CN115130396A - A Distributed Hydrological Modeling Method for Channel-type Reservoir Areas - Google Patents

Abstract

The invention provides a distributed hydrological model modeling method for a river channel type reservoir area, and belongs to the technical field of distributed hydrological modeling. Firstly, obtaining basic data required for building a reservoir area distributed hydrological model, including topographic and geomorphic data and hydrological meteorological data, seeking a proper grid scale and providing a basis for modeling the distributed hydrological model. And secondly, dividing the sub-watershed and the sloping surface watershed, and calculating the convergence. And finally, evaluating the forecasting performance of each flood according to the traditional evaluation indexes, and carrying out parameter calibration. According to the method, the reservoir area is divided into the sub-watersheds and the slope watershed through the establishment of the distributed hydrological model, the slope watershed is divided into the more refined slope hydrological response units, the spatial variation characteristics of the underlying surface of the slope unit can be fully reflected, and the inflow condition of the slope watershed can be simulated more accurately.

Description

A Distributed Hydrological Modeling Method for Channel-type Reservoir Areas

technical field

The invention belongs to the technical field of distributed hydrological modeling, and relates to a distributed hydrological model modeling method for a channel-type reservoir area.

Background technique

In the channel-type reservoir area, the banks on both sides of the main channel are mostly mountains and mountains. In addition to the confluence of the tributaries, the runoff will be directly scattered into the main channel from the slopes on both banks. Although the water-collecting area of the slope area is small, the spatial span is large and the topographic and geomorphological characteristics The variability is strong, which makes the spatial distribution of water flowing into the main channel very uneven.

At present, the existing lumped hydrological model does not fully consider the spatial heterogeneity of the slope area, and divides the slope area into a large watershed for processing, which can only simulate the runoff at the outlet of the watershed, and cannot obtain the slope surface. It is difficult to accurately describe the temporal and spatial variability of rainfall and runoff in the reservoir area because of the flow processes along the main channel in the basin. Although the distributed watershed hydrological model can simultaneously consider the spatial distribution of rainfall and its unevenness, as well as the influence of the spatial variation of the underlying surface on the runoff and confluence of the watershed, the modelling method for the channel-type reservoir area is not refined enough to meet the requirements of the river channel. Accuracy requirements of flood forecasting for important cross-sections along the reservoir area.

Therefore, according to the topographic and geomorphological characteristics of the channel-type reservoir area, it is urgent to invent a distributed hydrological model that can describe the runoff and runoff process of the slope basin along the main channel, so as to fully reflect the spatial variation characteristics of the underlying surface of the slope unit and accurately Simulate the inflow of a slope watershed.

SUMMARY OF THE INVENTION

Aiming at the problems existing in the prior art, the present invention provides a distributed hydrological model modeling method for a channel-type reservoir area.

The technical scheme adopted in the present invention is:

A distributed hydrological modeling method for a channel-type reservoir area, the key is to divide the reservoir area into sub-basins and slope watersheds. A more refined slope hydrological response unit. Specifically include the following steps:

The first step is the division of watersheds for different landform types

For channel-type reservoirs, precipitation-runoff-confluence can be divided into two situations according to the contact form with the main river channel (reservoir area), and the two situations are different in the way of confluence. The runoff and confluence of the sub-basins flow into the reservoir area in the form of point sources, and the runoff from the slopes along the main river channel flows into the reservoir area in a distributed manner along the course.

1.1 Division of sub-watersheds and slope watersheds

The invention divides the reservoir area into two types: sub-watershed and slope watershed. There is a stable water system in the sub-basin, and it is hydraulically connected with the main channel in the form of contact points. According to the modeling method for closed watersheds, it is not necessary to obtain the precipitation-runoff process of various parts of the river in the sub-basin, and only the output of this The runoff at the outlet of the sub-catchment is sufficient. The slope watershed is in direct contact with the main river channel in the form of a surface, and there is no stable and complete water system in the slope watershed, and the runoff will be scattered into the main river channel. The catchment area of the slope watershed is relatively small, but the slope watershed extends over a large area on both sides of the main channel, and the topographic and geomorphological features are highly variable. It is necessary to divide the slope watershed into smaller-scale slope hydrological response units, and fully consider The spatial heterogeneity of the underlying surface of the temporal and spatial variability of rainfall in the slope watershed satisfies the fine modeling on the spatial scale.

1.2 Determining the threshold value of the hydrological response unit for the slope watershed

Due to the different interaction modes and confluence characteristics between the sub-basin and the slope basin and the main channel, the requirements for the level of refinement of the modeling are also different. In order to make the slope watershed more refined, it is necessary to divide the slope area into finer slope hydrological response units by setting soil types, land use types, and slope thresholds.

In order to determine the refined hydrological response unit of the slope surface, the present invention refines the watershed unit by setting a combination of multiple and different thresholds, that is, the area thresholds of different land use types, the area thresholds of soil types, and the slope grades are respectively set. After the threshold, they are arranged and combined, and the impact of different threshold combinations on the division of hydrological response units is tested and analyzed, and finally a threshold combination that can take into account the simulation accuracy and calculation efficiency is obtained. The surface watershed is divided into the number of slope hydrological response units.

The second step, grid size selection

The distributed hydrological model can take into account the spatial distribution differences of hydrological elements in the watershed, and can truly simulate the spatial variation of the runoff and runoff process in the watershed. To reflect this advantage, the construction of the distributed hydrological model needs to be based on the watershed DEM data. It is divided into several grids that can reflect the spatially heterogeneous characteristics of the watershed. At the same time, because the size of the grid size will have a great impact on the simulation accuracy of the runoff and the model calculation efficiency, the invention proposes a grid division method framework that can satisfy the simulation accuracy and calculation efficiency at the same time.

2.1 Grid construction of different scales

In the construction of a hydrological model, the selection of different grid scales often has a greater impact on the simulation results of the hydrological model. Based on the national digital elevation data with a precision of 30m×30m, the invention resamples the original DEM data to different resolutions, thereby establishing DEMs with different grid scales.

The present invention adopts the three most commonly used resampling methods at present to resample the original DEM respectively, and the three methods are the nearest neighbor method, the bilinear interpolation method and the cubic convolution method. Based on the DEM data obtained after resampling, calculate its root mean square error (RMSE), and finally select the resampling method according to the lowest value of RMSE. The RMSE calculation formula is:

In the formula, Z _k is the grid elevation of the DEM after resampling, z _k is the elevation of the corresponding position of the reference data, and n is the number of grids after resampling.

2.2 DEM analysis and processing

The resampled DEM data is further analyzed and processed to obtain watershed attribute data, and a hydrological model is constructed based on this. For the acquired water system in the basin, through the steps of data correction, filling and leveling, determining the direction of the water flow, calculating the accumulation of the confluence, setting the threshold of the catchment area, and extracting the river network in the river basin, a match with the actual river network that conforms to the surface water flow mechanism is obtained. High degree of continuous river network data. Specific steps are as follows:

①DEM correction: The Agree method is used to detect and improve the accuracy of the surface elevation data of the DEM data. By fine-tuning the surface elevation of the DEM, it maintains continuity and consistency with the river vector data, so as to complete the correction of the DEM data.

②DEM depression and flat processing: The large amount of depression information existing in DEM data will lead to discontinuous water flow during water flow analysis. In order to ensure the continuity of the flow direction analysis, the two depression types are filled with different processing methods. For an independent depression in which the elevation data of eight adjacent points around a point are all greater than the data of the point, when at least one of the eight adjacent points is the outlet of the catchment area, the minimum value of the eight adjacent points around the point is assigned. At this point; when there is no exit of the catchment area around, find the point with the minimum elevation on the boundary line of the area, and assign its value to the filling point less than this value and eight adjacent points, thus completing the filling work of the independent depression. For the composite depression area with multiple valley bottom points, firstly starting from each valley bottom, determine the position of the depression edge point and the exit point through the reverse water flow, and then use the exit point elevation to replace the elevation of the point whose depression edge elevation is less than this value, and then use it as The foundation also assigns this elevation data to the point where the exit point elevation of the adjacent depression area is lower than this value, and thus the filling of the composite depression is completed.

For a grid point with the same elevation value as the eight surrounding grids, the flat land needs to be processed to ensure that water flow can be directed out of the flat land area. Search for a point whose elevation is greater than that of adjacent points on the boundary of the flat area, mark it as a safe point, and mark the remaining points as points to be processed, and then increase the elevation of the points to be processed by a small amount until all the points to be processed can be completely removed. Mark it as a safe point, so far the processing of the flat land is completed, and finally the DEM model can generate a continuous river system;

③Determination of water flow direction: In order to clarify the water flow direction in each cell, the present invention first compares the slope between the processed grid cell and the adjacent 8 grid cells, and connects the center of the processed grid cell with the adjacent grid cells. The center of the grid unit with the largest gradient among the 8 grid units, the connection direction is defined as the water flow direction of the processed grid unit, that is, the D8 algorithm is used to determine the flow rate of the water flow;

④ Determination of catchment area and water system: The cumulative amount of water flow on each grid point is the total number of all grids flowing into this grid, that is, the cumulative confluence matrix. Based on the water flow direction of the grid, the confluence accumulation matrix of each grid can be calculated, and the upstream confluence area of each unit grid can be obtained by multiplying the confluence accumulation matrix by the grid area.

However, not all grids with an upstream catchment area can form a river network. Only when the catchment area is greater than a certain threshold will the grid be counted as a grid within the river network. After the direction is connected, a river network can be formed. In order to determine the threshold value of the catchment area, the present invention ascertains the change of the relation coefficient by establishing the relational expression between the slope of the river network and the catchment area. Through trial calculation, when the threshold value of the catchment area approaches a certain value, the relationship coefficient tends to be stable, and the threshold value of the catchment area corresponding to this stable relationship coefficient is regarded as the most suitable threshold value of the catchment area.

The third step is the calculation of runoff and runoff in sub-watersheds and slope watersheds

3.1 Calculation of runoff in sub-watersheds and slope watersheds

The runoff process of the sub-watershed and the slope watershed is consistent. Based on the determination of the grid scale, the present invention adopts the two runoff mechanisms of full storage and over-osmosis and the influence of temperature on the runoff simulation results for both the sub-watershed and the slope watershed. The distributed hydrological model of the runoff model.

The runoff calculations for subwatersheds and slope watersheds are as follows:

The infiltration capacity of grid cells varies with space and can be expressed by the following formula:

f=f _m [1-(1-C) ^1/B ] (2)

where f is the infiltration capacity, f _m is the maximum infiltration capacity, C is the area ratio of the infiltration capacity less than or equal to f, and B is the infiltration capacity shape parameter.

According to the water balance formula, in a given period of time P can be deduced:

P=R ₁ (y)+R ₂ (y)+ΔW(y) (3)

y=R ₁ (y)+ΔW(y) (4)

Among them, p represents the rainfall during the period; y represents the vertical depth; R ₁ (y) represents the full storage runoff; R ₂ (y) represents the hyperosmotic runoff; ΔW(y) represents the change of soil moisture content;

In equations (3) and (4), the full storage runoff (R ₁ ) and soil moisture content change (ΔW) can be expressed as:

Among them, im represents the maximum water storage capacity; _{i 0} represents the water storage capacity at a certain point; b represents the water storage shape coefficient;

From the above formula, it can be inferred that W _p , the expression of R ₂ is:

Among them, B is the shape coefficient of infiltration capacity; Δt is the calculation time step;

The base flow calculation adopts the ARNO model, and the calculation formula is:

为下层土壤的初始含水量，

为下层土壤的最大含水量，W_s为下层土壤含水比例，R_b为基流。where D _m is the maximum base flow, D _s is the ratio of the current base flow to D _m ,

is the initial water content of the subsoil,

is the maximum water content of the lower soil, W _s is the water ratio of the lower soil, and R _b is the base flow.

3.2 Calculation of Confluence in Sub-basin and Slope Watershed

The present invention simulates the spatial evolution of precipitation and runoff in a channel-type reservoir, adopts the confluence method of "first integration and then evolution" to calculate the confluence process of the sub-watershed and the slope watershed, and finely simulates the evolution process of runoff between grids.

For sub-watersheds, the grid is divided into two types: slope grid (excluding tributary channels) and channel grid (including tributary channels) according to whether the grid contains tributary channels. In the slope grid, the runoff first merges into the nearby river grid according to the topological relationship of the grid. This process is calculated by the unit line method based on probability distribution, and then the evolution process of the runoff in the channel grid uses kinematic waves. Equations or impulse response functions are used to describe, and the calculation is performed step by step along the channel grid unit to the sub-basin outlet section until it merges into the main channel.

In the sub-watershed confluence, the topological relationship between grids is determined according to the D8 flow direction algorithm introduced above, and the movement of runoff in each grid is calculated according to the slope confluence method.

In the slope watershed, the runoff is dispersed and directly merges into the main channel, which is in direct contact with the main channel in the form of planes. There is no downward calculation process along the river channel. The process of the slope grid entering the main channel adopts a probability distribution The unit line method is described.

①Slope confluence calculation

In the calculation of slope confluence, the present invention deduces the unit line from the probability distribution.

The probability density function of the two-parameter lognormal distribution is:

In the formula, x>0, -∞<μ<∞, σ>0.

t _p =exp(μ-σ ² ) (11)

Bring the above formula into the probability density function to solve, there are:

Multiply t _p and f(t _p ) formula to get:

Taking the natural logarithm of both sides of the above equation, we get:

ln(t _p )=μ-σ ² (14)

but

μ=σ ² +ln(t _p ) (15)

Given t _p and q _p , μ and σ can be solved by the formula. In addition to the two-parameter lognormal distribution, the unit line can also be derived from probability distribution functions such as the two-parameter Pareto distribution, the two-parameter Weibull distribution, and the two-parameter Frechet distribution.

②Calculation of river confluence

Aiming at the channel-type reservoir, the present invention adopts the existing mainstream efficient calculation method of confluence to calculate the confluence, that is, two river confluence schemes of kinematic wave (KWT) and impulse response function (IRF) are used to calculate the river basin confluence, wherein the KWT method uses For rivers with large river bottom gradients, the IRF method is used for general rivers. The two methods are calculated as follows:

a. Kinematic wave method (KWT)

The KWT method can calculate the wave speed or flow from the hydrological response unit to an independent river reach in each time period. It approximately assumes that the river channel is rectangular, the hydraulic width, the wave speed C is the channel width ω, the Manning coefficient n, and the river bottom gradient S ₀ , and a function of flow q. Among them, the river bottom gradient can be calculated from the river network data, and the flow q can be obtained from the above-mentioned slope confluence. The formula for determining the channel width is as follows:

The calculation formula of wave speed in the river reach is as follows:

If the calculated estimated outflow time is before the end of the period, the wave flows into the downstream reaches, otherwise it is considered to stay in the current period.

b. Impulse Response Function (IRF)

The IRF method is used to perform confluence calculations on the runoff output of a gridded land surface model, using either a gridded river network or a vector river network.

The mathematical development of the IRF method is based on the one-dimensional diffusion wave equation derived from the one-dimensional Saint-Venant equation

In the formula, q is the flow rate of the water-passing section, x is the distance along the channel, C is the flow velocity, and D is the diffusion coefficient.

The formula can be solved by convolution integration

in

In the formula, U(t-s) is the runoff depth generated at the time of t-s.

When using the IRF method, it is only necessary to perform unit line integration on each upstream reach, and then accumulate the confluence routing flows of all upstream reaches in the outlet reach to obtain the flow process. Compared with the KWT method, this method does not require Paying attention to the sequence of the reach is more convenient to calculate.

The fourth step, distributed hydrological model parameter calibration

For the distributed hydrological model, in order to avoid falling into a local convergence point, a multi-objective optimization algorithm is used for parameter calibration. The core of this algorithm is to coordinate the relationship between the objective functions and find out the Pareto optimal solution set that makes each objective function as large or small as possible. The present invention adopts NSGA-II algorithm for calibration.

The flood forecast in the reservoir area focuses on the flood peak, flood volume, flood hydrograph, and flood spatiotemporal distribution. Therefore, when optimizing the parameters of the runoff model of the distributed hydrological model, two objective functions are set in the NSGA-II algorithm. , respectively, the mean value of the absolute value of the relative error of the flood runoff depth and the peak discharge is the smallest as the objective function.

In the formula, n is the number of flood fields, R _f (i) represents the simulated runoff depth of the i-th flood, and R _m (i) represents the measured runoff depth of the i-th flood. Q _f (i) represents the simulated flood peak of the i-th flood, and Q _m (i) represents the measured flood peak of the i-th flood.

When optimizing the parameters of the confluence model, the maximum mean value of the flood certainty coefficient (DC) is taken as the objective function:

总流量过程的平均实测流量。In the formula, Q _if and Q _im are the simulated flow and the measured flow of the i-th flood in the total flow process, respectively,

The average measured flow of the total flow process.

So far, the parameters of the model are obtained, and the construction of the distributed hydrological model of the channel-type reservoir area is completed.

The effects and benefits of the present invention are as follows: the present invention divides the reservoir area into sub-watersheds and slope watersheds through the establishment of a distributed hydrological model, and divides the slope watershed into finer slope hydrological response units, which can fully reflect the slope The spatial variation characteristics of the underlying surface of the surface unit can be used to simulate the inflow of the slope watershed more accurately.

Description of drawings

FIG. 1 is a DEM elevation map of the Three Gorges Reservoir area with a resolution of 30 m according to an embodiment of the present invention.

FIG. 2 is a diagram showing the results of grid division of the Three Gorges Reservoir area according to an embodiment of the present invention.

FIG. 3 is an output node diagram of a distributed hydrological model in the Three Gorges Reservoir area according to an embodiment of the present invention.

Fig. 4 is a schematic diagram of basin division of the present invention.

FIG. 5 is a schematic diagram of the confluence process of the sub-watershed and the slope watershed according to the present invention.

FIG. 6 is a technical roadmap of the present invention for modeling a distributed hydrological model in a channel-type reservoir area.

Detailed ways

Based on distributed hydrological modeling, the invention proposes a distributed hydrological modeling method for a channel-type reservoir area.

The present invention will be further described below through examples.

The Three Gorges Reservoir is a typical large-scale channel-type reservoir, and the water flow in the reservoir area presents an unsteady flow state. Under the influence of the backwater and the inflow flow of the reservoir, the water surface of the reservoir is not horizontal, and the wedge-shaped storage capacity (dynamic storage capacity) formed by the backwater above the horizontal surface It also plays a role in regulating floods, and its flood regulation role is often greater than that of static storage capacity below the water level. At the same time, the Three Gorges Reservoir area has a humid subtropical monsoon climate, and the climate is significantly affected by the canyon topography. The annual rainfall is abundant, with an average annual rainfall of 1150mm. The rivers in the reservoir area are vertical and horizontal, the water system is developed, and there are many tributaries. Except for the main stream of the Yangtze River, Jialing River and Wujiang River In addition to the major tributaries, there are also tributaries such as Dahong River, Quxi River, Long River, Xiaojiang River and Xiangxi River that continuously flow into the main river channel. In addition, the runoff will also be scattered into the reservoir from the surrounding slopes of the reservoir, causing heavy rain and floods in the reservoir area. The rules are very complicated. Therefore, this area is taken as an example to explore the refined runoff theory and related distributed hydrological model modeling methods in the Three Gorges Reservoir area, and accurately simulate the temporal and spatial evolution of precipitation and runoff in the Three Gorges Reservoir area. Specific steps are as follows:

Step 1: It is necessary to collect basic hydrometeorological data, topographic data and other basic data required for the hydrological modeling of the reservoir area to drive the hydrological model to conduct refined modeling of the runoff and runoff process in the Three Gorges Reservoir area. DEM elevation data of the Three Gorges Reservoir area Select the DEM elevation data with a resolution of 30m, as shown in Figure 1, and set 0.05° as the horizontal resolution of the distributed hydrological model grid, and then divide the entire Three Gorges Reservoir area into 2854 0.05°× Orthogonal grids with 0.05° resolution, as computational units for the runoff model of the watershed. as shown in picture 2.

Step 2: Based on the DEM elevation data of the Three Gorges Reservoir area, use ArcGIS tools to extract information such as the boundary of the study area and the river network and sub-basin in the region through processes such as subsidence filling, flow direction generation, cumulative flow calculation, and sub-basin division. . For the sub-watershed, the method of combining slope and river confluence is used to calculate the confluence.

Step 3: Through the constructed distributed hydrological model, the flow process of 54 nodes including 26 sub-basins and 28 main river slopes and inflows in the sub-basin interval is output at the same time, as shown in Figure 3. When calculating the flow process of the watershed node, the watershed is first divided, as shown in Figure 4, including 6 sub-watersheds and 2 slope watersheds (including 6 and 3 hydrological response units respectively). The specific details are as follows:

①For sub-basin 1 and sub-basin 2: the runoff of sub-basin 1 flows into the main channel through node 6, and the sub-basin 2 interacts with the main channel through node 9;

②For the slope watershed: the main river runs through it, and the slope watershed 1 is divided into 6 slope hydrological response units, and the slope watershed 2 is divided into 3 slope hydrological response units. The runoff from hydrological response units 1 to 6 in the slope watershed flows into the main channel through nodes 1 to 6 in sequence, and the runoff from hydrological response units 7 to 9 in the slope watershed flows into the main channel through nodes 7 to 9 in sequence;

③ It indicates that the runoff simulation values of nodes 1 to 9 in the watershed can be used as the continuous output of the hydrological model in space.

Subsequently, the confluence calculation is performed, and the specific details are as follows:

The confluence process of the sub-watershed and the slope watershed is shown in Figure 5 (compared to Figure 4 in this example, the slope watershed is divided into more response units, a total of 10), the details are as follows:

① Sub-basin confluence

a. The runoff of grid A first flows to the nearby river grid H through the slope grids D, E and G. After the runoff enters the river channel in grid H, it flows downstream along the river channel according to the mode of river confluence. Calculation, and finally enter the main channel through node 10.

b. The runoff of grid B flows into river grid H through slope grids C and F.

The confluence path of each grid, that is, the topological relationship between grids, is determined according to the D8 flow direction algorithm, and the movement of runoff in each grid is calculated according to the slope confluence method.

②Slope watershed (hydrological response unit)

The slope watershed is divided into 10 slope hydrological response units. In slope hydrological response units 3, 4, 9, and 10, the runoff flows into the main channel grids J, K, L, and M along the slope, and directly passes through Nodes 3, 4, 9, 10 in grids J, K, L, M flow out.

Finally, the flow process can be obtained by using the river confluence calculation described above in the present invention.

Step 4: Use the multi-objective genetic algorithm to calibrate the parameters of the distributed hydrological model of the Yandu River Basin in the reservoir area, select 26 floods in the basin from 2014 to 2021, and the flood peak range is 104-903 m ³ /s, covering large, medium, and Small floods of various magnitudes are well representative. The parameter calibration results are shown in Table 1. This example can fully prove the accuracy and reliability of the simulation results of the distributed hydrological model for the channel-type reservoir area of the present invention.

Table 1 Flood simulation results in the Yandu River Basin

The absolute value of the average relative error of the flood volume is 13.86%, the absolute value of the average relative error of the flood peak is 19.32%, and the error of the flood peak does not exceed 20% of the measured value. Hong Feng qualified for 16 games, with a pass rate of 61.5%. The average time deviation of the secondary flood peaks in the Yandu River Basin was 1.38 hours, of which 17 peak time deviations were less than or equal to 1 hour, accounting for 65.38%. Therefore, the distributed hydrological model proposed by the present invention has a good simulation effect on the runoff and flood peak of the Yandu River Basin.

The above-mentioned embodiments only represent the embodiments of the present invention, but should not be construed as a limitation on the scope of the present invention. It should be pointed out that for those skilled in the art, without departing from the concept of the present invention, Several modifications and improvements can also be made, which all belong to the protection scope of the present invention.

Claims (5)

1. a distributed hydrological model modeling method in a river channel type reservoir area, is characterized in that, comprises the following steps: The first step is the division of watersheds for different landform types For channel-type reservoirs, precipitation-runoff-confluence in the basin area can be divided into two cases according to the contact form with the main river channel, and the confluence methods are different: the runoff and runoff of the sub-basin flows into the reservoir area in the form of point sources, and the main The strip-shaped slope production and confluence along the river channel are distributed into the reservoir area along the way; 1.1 Division of sub-watersheds and slope watersheds The reservoir area is divided into two types: sub-watershed and slope watershed; the sub-basin has a stable water system, which is hydraulically connected with the main channel in the form of contact points. For the precipitation-runoff process in various parts of the river, it is only necessary to output the runoff at the outlet of the sub-basin; the slope watershed is in direct contact with the main river channel in the form of a surface, and there is no stable and complete water system in the slope watershed. Runoff will be scattered into the main channel; 1.2 Determining the threshold value of the hydrological response unit for the slope watershed In order to determine the refined slope hydrological response unit, the watershed unit is refined by setting various combinations of different thresholds, that is, after setting the area thresholds of different land use types, soil type area thresholds, and slope grade thresholds respectively They are arranged and combined, and the influence of different threshold combinations on the division of hydrological response units is tested and analyzed, and a threshold combination that can take into account the simulation accuracy and calculation efficiency is obtained. The number of slope hydrological response units; The second step, grid size selection Based on the DEM data of the watershed, a distributed hydrological model is constructed, and the watershed is divided into several grids that can reflect the spatial heterogeneity of the watershed; and a method framework for grid division that can meet the simulation accuracy and computational efficiency at the same time is proposed. 2.1 Constructing grids of different scales Resampling the original DEM data to different resolutions to build DEMs with different grid scales; 2.2 DEM analysis and processing The resampled DEM data is further analyzed and processed to obtain the watershed attribute data, and a hydrological model is constructed based on this. Water area threshold, extract the river network of the river basin, and obtain the continuous river network data that conforms to the surface water flow mechanism and matches the actual river network with high degree; ①DEM correction; ②DEM depression and flat processing: In order to ensure the continuity of the flow direction analysis, different processing methods are used to fill the independent depression and the composite depression; When at least one of the surrounding eight adjacent points is the outlet of the catchment area, the minimum value of the eight adjacent points around the point is assigned to this point; when there is no outlet of the watershed area, the minimum elevation on the boundary line of the area is searched. point, assign its value to the filling point less than this value and the eight adjacent points, thus completing the filling work of the independent depression; for the composite depression area with multiple valley bottom points, first start from each valley bottom, and determine the depression through the reverse water flow. The location of the edge point and the exit point, and then use the exit point elevation to replace the elevation of the point with the depression edge elevation less than this value, and then also assign this elevation data to the point where the exit point elevation in the adjacent depression area is lower than this value, so far Filling of composite depressions; For a grid point with the same elevation value as the surrounding eight grids, the flat land needs to be processed to ensure that the water flow can flow out of the flat land area in a directional manner; a point with an elevation greater than that of adjacent points is found on the boundary of the flat land area, and the It is marked as a safe point, and the rest of the points are marked as pending points, and then the elevation of the pending points is increased by a small amount until all the pending points can be marked as safe points. At this point, the processing of the flat ground is completed, and finally the DEM model can be generated. continuous river system; ③Determination of water flow direction: In order to clarify the water flow direction in each cell, first compare the slope between the processed grid cell and the adjacent 8 grid cells, and connect the center of the processed grid cell with the adjacent 8 grid cells. The center of the grid unit with the largest gradient among the grid units, and the connection direction is defined as the water flow direction of the processed grid unit, that is, the D8 algorithm is used to determine the flow rate of the water flow; ④Determination of catchment area and water system: The cumulative amount of water flow at each grid point is the total number of all grids flowing into this grid, that is, the cumulative matrix of confluence; based on the direction of water flow of the grid, each grid is calculated. The confluence accumulation matrix is obtained by multiplying the confluence accumulation matrix by the grid area to obtain the upstream confluence area of each unit grid; In order to determine the threshold value of the catchment area, the relationship between the slope of the river network and the catchment area is established to verify the change of the relationship coefficient; through trial calculation, when the threshold value of the catchment area approaches a certain value, the relationship coefficient tends to be stable , then the catchment area threshold corresponding to this stable relation coefficient is regarded as the most suitable catchment area threshold; The third step is the calculation of runoff and runoff in sub-watersheds and slope watersheds 3.1 Calculation of runoff in sub-watersheds and slope watersheds The runoff processes of the sub-watershed and the slope watershed are consistent. Based on the determination of the grid scale, a distribution that can simultaneously consider the two runoff mechanisms of full storage and over-osmosis and the influence of temperature on the runoff simulation results are adopted for both the sub-watershed and the slope watershed. The runoff model of the formula hydrological model; The runoff calculations for sub-watersheds and slope watersheds are as follows: The infiltration capacity of grid cells varies with space and is expressed by the following formula: f=f _m [1-(1-C) ^1/B ] (1) where f is the infiltration capacity, f _m is the maximum infiltration capacity, C is the area ratio of the infiltration capacity less than or equal to f, and B is the infiltration capacity shape parameter; According to the water balance formula, it is introduced in a given time period P: P=R ₁ (y)+R ₂ (y)+ΔW(y) (2) y=R ₁ (y)+ΔW(y) (3) Among them, p represents the rainfall during the period; y represents the vertical depth; R ₁ (y) represents the full storage runoff; R ₂ (y) represents the hyperosmotic runoff; ΔW(y) represents the change in soil moisture content; In equations (3) and (4), the full storage runoff R ₁ and the change in soil moisture content ΔW can be expressed as:

Among them, im represents the maximum water storage capacity; _{i 0} represents the water storage capacity at a certain point; b represents the water storage shape coefficient; And then get W _p , the expression of R ₂ is:

Among them, B is the shape coefficient of infiltration capacity; Δt is the calculation time step; The base flow calculation adopts the ARNO model, and the calculation formula is:

where D _m is the maximum base flow, D _s is the ratio of the current base flow to D _m ,

is the initial water content of the subsoil,

is the maximum water content of the lower soil, W _s is the water ratio of the lower soil, and R _b is the base flow; 3.2 Calculation of Confluence in Sub-basin and Slope Watershed In order to simulate the spatial evolution of precipitation and runoff in channel-type reservoirs, the confluence method of "first integration and then evolution" is used to calculate the confluence process of the sub-watershed and the slope watershed, and the evolution process of runoff between grids is simulated finely; For sub-catchments, the grid is divided into two types: slope grid and channel grid according to whether the grid contains tributary channels. The slope grid does not contain tributary channels, and the river grid contains tributary channels; In the surface grid, the runoff first merges into the nearby river grid according to the topological relationship of the grid. This process is calculated by the unit line method based on the probability distribution, and then the evolution process of the runoff in the channel grid uses the kinematic wave equation or The impulse response function is used to describe, and the calculation is carried out gradually along the channel grid unit to the sub-basin outlet section until it merges into the main channel; In sub-watershed confluence, the topological relationship between grids is determined according to the D8 flow direction algorithm, and the movement of runoff in each grid is calculated according to the method of slope confluence; In the slope watershed, the runoff is dispersed and directly merges into the main channel, which is in direct contact with the main channel in the form of planes. There is no downward calculation process along the river channel. The process of the slope grid entering the main channel adopts a probability distribution The unit line method is described; ①Slope confluence calculation When calculating slope confluence, the unit line is derived from the probability distribution; ②Calculation of river confluence For channel-type reservoirs, two channel confluence schemes, kinematic wave KWT and impulse response function IRF, are used to calculate the river basin confluence. The KWT method is used for rivers with large gradients, and the IRF method is used for general rivers; The fourth step, distributed hydrological model parameter calibration For the distributed hydrological model, in order to avoid falling into the local convergence point, a multi-objective optimization algorithm is used for parameter calibration; the core of this algorithm is to coordinate the relationship between the objective functions, and find out the Pareto that makes each objective function as large or small as possible. Optimal solution set; use NSGA-II algorithm for calibration; Flood forecasting in the reservoir area focuses on the flood peak, flood volume, flood hydrograph, and flood spatiotemporal distribution. Therefore, when optimizing the parameters of the runoff model of the distributed hydrological model, two objective functions are set in the NSGA-II algorithm, namely flood runoff. The minimum mean value of the relative error absolute value of deep and peak flow is the objective function;

In the formula, n is the number of flood fields, R _f (i) represents the simulated runoff depth of the i-th flood, R _m (i) represents the measured runoff depth of the i-th flood; Q _f (i) represents the i-th flood’s runoff depth. The simulated flood peak, Q _m (i) represents the measured flood peak of the i-th flood; When optimizing the parameters of the confluence model, the maximum mean value of the flood certainty coefficient DC is taken as the objective function:

In the formula, Q _if and Q _im are the simulated flow and the measured flow of the i-th flood in the total flow process, respectively,

The average measured flow of the total flow process; So far, the parameters of the model are obtained, and the construction of the distributed hydrological model of the channel-type reservoir area is completed. 2. the distributed hydrological model modeling method of a kind of river channel type reservoir area according to claim 1, is characterized in that, in described second step step 2.1), the method of resampling is respectively nearest neighbor method, Bilinear interpolation method and cubic convolution method; based on the DEM data obtained after resampling, calculate its root mean square error RMSE, and finally select the resampling method according to the minimum value of RMSE. The RMSE calculation formula is:

In the formula, Z _k is the grid elevation of the DEM after resampling, z _k is the elevation of the corresponding position of the reference data, and n is the number of grids after resampling. 3. the distributed hydrological model modeling method of a kind of river channel type reservoir area according to claim 1, is characterized in that, in described step 2.2 ①): adopt Agree method to carry out the surface elevation data precision of DEM data. For detection and improvement, the DEM data is corrected by fine-tuning the surface elevation of the DEM to maintain continuity and consistency with the river vector map data. 4. the distributed hydrological model modeling method of a kind of river channel type reservoir area according to claim 1, is characterized in that, in described 3rd step step 3.2 ①), adopts two-parameter logarithmic normal distribution, Unit lines are derived from the 2-parameter Pareto distribution, the 2-parameter Weibull distribution, the 2-parameter Frechet distribution, or other probability distribution functions. 5. the distributed hydrological model modeling method of a kind of river channel type reservoir area according to claim 1, is characterized in that, in described 3rd step step 3.2 ②), kinematic wave KWT and impulse response function IRF calculate Methods as below: a. Kinematic wave method KWT The KWT method can calculate the wave speed or flow from the hydrological response unit to an independent river reach in each time period. It approximately assumes that the river channel is rectangular, the hydraulic width, the wave speed C is the channel width ω, the Manning coefficient n, and the river bottom gradient S ₀ , and a function of flow q; among them, the gradient of the river bottom can be calculated from the river network data, and the flow q can be obtained from the above-mentioned slope confluence; the formula for determining the width of the river channel is as follows:

The calculation formula of wave speed in the river reach is as follows:

If the calculated estimated outflow time is before the end of the period, then the wave flows into the downstream reaches, otherwise it is considered to stay in the current period; b. Impulse response function IRF The IRF method is used to perform confluence calculation on the runoff output of the gridded land surface model, and a gridded river network or a vector river network can be used. The flow process can be obtained by accumulating the confluence routing flows of all upstream reaches in the outlet reaches, and this method does not need to pay attention to the sequence of the reaches.

Images