir water level, reservoir inflow, and reservoir outflow.

[0082] In S3, the medium-term is calculated on a ten-day period, outputting the daily rainfall of the area above each control section; the long-term is calculated on a monthly, quarterly, or annual basis, outputting the monthly rainfall of the area above each control section.

[0083] In S3, the hybrid downscaling model refers to a medium- to long-term rainfall forecasting model constructed by coupling dynamic downscaling, statistical downscaling, and artificial intelligence methods, specifically:

[0084] S301. Obtain output data from different global climate models. for The rainfall output data of the k-th climate model, where K represents the type of climate model used in the calculation, and is compared with the measured data of the study area. Comparative analysis was conducted to evaluate the prediction accuracy (P) of different climate models. k The availability of different climate models was ranked from highest to lowest based on their prediction accuracy. Among them, P k Let be the prediction accuracy of the k-th climate model; The prediction accuracy of the control section in the k-th climate model; The prediction accuracy of the controlling reservoir in the k-th climate model;

[0085] S302, according to the evaluation results, combined with the characteristics of the study area, select one or more climate models with high availability, set the initial conditions and boundary conditions respectively, and perform simulation operation to obtain a regional climate model with good prediction effect for the study area;

[0086] If a climate model is selected, the regional climate model calculation result is equal to the output data of the global climate model

[0087] If multiple climate models are selected, the regional climate model calculation result is equal to the arithmetic mean of the multiple model calculation results, and the calculation formula is:

[0088]

[0089] S303, analyze the simulation results, evaluate the error in the global climate model, use deep learning artificial intelligence algorithm to fit the error distribution law, and obtain the error correction calculation formula e(x), wherein e is the error correction data; x is a matrix composed of factors affecting the error distribution law;

[0090] S304, use the regional climate model to predict the long-term rainfall in the study area, and based on the error correction calculation formula, revise the prediction results as the final rainfall prediction results, that is, the final rainfall prediction results

[0091] In the S4, the hydrological model includes Xin'anjiang model, rainfall runoff correlation diagram, SAC model, SWAT model, TOPMODEL model and hydrological model applicable to monthly scale runoff calculation;

[0092] For medium-term runoff calculation within a ten-day scale, Xin'anjiang model, rainfall runoff correlation diagram, SAC model, SWAT model, TOPMODEL model are used according to the climate division and runoff characteristics of the study area, and daily runoff prediction is output;

[0093] For long-term runoff calculation of monthly scale and above, SWAT model, TOPMODEL model and hydrological model applicable to monthly scale runoff calculation are used for calculation, and monthly runoff prediction is output.

[0094] In the S5, when carrying out river confluence calculation, if the rainfall surface distribution in the study area is uneven, the equal flow time line method is used;

[0095] If the rainfall spatial and temporal distribution is uniform and the net rain duration is short within the calculation period, the time period unit line method is used;

[0096] If there is a significant runoff change process predicted in the calculation period, and the net rainfall intensity is uniformly distributed in the unit period, the instantaneous unit line method is adopted.

[0097] In S6, when the river channel evolves, if the river channel is straight and the flow changes linearly along the river, the Muskingum method is adopted.

[0098] If the river channel evolution length is short and the tributary is merged, the synthetic flow method is adopted.

[0099] If the river channel topography is complex and the flow changes in various forms, the hydrodynamic method is adopted.

[0100] Comprising at least one processor, and the memory connected with at least one processor in communication, wherein,

[0101] The memory stores instructions executed by the processor, and the instructions are executed by the processor to implement a four-dimensional coupling long-term runoff chain prediction method of air-land reservoir.

[0102] Embodiment

[0103] Now taking the Hanjiang River upstream watershed system with the Danjiangkou Reservoir as the core as an example, the medium and long-term decadal scale runoff prediction calculation is carried out to verify the feasibility and effectiveness of the method.

[0104] Figure 2 、 Figure 3 The Hanjiang River upstream watershed system and rainfall prediction zoning are drawn respectively, Figure 2 and Figure 3 It can be seen that the medium and long-term runoff prediction of the Hanjiang River upstream watershed system can be divided into three rainfall prediction modules of the subzone, a multi-zone runoff module composed of different trunk and branch streams, and a river evolution module with Shiquan Reservoir, Ankang Reservoir, Pankou Reservoir, Huanglongtan Reservoir and Danjiangkou Reservoir as control reservoir nodes. When there is a rainfall process in the watershed, the method is used to calculate the surface rainfall of the three subzones by using the hybrid downscaling model, and then the runoff, confluence and evolution calculation are carried out respectively. For the five reservoirs, the scheduling rules are constructed and called to realize continuous calculation of the river system, and the runoff process of the Danjiangkou Reservoir section is output as the decadal scale (i.e. 10 days in the future) runoff prediction result of the Hanjiang River upstream.

[0105] Table 1 shows the runoff results of the main control section Yangxian, Ankang station and the control reservoir section Danjiangkou Reservoir. The specific runoff process of each section is shown in Figures 4 to 7 .

[0106] Table 1 Unit 1-Unit 3 specific scheduling process information table

[0107]

[0108]

[0109] By Figures 4 to 7 It can be known that the method adopts the four-dimensional coupling medium and long term runoff chain prediction method of air-land-warehouse-water, can quickly realize the calculation of rainfall, runoff, confluence and evolution of the research area, obtains the medium and long term runoff process of the required section, the prediction result has good fitting effect with the actual process, the prediction precision is improved obviously, and the feasibility and effectiveness of the method are proved.

[0110] According to the above analysis, the method has strong practicability, and can effectively solve the problem of insufficient precision in medium and long term runoff prediction.

[0111] In summary, the method has practicability and operability, can quickly realize the medium and long term runoff prediction of the research area, and obtain the runoff prediction result of the important section, so as to provide more scientific and efficient technical support for the basin hydrology and water resources prediction and reservoir scheduling.

[0112] The above-described embodiments only express the implementation of the present application, and the description is more specific and detailed, but it cannot be understood as the limitation of the scope of the present application. It should be pointed out that for ordinary skilled persons in the art, without departing from the concept of the present application, a number of modifications and improvements can be made, which belong to the protection scope of the present application. Therefore, the protection scope of the present application patent should be the appended claims.

Claims

1. A method for predicting a long-term runoff chain in a four-dimensional coupling of air, land, reservoir and water, characterized in that, Comprise: S1, select the study area, determine the control section, control reservoir, collect and collate meteorological, hydrological long sequence historical data, Control section meteorological data is denoted as A f = {a1, a2, a3,..., an}, where n is the number of control sections. i = {a1, a2, a3,..., an}, where n is the number of control sections. N = {a1, a2, a3,..., an} Control section hydrograph data is denoted as H f = {h1, h2, h3,..., h i ,...,h N} Controlled reservoir weather data is denoted as A r = {b1, b2, b3,..., b j ,...,b M} Hydrological data of controlled reservoirs r = {l1, l2, l3,..., ln} ; where n is the number of reservoirs j = {l1, l2, l3,..., ln} ; where n is the number of reservoirs M} ; where A f represents meteorological data of the control section; a i represents meteorological data of the control section i; H f represents hydrological data of the control section; h i represents hydrological data of the control section i; N represents the number of control sections; A r represents meteorological data of the control reservoir; b j represents meteorological data of the control reservoir j; H r represents hydrological data of the control reservoir; l j represents hydrological data of the control reservoir j; M represents the number of control reservoirs; S2, establishing a rainfall runoff concentration evolution chain prediction framework, performing series calculation of meteorological rainfall module R(r1, r2,..., r x ), land surface runoff module Q(R, S, Pa, M), river concentration module M(Q, L, G, X) and runoff evolution module E(M, L, G, O, Y). where R(r1, r2,..., r x ) represents the rainfall calculation equation; r x represents different factors affecting the rainfall value; Q(R, S, Pa, M) represents the runoff calculation equation, R is the rainfall value; S is the watershed catchment area; Pa is the previous influence rainfall; M is the runoff mode, including storage runoff, excess infiltration runoff; M(Q, L, G, X) represents the concentration calculation equation; Q is the runoff value; L is the concentration river length; G is the concentration river ratio; X is the factor affecting the concentration; E(M, L, G, O, Y) is the evolution calculation equation; M is the concentration value; O is the storage influence of the evolution process reservoir; Y is the factor affecting the evolution; S3, in the meteorological rainfall module, build a mixed downscaling model R', carry out long-term rainfall prediction in the study area; S4, in the land surface runoff module, taking the rainfall forecast as the input, using a hydrological model, carrying out runoff calculation of the control section corresponding to each subzone of the research area, assuming S m For the mth subzone of the research area, the runoff is Q m = Γ(S m ), wherein Γ(S m ) is a runoff calculation formula of the mth subzone. S5, in the river concentration module, combined with the isochrone method, the time interval unit line method, the instantaneous unit line method, carry out river concentration calculation, and obtain the flow process of different control sections; S6, in the runoff evolution module, the method of Muskingum, synthetic flow, hydrodynamics is used for river evolution calculation, and the flow process of different control sections after river evolution is obtained; S7, from upstream to downstream, each node is calculated in the same time scale, until the runoff process of each control section under the time scale is output; S8, for the study area, the runoff process calculation of different time scales of ten, month, season and year is completed.

2. The four-dimensional coupled air-land reservoir water medium and long-term runoff chain prediction method according to claim 1, characterized in that: In the S1, the method for determining the control section and the controlling reservoir is as follows: on the premise of ensuring the flood control safety of the basin, the section where the important flood control object is located, the concerned section and the reservoir with the flood control capacity greater than or equal to 100 million m 3 or the reservoir with the total reservoir capacity ranking in the top 5-10% of the total reservoir capacity of the basin are selected as the control section and the controlling reservoir.

3. The four-dimensional coupled air-land reservoir water long-term runoff chain prediction method according to claim 2, characterized in that: In S1, the meteorological data includes: rainfall, temperature, wind speed, radiation; The hydrological data includes: flow, water level, evaporation, reservoir water level, reservoir inflow, reservoir outflow.

4. The four-dimensional coupled air-land reservoir water medium and long-term runoff chain prediction method according to claim 3, characterized in that: In S3, the medium term is calculated by ten, and the daily rainfall of the area above each control section is output; the long term is calculated by month, season and year, and the monthly rainfall of the area above each control section is output.

5. The air-land reservoir water four-dimensional coupled medium and long-term runoff chain prediction method according to claim 4, characterized in that: In S3, the mixed downscaling model refers to the coupled dynamic downscaling, statistical downscaling and artificial intelligence method, which is used to build a medium and long term rainfall prediction model, specifically: S301. Obtain output data from different global climate models. for The rainfall output data of the k-th climate model, where K represents the type of climate model used in the calculation, and is compared with the measured data of the study area. Comparative analysis was conducted to evaluate the prediction accuracy (P) of different climate models. k The availability of different climate models was ranked from highest to lowest based on their prediction accuracy. Among them, P k Let be the prediction accuracy of the k-th climate model; The prediction accuracy of the control section in the k-th climate model; The prediction accuracy of the controlling reservoir in the k-th climate model; S302, according to the evaluation results, combined with the characteristics of the study area, select one or more climate models with high availability, set the initial conditions and boundary conditions respectively, and carry out simulation operation, obtain the regional climate model with good prediction effect for the study area; If a climate model is selected, the regional climate model calculation results are equal to the output data of the global climate model If multiple climate models are selected, the regional climate model calculation result is equal to the arithmetic mean of the multi-model calculation results, and the calculation formula is: S303, analyze the simulation results, evaluate the error in the global climate model, use the deep learning artificial intelligence algorithm to fit the error distribution law, obtain the error correction calculation formula e(x), wherein e is the error correction data; x is the matrix composed of factors affecting the error distribution law; S304. Utilize regional climate models to predict medium- to long-term rainfall in the study area, and revise the prediction results based on error correction formulas, outputting the final rainfall prediction result.

6. The air-ground-ocean water four-dimensional coupled medium and long-term runoff chain prediction method according to claim 5, characterized in that: In S4, the hydrological model includes Xin'anjiang model, rainfall runoff correlation diagram, SAC model, SWAT model, TOPMODEL model and hydrological model suitable for month scale and above runoff calculation; For the calculation of medium-term runoff within the scale of ten days, the Xin'anjiang model, the rainfall-runoff correlation diagram, the SAC model, the SWAT model, and the TOPMODEL are used according to the climate division and the runoff characteristics of the research area to output daily runoff prediction; For the calculation of long-term runoff at the scale of a month or above, the SWAT model, the TOPMODEL, and the hydrological model suitable for runoff calculation at the scale of a month or above are used to output monthly runoff prediction.

7. The four-dimensional coupled air-land reservoir water medium long-term runoff chain prediction method according to claim 6, characterized in that, In the S5, when the river confluence calculation is carried out, if the rainfall surface distribution in the research area is uneven, the equiflux isochrone method is used; If the rainfall space-time distribution is uniform and the net rainfall duration is short in the calculation period, the time interval unit line method is used; If there is a significant runoff change process in the prediction in the calculation period, and the net rainfall intensity is uniformly distributed in the unit time interval, the instantaneous unit line method is used.

8. The four-dimensional coupled air-land reservoir water medium long-term runoff chain prediction method according to claim 7, characterized in that: In the S6, when the river evolution calculation is carried out, if the river is straight and the flow changes linearly along the river, the Muskingum method is used; If the river evolution length is short and many tributaries flow into the river, the synthetic flow method is used; If the river terrain is complex and the flow changes in various forms along the river, the hydrodynamic method is used.

9. The air-land reservoir water four-dimensional coupled medium and long-term runoff chain prediction method according to claim 8, characterized in that: The device comprises at least one processor and a memory connected to the at least one processor in communication, wherein The memory stores instructions executed by the processor, and the instructions are executed by the processor to implement the four-dimensional coupled air-land reservoir water long-term runoff chain prediction method of any one of claims 1 to 8.

Images