WO2022032872A1 - Big data-based hydrologic forecasting method - Google Patents

Abstract

A big data-based hydrologic forecasting method, comprising: step 1, dividing a whole drainage basin into basic calculation units according to a range; step 2, performing runoff calculation on each calculation unit; step 3, performing confluence calculation on each calculation unit according to a runoff calculation result; step 4, performing river network evolution superposition calculation on the flow output process of each calculation unit to finally obtain the outlet flow of the section of the whole drainage basin; and performing hydrologic forecasting according to the outlet flow of the section of the drainage basin. The present invention solves the problem in the prior art that a forecasting result calculated by a Xinanjiang model is poor in accuracy due to the fact that the parameters of the Xinanjiang model mainly used for hydrologic forecasting cannot be accurately calibrated, and the problem of being difficult to carry out vigorous technical popularization and application to a region having no hydrologic data due to the difficulty of parameter calibration and the limitation of an application region.

Description

A hydrological forecast method based on big data technical field

The invention belongs to hydrological forecasting technology, and particularly relates to a hydrological forecasting method based on big data.

Background technique

Hydrological forecasting plays an important role in flood control, drought resistance and rational utilization of water resources. In the prior art, the commonly used hydrological forecasting model in my country is the Xin'anjiang model. At present, the Xin'anjiang model has certain limitations for hydrological forecasting in areas with no data and in arid regions. In the Xin'anjiang model, the soil evaporation mechanism adopts a three-layer evaporation model. For areas without hydrological data, it is difficult to determine the evaporation parameters of the model, and it is difficult to evaluate the evaporation parameters through the underlying surface factors of the basin; In addition, the runoff calculation mechanism of the Xin'anjiang model is only suitable for humid regions, not for arid regions. The difficulty of parameter calibration and the limitation of application areas make it difficult to popularize the Xin'anjiang model on a large scale, especially in areas without data. The prediction results calculated by the Anjiang model have problems such as poor accuracy.

SUMMARY OF THE INVENTION

The technical problem to be solved by the present invention is: to provide a hydrological forecasting method based on big data, so as to solve the problem that the hydrological forecasting in the prior art mainly adopts the Xin'anjiang model because the parameters of the model cannot be accurately calibrated, which eventually leads to new The prediction results calculated by the Anjiang model have problems such as poor accuracy; the difficulty of parameter calibration and the limitation of application areas make it difficult for the Xin Anjiang model to vigorously promote and apply technical problems in areas without hydrological data.

The technical scheme of the present invention is:

A hydrological forecasting method based on big data, which includes:

Step 1. Divide the entire watershed into basic calculation units according to the scope.

Step 2, carry out the yield calculation for each calculation unit;

Step 3, carry out confluence calculation for each calculation unit according to the calculation result of yield flow;

Step 4: Perform river network evolution and superposition calculation on the output flow process of each calculation unit, and finally obtain the outlet flow of the entire watershed section; according to the hydrological forecast of the outlet flow of the watershed section.

It also includes: step 5, to calibrate the maximum infiltration rate F ₀ , the stable infiltration rate F _c , the Horton infiltration formula coefficient Beta, the field water holding capacity of the basin soil Wm and the parameter Em.

The method for dividing the basic computing units in step 1 includes: taking watersheds that are not nested in the watershed as basic computing units; one computing unit corresponds to one computing point, and the computing points are divided into two types: leaf and interval. Stringen's algorithm connects with downstream computing points.

The precipitation input of each calculation unit adopts 5 square kilometers of precipitation raster data of the meteorological department; it is extracted separately according to the range of each watershed in the basin.

The method for calculating the flow rate of each computing unit described in step 2 includes:

Step 2.1. When the precipitation or rainfall h reaches the ground, deduct the precipitation according to the basin evaporation capacity Em to obtain the deducted rainfall PE, and the lost rainfall is recorded as E _S , that is, PE+E _S =h ; When PE<0, activate soil evaporation calculation, and set PE to 0; after activating soil evaporation calculation, perform step 2.4 to complete an iterative calculation of runoff;

Step 2.2. After the precipitation is deducted, if PE>0, then according to the hydrological standard algorithm, the full storage and runoff is calculated through the water storage capacity curve of the basin; the net rain R and the full storage absorption DW are output, and the output net rain R is passed through. The two water source algorithm obtains the net rain on the ground Rs and the net rain on the ground hg; after obtaining the net rain on the ground Rs, the infiltration rate f _v is calculated according to the Horton infiltration formula combined with the soil water content W;

Step 2.3. According to the current infiltration rate f _v and the runoff area ratio a, carry out the first distribution of net rain in the soil, and obtain the first part of the net rain in the soil dhss1 and the net rain on the surface hs; Net rain distribution, obtain the second part of the net rain in the soil dhss2 from the full absorption DW, and correct the DW; finally combine dhss1 and dhss2 to obtain the final net rain in the soil hss;

Step 2.4. Through the seepage mechanism, according to the free water conversion rate SSCR in the soil and the groundwater leakage conversion rate GCR, the net rain in the soil h _ss to the underground net rain h _g , and the underground net rain h _g to the deep groundwater h _dg are realized. convert.

The content of the confluence calculation for the calculation unit includes: the calculation of the surface confluence Ds, the confluence in the soil Dss and the underground confluence Dg; the surface confluence is realized by the instantaneous unit line method, and the confluence in the soil Dss and the underground confluence Dg are realized by the linear reservoir method;

Algorithm to convert surface net rain to flow process through instantaneous unit line:

Among them, U _i is the value corresponding to the i-th moment of the instantaneous unit line; D _ui is the i-th value corresponding to the surface net rain h _s

The flow value at time, F is the area of the watershed.

The method for performing river network evolution and stacking is the Muskingen channel evolution algorithm.

For an interval node, the outgoing flow Q is equal to the outflow Q _uO of the upstream inflow Q _uI evolving to the node plus the aggregated flow D of the node interval, namely:

D _i =D _si +D _ssi +D _gi

Q _i =Q _uOi +D _i

Among them, Q _uOi-1 represents the evolution and egress flow of the interval channel at the i-1th time, and Q _uIi represents the inflow and outflow of the interval channel at the i-th time,

is the average flow velocity, X is the flow proportion coefficient, L is the length of the interval channel, and _QuOi represents the outflow of the interval channel at the i-th time.

The calibration methods for the maximum infiltration rate F ₀ , the stable infiltration rate F _c , the coefficient Beta of the Horton infiltration formula, the field water holding capacity of the basin soil Wm and the parameter Em include:

According to soil texture and generalized soil porosity, the maximum infiltration rate F ₀ , the stable infiltration rate F _c and the coefficient Beta of the Horton infiltration formula are deduced;

Trial calculation of Wm according to Horton's infiltration formula, and correction trial calculation by citing the influence coefficient of land use type; verification based on soil water content partitions and measured samples, and corresponding feedback optimization; The results of the final output Wm, F ₀ , F _c and Beta grid results;

The method for setting the parameter Em is:

Calculate the drought index and flood season weight coefficient based on 30-year rain station statistics; draw up the drought index curve according to the empirical relationship;

According to the formula of the drought index, the annual evaporation capacity is reversed, and the evaporation capacity in the flood season is calculated in combination with the weight coefficient of the flood season;

Compare the measured samples, and then feedback to optimize the drought index curve;

After the results are obtained, the land use influence coefficient is added for further optimization; until the final result is obtained; the Em grid is finally output.

The calculation method of the soil evaporation is:

Establish a correlation curve between soil water content _w and soil evaporation Ew, referred to as soil evaporation curve;

The evaporation characteristics of different watersheds are reflected by Em, Wm and Wb in the soil evaporation curve; Em is the daily evaporation capacity of the watershed (mm/d), Wm is the field water holding capacity of the watershed soil (mm), and Wb is the soil capillary fracture content of the watershed. Water volume (mm). The mathematical formula for the soil evaporation curve is as follows:

In the formula, K ₁ and K ₂ are the global normalized shape coefficients, which are used to control the curvature of the curve in the two regions of Wm and Wb;

Calculate the soil evaporation Ew at a certain time according to the soil water content w in the watershed at this time; the calculation of soil evaporation must limit Ew+Es<=Em.

The current infiltration rate f _v is derived from the logical expression:

f _v =F(W,F ₀ ,F _c ,Beta)

Among them, F ₀ is the maximum infiltration rate; F _c is the stable infiltration rate; Beta is the coefficient of the Horton infiltration formula.

The formula for the distribution of net rain in the first soil is:

hs=Rs-dhss1

The formula for calculating the net rain in the second soil is:

hss=dhss1+dhss2

When calculating hss in the above steps, it is necessary to add hhs to the total amount of free water in the soil Wss, and Wss should be controlled by MaxWss, the upper limit of the total free water in the soil. It is classified as surface net rain hs; the value of MaxWss is equal to the product of WM and soil free water capacity coefficient WSSC;

dhss=F(W,WM,F _c ,SSCR,h _ss ,h _g )

h _dg =F(F _c ,GCR,h _g )

In the above formula, dhss is the conversion amount of net rain in soil to underground net rain; h _dg is the conversion amount of underground net rain to deep groundwater; in the calculation, it is assumed that deep groundwater no longer produces sink flow.

For each runoff iterative calculation, the following formula must be established according to the principle of water balance:

PE=h _s +h _ss +h _g +h _dg .

Beneficial effects of the present invention:

The present invention is based on the basic principle of traditional hydrological calculation, and a hydrological calculation method reconstructed with reference to some algorithm ideas of the Xin'anjiang model; the present invention combines the physical factors of the underlying surface such as soil, vegetation, land use and geology with the EC model. It can effectively correlate with the parameter system of the watershed, so as to solve the technical problem of hydrological forecasting in areas without data; the water source division of the present invention adopts a layered model, and combines the watershed infiltration curve and dynamic infiltration algorithm to realize the water source division mechanism; The infiltration rate F _c is used to control the distribution of groundwater, the infiltration rate f _v is used to control the net rainfall distribution in the soil, and the total amount of the linear reservoir in the soil is controlled by WSSC; the two parameters of SSCR and GCR are related to the stable infiltration F . _c is strictly related, and can be used to simulate the runoff phenomenon in arid areas to a certain extent; the present invention is based on the instantaneous unit line and dynamically calculated according to the net rainfall intensity, which improves the division density of the basin calculation unit.

The present invention makes full use of modern GIS (remote sensing telemetry and geographic information) system technology, utilizes the powerful functions of GIS to extract basin terrain, soil and vegetation, establishes a big data hydrological forecasting method with physical concept basis, makes the calculation parameters objective, and avoids as much as possible. The subjectivity of artificial debugging or selection of parameters can enhance the versatility of the algorithm to solve the difficulty of flood forecasting in small and medium-sized watersheds in areas without data. The big data analysis method is used to optimize the parameters feedback, which improves the adaptability of algorithm parameters and helps To promote and apply technology to areas with no data.

Advantages of the present invention:

Improved accuracy of model rainfall input.

The estimation and calibration of relevant evaporation parameters are simplified through soil evaporation curves.

The infiltration curve of the soil is fully utilized, and the three water sources are divided through an additional leakage mechanism, which improves the adaptability of the model to the calculation of runoff in arid areas.

Through the big data analysis method, the mapping relationship between the model parameters and the underlying physical factors such as soil, vegetation, land use and geology is established; it provides technical support for flood forecasting in areas without data.

It solves the problems that the existing hydrological forecast mainly uses the Xin'anjiang model because the parameters of the model cannot be accurately calibrated, which eventually leads to the poor accuracy of the forecast results calculated by the Xin'anjiang model; the difficulty of parameter calibration and The limitation of the application area makes it difficult for the Xin'anjiang model to be vigorously promoted and applied in areas without hydrological data.

detailed description

A hydrological forecasting method based on big data, which includes:

Step 1. Calculation unit division: Divide the entire watershed into basic calculation units according to the scope;

The method for dividing the entire watershed into basic computing units according to the scope described in step 1 includes: taking non-nested watersheds as basic computing units; one computing unit corresponds to one computing point, and the computing points are divided into two types: leaf and interval, The upstream computing point is connected with the downstream computing point through Muskingen algorithm.

The precipitation input of each calculation unit of the present invention adopts the rainfall grid data of 5 square kilometers from the meteorological department; it is extracted according to the range of each watershed in the watershed; and the traditional Xin'anjiang model application is generally based on the distribution of rainfall stations and the Tyson polygon. Therefore, the hydrological prediction accuracy of the present invention is greatly improved compared with the prior art.

The present invention uses 30-meter DEM data to establish the data set results of the whole watershed and the watershed of the small watershed below 30km^2, and the calculation unit division is more reasonable and precise than the traditional Xin'anjiang model.

Step 2, carry out the yield calculation for each calculation unit;

The method for calculating the flow rate of each computing unit described in step 2 includes:

Step 2.1. When the precipitation or rainfall h reaches the ground, deduct the precipitation according to the basin evaporation capacity Em, and obtain the precipitation PE after the deduction. The lost rainfall is recorded as E _S , that is, PE+E _S =h ;

When PE<0, activate soil evaporation calculation, and set PE to 0; after activating soil evaporation calculation, perform step 2.4 to complete an iterative calculation of runoff.

The calculation method of the soil evaporation is:

Establish a correlation curve between soil water content _w and soil evaporation Ew, referred to as soil evaporation curve;

The evaporation characteristics of different watersheds are reflected by Em, Wm and Wb in the soil evaporation curve; Em is the daily evaporation capacity of the watershed (mm/d), Wm is the field water holding capacity of the watershed soil (mm), and Wb is the soil capillary fracture content of the watershed. Water volume (mm). The mathematical formula for the soil evaporation curve is as follows:

where K ₁ and K ₂ are global normalized shape coefficients, which are used to control the curvature of the curve in the Wm and Wb regions.

In the runoff calculation, the soil evaporation Ew at this moment can be calculated according to the soil water content w in the watershed at this time; the calculation of soil evaporation must limit Ew+Es<=Em, otherwise the value of Ew should be corrected to ensure Ew+Es =Em.

Step 2.2. After the precipitation is deducted, if PE>0, then according to the hydrological standard algorithm, the full storage and runoff is calculated through the water storage capacity curve of the basin; the net rain R and the full storage absorption DW are output, and the output net rain R is passed through. The two-water source algorithm obtains the above-ground net rain Rs and the underground net rain hg; after obtaining the above-ground net rain Rs, the infiltration rate f _v is calculated according to the Horton infiltration formula combined with the soil water content W.

Step 2.3. According to the current infiltration rate f _v and the runoff area ratio a, carry out the first distribution of net rain in the soil, and obtain the first part of the net rain in the soil dhss1 and the net rain on the surface hs; For net rain distribution, the second part of net rain in soil dhss2 is obtained from the full absorption DW, and DW is corrected; finally, dhss1 and dhss2 are combined to obtain the final net rain in soil hss.

The current infiltration rate f _v is derived from the logical expression:

f _v =F(W,F ₀ ,F _c ,Beta)

Among them, F ₀ is the maximum infiltration rate; F _c is the stable infiltration rate; Beta is the coefficient of the Horton infiltration formula.

The formula for the distribution of net rain in the first soil is:

hs=Rs-dhss1

The formula for calculating the net rain in the second soil is:

hss=dhss1+dhss2

When calculating hss in the above steps, it is necessary to add hhs to the total amount of free water in the soil Wss, and Wss should be controlled by MaxWss, the upper limit of the total free water in the soil. It is classified as surface net rain hs. The value of MaxWss is equal to the product of WM and the free water capacity coefficient WSSC in the soil.

Step 2.4. Through the seepage mechanism, according to the free water conversion rate SSCR in the soil and the groundwater leakage conversion rate GCR, the net rain in the soil h _ss to the underground net rain h _g , and the underground net rain h _g to the deep groundwater h _dg is realized. convert.

dhss=F(W,WM,F _c ,SSCR,h _ss ,h _g )

h _dg =F(F _c ,GCR,h _g )

In the above formula, dhss is the conversion amount of net rain in soil to underground net rain; h _dg is the conversion amount of underground net rain to deep groundwater. In the model, it is assumed that deep groundwater no longer generates sink flow.

For each runoff iterative calculation, the following formula must be established according to the principle of water balance:

PE=h _s +h _ss +h _g +h _dg

Step 3, carry out confluence calculation for each calculation unit according to the calculation result of yield flow;

The content of the confluence calculation performed by the computing unit includes: the calculation of the surface confluence Ds, the confluence in the soil Dss and the underground confluence Dg. The surface confluence is realized by the instantaneous unit line method, and the confluence in the soil Dss and the underground confluence Dg are realized by the linear reservoir method.

Algorithm to convert surface net rain to flow process through instantaneous unit line:

Among them, U _i is the value corresponding to the i-th moment of the instantaneous unit line; D _ui is the flow value at the i-th moment corresponding to the surface net rain h _s , and F is the watershed area.

The key problem in determining the instantaneous unit line is to determine the flood lag time TL in the basin;

The present invention adopts the formula:

T _L =T(H,F,C,J,Ke,LLR)

Calculate; where T(X) is the time delay probability function, H is the surface net rainfall intensity, F is the watershed area, C is the slope velocity coefficient, J is the average gradient ratio of the watershed, Ke is the shape coefficient, and LLR is Lag factor.

Step 4. Perform river network evolution superposition calculation on the output flow process of each calculation unit, and finally obtain the outlet flow of the entire watershed section.

The method for performing river network evolution and stacking is the Muskingen channel evolution algorithm.

For an interval node, the outgoing flow Q is equal to the outflow Q _uO of the upstream inflow _QuI evolving to the node plus the aggregate flow D of the node interval. which is:

D _i =D _si +D _ssi +D _gi

Q _i =Q _uOi +D _i

Among them, Q _uOi-1 represents the evolution and egress flow of the interval channel at the i-1th time, and Q _uIi represents the inflow and outflow of the interval channel at the i-th time,

is the average flow velocity, X is the flow proportion coefficient, L is the length of the interval channel, and _QuOi represents the outflow of the interval channel at the i-th time.

The maximum infiltration rate F ₀ , the stable infiltration rate F _c , the Horton infiltration formula coefficient Beta and the watershed soil field water holding capacity Wm are determined:

1. According to the soil texture, generalize the soil gap, and derive the maximum infiltration rate F ₀ , the stable infiltration rate F _c and the coefficient Beta of the Horton infiltration formula.

2. Calculate Wm according to Horton's infiltration formula, and cite the influence coefficient of land use type to correct the calculation.

3. Validate based on the soil water content zoning and measured samples in China, and carry out corresponding feedback optimization.

4. Repeat multiple cycles to optimize parameters. until a satisfactory result is obtained. Finally output Wm, F ₀ , F _c and Beta grid results.

parameter Em rate

1. Calculate the drought index and flood season weight coefficient based on the 30-year rain station statistics; draw up the drought index curve according to the empirical relationship.

2. According to the formula of drought index, the annual evaporation capacity is inversely inferred, and the evaporation capacity during the flood season is calculated by combining with the weight coefficient of the flood season.

3. Add model calculation, compare the measured samples, and then feedback to optimize the drought index curve.

4. After getting better results, add the land use influence coefficient for further optimization.

5. Perform multiple optimization steps until satisfactory results are obtained. Finally, output the Em raster.

The calibration of parameters is an important part of model application. Most of the watershed hydrological models, especially some parameters of small and medium watersheds, cannot be directly determined by observational experiments. Their values have a certain relationship with the characteristics of the underlying surface of the watershed, but in reality they cannot directly establish a relationship with the characteristics of the underlying surface of the watershed, so it is still difficult to determine the parameters for the watershed hydrological model. question.

The present invention obtains a relatively reasonable parameter estimation result through continuous iteration and feedback trial calculation based on some theoretical mathematical formulas by means of the method of big data analysis, so as to solve the deficiencies of the prior art.

Claims (10)

1. A hydrological forecasting method based on big data, which includes:

   Step 1. Divide the entire watershed into basic calculation units according to the scope.

   Step 2, carry out the yield calculation for each calculation unit;

   Step 3, carry out confluence calculation for each calculation unit according to the calculation result of yield flow;

   Step 4: Perform river network evolution and superposition calculation on the output flow process of each calculation unit, and finally obtain the outlet flow of the entire watershed section; according to the hydrological forecast of the outlet flow of the watershed section.
2. A big data-based hydrological forecasting method according to claim 1, characterized in that: it further comprises: step 5, for the maximum infiltration rate F ₀ , the stable infiltration rate F _c , and the Horton infiltration formula coefficient Beta , watershed soil field water holding capacity Wm and parameter Em were calibrated.
3. The method for hydrological forecasting based on big data according to claim 1, characterized in that: the method for dividing the basic computing unit in step 1 comprises: taking watersheds that are not nested in the watershed as basic computing units; a computing unit Corresponding to a calculation point, the calculation point is divided into two types: leaf and interval. The upstream calculation point is connected with the downstream calculation point through the Muskingen algorithm.
4. A big data-based hydrological forecasting method according to claim 1, characterized in that: the precipitation input of each calculation unit adopts 5 square kilometers of precipitation raster data of the meteorological department; and it is extracted according to each watershed range in the watershed.
5. A big data-based hydrological forecasting method according to claim 1, characterized in that: the method for performing runoff calculation for each computing unit described in step 2 comprises:

   Step 2.1. When the precipitation or rainfall h reaches the ground, deduct the precipitation according to the basin evaporation capacity Em to obtain the deducted rainfall PE, and the lost rainfall is recorded as E _S , that is, PE+E _S =h ; When PE<0, activate soil evaporation calculation, and set PE to 0; after activating soil evaporation calculation, perform step 2.4 to complete an iterative calculation of runoff;

   Step 2.2. After the precipitation is deducted, if PE>0, then according to the hydrological standard algorithm, the full storage and runoff is calculated through the water storage capacity curve of the basin; the net rain R and the full storage absorption DW are output, and the output net rain R is passed through. The two water source algorithm obtains the net rain on the ground Rs and the net rain on the ground hg; after obtaining the net rain on the ground Rs, the infiltration rate f _v is calculated according to the Horton infiltration formula combined with the soil water content W;

   Step 2.3. According to the current infiltration rate f _v and the runoff area ratio a, carry out the first distribution of net rain in the soil, and obtain the first part of the net rain in the soil dhss1 and the net rain on the surface hs; Net rain distribution, obtain the second part of the net rain in the soil dhss2 from the full absorption DW, and correct the DW; finally combine dhss1 and dhss2 to obtain the final net rain in the soil hss;

   Step 2.4. Through the seepage mechanism, according to the free water conversion rate SSCR in the soil and the groundwater leakage conversion rate GCR, the net rain in the soil h _ss to the underground net rain h _g , and the underground net rain h _g to the deep groundwater h _dg are realized. convert.
6. A big data-based hydrological forecasting method according to claim 1, characterized in that: the content of the confluence calculation performed on the computing unit includes: the calculation of the surface confluence Ds, the soil confluence Dss and the underground confluence Dg; the surface confluence adopts the instantaneous The unit line method is implemented, and the confluence Dss in the soil and the underground confluence Dg are implemented by the linear reservoir method;

   Algorithm to convert surface net rain to flow process through instantaneous unit line:

   

   Among them, U _i is the value corresponding to the i-th moment of the instantaneous unit line; D _ui is the flow value at the i-th moment corresponding to the surface net rain h _s , and F is the watershed area.
7. The method for hydrological forecasting based on big data according to claim 1, wherein the method for performing river network evolution and stacking is Muskingen river channel evolution algorithm.

   For an interval node, the outgoing flow Q is equal to the outflow Q _uO of the upstream inflow Q _uI evolving to the node plus the aggregated flow D of the node interval, namely:

   

   D _i =D _si +D _ssi +D _gi

   Q _i =Q _uOi +D _i

   Among them, Q _uOi-1 represents the evolution and egress flow of the interval channel at the i-1th time, and Q _uIi represents the inflow and outflow of the interval channel at the i-th time,

   

   is the average flow velocity, X is the flow proportion coefficient, L is the length of the interval channel, and _QuOi represents the outflow of the interval channel at the i-th time.
8. A big data-based hydrological forecasting method according to claim 2, characterized in that: maximum infiltration rate F ₀ , stable infiltration rate F _c , Horton infiltration formula coefficient Beta, watershed soil field water holding capacity Wm The calibration methods of and parameter Em include:

   According to soil texture and generalized soil porosity, the maximum infiltration rate F ₀ , the stable infiltration rate F _c and the coefficient Beta of the Horton infiltration formula are deduced;

   Trial calculation of Wm according to Horton's infiltration formula, and correction trial calculation by citing the influence coefficient of land use type; verification based on soil water content partitions and measured samples, and corresponding feedback optimization; The results of the final output Wm, F ₀ , F _c and Beta grid results;

   The method for setting the parameter Em is:

   Calculate the drought index and flood season weight coefficient based on 30-year rain station statistics; draw up the drought index curve according to the empirical relationship;

   According to the formula of drought index, the annual evaporation capacity is reversed, and the evaporation capacity during the flood season is calculated by combining with the weight coefficient of the flood season;

   Compare the measured samples, and then feedback to optimize the drought index curve;

   After the results are obtained, the land use influence coefficient is added for further optimization; until the final result is obtained; the Em grid is finally output.
9. A big data-based hydrological forecasting method according to claim 5, wherein the calculation method of the soil evaporation is:

   Establish a correlation curve between soil water content _w and soil evaporation Ew, referred to as soil evaporation curve;

   The evaporation characteristics of different watersheds are reflected by Em, Wm and Wb in the soil evaporation curve; Em is the daily evaporation capacity of the watershed (mm/d), Wm is the field water holding capacity of the watershed soil (mm), and Wb is the soil capillary fracture content of the watershed. Water volume (mm). The mathematical formula for the soil evaporation curve is as follows:

   

   In the formula, K ₁ and K ₂ are the global normalized shape coefficients, which are used to control the curvature of the curve in the two regions of Wm and Wb;

   Calculate the soil evaporation Ew at a certain time according to the soil water content w in the watershed at this time; the calculation of soil evaporation must limit Ew+Es<=Em.
10. A kind of hydrological forecasting method based on big data according to claim 5, is characterized in that: infiltration rate f _v derivation logical expression is:

    f _v =F(W,F ₀ ,F _c ,Beta)

    Among them, F ₀ is the maximum infiltration rate; F _c is the stable infiltration rate; Beta is the coefficient of the Horton infiltration formula.

    The formula for the distribution of net rain in the first soil is:

    

    hs=Rs-dhss1

    Among them, f _v is the current infiltration rate, which is derived from the infiltration curve and soil water content, and a is the runoff area ratio;

    The formula for calculating the net rain in the second soil is:

    

    hss=dhss1+dhss2

    When calculating hss in the above steps, hhs needs to be added to the total amount of free water in the soil Wss, and Wss should be controlled by MaxWss, the upper limit of the total free water in the soil. If the upper limit is exceeded, hss needs to be re-corrected and the excess part It is classified as surface net rain hs; the value of MaxWss is equal to the product of WM and soil free water capacity coefficient WSSC;

    dhss=F(W,WM,F _c ,SSCR,h _ss ,h _g )

    h _dg =F(F _c ,GCR,h _g )

    In the above formula, dhss is the conversion amount of net rain in soil to underground net rain; h _dg is the conversion amount of underground net rain to deep groundwater; in the calculation, it is assumed that deep groundwater no longer produces sink flow.

    For each runoff iterative calculation, the following formula is guaranteed according to the principle of water balance:

    PE=h _s +h _ss +h _g +h _dg .