CN118112685A - Watershed hydrologic efficient high-precision forecasting method for non-data area - Google Patents
Watershed hydrologic efficient high-precision forecasting method for non-data area Download PDFInfo
- Publication number
- CN118112685A CN118112685A CN202311361992.6A CN202311361992A CN118112685A CN 118112685 A CN118112685 A CN 118112685A CN 202311361992 A CN202311361992 A CN 202311361992A CN 118112685 A CN118112685 A CN 118112685A
- Authority
- CN
- China
- Prior art keywords
- rainfall
- watershed
- data
- forecasting
- hydrologic
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000013277 forecasting method Methods 0.000 title claims abstract description 9
- 238000000034 method Methods 0.000 claims abstract description 109
- 230000008569 process Effects 0.000 claims abstract description 71
- 238000010801 machine learning Methods 0.000 claims abstract description 29
- 238000004519 manufacturing process Methods 0.000 claims abstract description 17
- 238000005259 measurement Methods 0.000 claims abstract description 10
- 238000012545 processing Methods 0.000 claims abstract description 7
- 238000012549 training Methods 0.000 claims description 20
- 238000004422 calculation algorithm Methods 0.000 claims description 16
- 238000005516 engineering process Methods 0.000 claims description 15
- 238000004458 analytical method Methods 0.000 claims description 7
- 230000006870 function Effects 0.000 claims description 7
- 230000008595 infiltration Effects 0.000 claims description 7
- 238000001764 infiltration Methods 0.000 claims description 7
- 230000008020 evaporation Effects 0.000 claims description 6
- 238000001704 evaporation Methods 0.000 claims description 6
- 239000000463 material Substances 0.000 claims description 6
- 238000012876 topography Methods 0.000 claims description 6
- 238000012360 testing method Methods 0.000 claims description 5
- 238000012795 verification Methods 0.000 claims description 5
- 238000011160 research Methods 0.000 claims description 4
- 238000012216 screening Methods 0.000 claims description 4
- 238000005520 cutting process Methods 0.000 claims description 3
- 238000003874 inverse correlation nuclear magnetic resonance spectroscopy Methods 0.000 claims description 3
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 15
- 238000010276 construction Methods 0.000 description 3
- 238000011161 development Methods 0.000 description 3
- 238000013439 planning Methods 0.000 description 3
- 230000002265 prevention Effects 0.000 description 3
- 238000003860 storage Methods 0.000 description 3
- 238000012271 agricultural production Methods 0.000 description 2
- 230000002950 deficient Effects 0.000 description 2
- 238000013461 design Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 230000001970 hydrokinetic effect Effects 0.000 description 2
- 238000007726 management method Methods 0.000 description 2
- 238000004088 simulation Methods 0.000 description 2
- 238000012935 Averaging Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000012937 correction Methods 0.000 description 1
- 238000009826 distribution Methods 0.000 description 1
- 238000007499 fusion processing Methods 0.000 description 1
- 238000003973 irrigation Methods 0.000 description 1
- 230000002262 irrigation Effects 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 238000010248 power generation Methods 0.000 description 1
- 238000007670 refining Methods 0.000 description 1
- 230000004044 response Effects 0.000 description 1
Landscapes
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a watershed hydrologic high-efficiency high-precision forecasting method aiming at a non-data area. Processing the public data based on ArcGIS to obtain modeling data; constructing a watershed two-dimensional hydrological hydrodynamic model based on the modeling data; simulating by using a watershed two-dimensional hydrological hydrodynamic model to obtain a watershed rainfall production converging process under each rainfall scene, establishing a watershed rainfall-runoff process database, and establishing a rainfall characteristic parameter-runoff process database based on the watershed rainfall-runoff process database; constructing a machine learning-based prediction model by taking the rainfall characteristic parameter-runoff process database as a learning object; inputting actual measurement or forecast rainfall to a forecast model, and forecasting the runoff process of the river basin. The invention solves the problems of difficult hydrologic forecasting, low efficiency and insufficient precision in the non-data area.
Description
Technical Field
The invention belongs to the technical field of drainage basin flood forecasting, and particularly relates to a drainage basin hydrologic high-efficiency high-precision forecasting method aiming at a non-data area.
Background
Hydrology is an important support for realizing the optimal allocation and sustainable development of water resources, maintaining the safety of water environment and the like; the watershed hydrologic forecasting is one of classical subjects of hydrologic science, and refers to a process of predicting hydrologic conditions in a future period of time through a scientific method according to information such as historical meteorological data, hydrologic data, meteorological forecasting data and the like. Watershed hydrologic forecasting plays a vital role in ensuring reasonable utilization of water resources, disaster prevention and control, ecological environment protection, agricultural production, urban planning and the like. For example, the river basin production and confluence process in a period of time in the future is predicted in advance, so that the water supply condition of the reservoir is determined, important reference data is provided for dam construction, the design capacity and the water storage level of the dam are determined, the problems of flood discharge, too high or too low water level of the reservoir and the like caused by improper design can be avoided through reasonable hydrologic prediction, and the safety and the stability of the dam are ensured; the method can also help reservoir managers to reasonably arrange water resource scheduling, and reasonable scheduling can utilize the water storage of the water reservoir to the greatest extent, meet the requirements of irrigation, urban water supply, power generation and the like, and reduce the influence of flood and drought period. Along with the rapid development of hydrology, the hydrologic forecasting technology has become an important scientific basis for reasonable allocation of water resources, planning construction of hydraulic engineering, flood prevention, drought resistance and other engineering construction. However, in some areas, because ground stations are not established or management is poor, a large amount of hydrologic data is deficient, and accuracy and efficiency of hydrologic forecasting of a river basin by adopting a traditional method are difficult to meet requirements, so hydrologic forecasting of a data-free area becomes a critical problem to be solved.
Disclosure of Invention
The purpose of the invention is that: the method for forecasting the watershed hydrology in the data-free area in a high-efficiency and high-precision mode is provided. The invention solves the problems of difficult hydrologic forecasting, low efficiency and insufficient precision in the non-data area.
The technical scheme of the invention is as follows: a watershed hydrologic high-efficiency high-precision forecasting method for a non-data area is based on ArcGIS to process public data to obtain modeling data; constructing a watershed two-dimensional hydrological hydrodynamic model based on the modeling data; simulating by using a watershed two-dimensional hydrological hydrodynamic model to obtain a watershed rainfall production converging process under each rainfall scene, establishing a watershed rainfall-runoff process database, and establishing a rainfall characteristic parameter-runoff process database based on the watershed rainfall-runoff process database; constructing a machine learning-based prediction model by taking the rainfall characteristic parameter-runoff process database as a learning object; inputting actual measurement or forecast rainfall to a forecast model, and forecasting the runoff process of the river basin.
In the method for efficiently and accurately forecasting the hydrologic domain of the non-data area, the modeling data is obtained as follows: and carrying out refinement treatment on the disclosed DEM elevation data by adopting a spatial interpolation and topography analysis technology, and dividing the basin land utilization type by taking the disclosed image data as a reference.
In the method for efficiently and accurately forecasting the watershed hydrology of the non-data area, the method for acquiring the wide-range and public DEM elevation data comprises the following steps of: firstly, downloading data in blocks, and then, fusing the data downloaded in the blocks by utilizing a mobile to NEW RASTER or a mobile tool in the ArcGIS.
In the method for efficiently and accurately forecasting the hydrologic domain of the non-data area, the spatial interpolation method is as follows:
Importing the publicly downloaded DEM elevation data into an ArcGIS, and performing spatial interpolation by using INVERSE DISTANCE WEIGHTED interpolation tools in the ArcGIS; the method comprises the following steps: firstly, selecting an interpolation method, and setting the size of an output grid, an interpolation range and interpolation weights; and secondly, operating an interpolation tool, and automatically interpolating the imported DEM elevation data by the ArcGIS to generate high-resolution DEM elevation data.
In the method for efficiently and accurately forecasting the hydrologic state of the river basin in the area without data, the terrain analysis technology is implemented as follows: referring to the disclosed image data, establishing a surface file by using an ArcGIS, starting an editing function, manually drawing a region with complex topography in a region by using a Create Features tool in the ArcGIS, cutting and manually correcting the DEM elevation data subjected to spatial interpolation processing by using a Clip or Extract by Mask tool in the ArcGIS, and fusing the cut and corrected region with the DEM elevation data subjected to spatial interpolation processing.
In the method for efficiently and accurately forecasting the hydrologic state of the river basin in the area without data, the classification of the land utilization types of the river basin is as follows: referring to the disclosed image data, an arcGIS is utilized to establish a surface file, an editing function is started, and various land utilization types are manually drawn by utilizing a Create Features tool in the arcGIS.
In the method for efficiently and accurately forecasting the watershed hydrology in the non-data area, the watershed two-dimensional hydrologic model is constructed according to the finely processed DEM elevation data and the watershed land utilization type and by combining the watershed hydrologic characteristic data; the watershed hydrologic characteristic data comprise rainfall, infiltration and evaporation.
In the method for efficiently and accurately forecasting the watershed hydrology in the area without data, the constructed watershed two-dimensional hydrology hydrodynamic model performs accuracy verification by using the actually measured rainfall-runoff data of the watershed.
In the method for efficiently and accurately forecasting the watershed hydrology of the non-data area, the machine learning model is constructed as follows:
Extracting rainfall characteristic parameters which mainly influence the drainage basin production and collection process, and setting various rainfall situations by combining the rainfall characteristics of the research area; simulating a basin production and convergence process under each rainfall scene by using the basin two-dimensional hydrologic hydrodynamic model, and establishing a basin rainfall-runoff process database under each rainfall scene; extracting rainfall characteristic parameters, and establishing a rainfall characteristic parameter-runoff process database; and training and learning a rainfall characteristic parameter-runoff process database by adopting a K neighbor machine learning algorithm, and establishing a machine learning-based watershed hydrologic forecasting model in the non-data area.
In the method for forecasting the watershed hydrology in the non-data area with high efficiency and high precision, the forecasting of the watershed runoff process is as follows: inputting measured rainfall or weather forecast rainfall data of the river basin into a forecast model; respectively calculating the distance between the predicted sample and each original sample in the training set by using a Euclidean distance formula, and screening K samples closest to the predicted sample from the distances; and distributing the prediction results of the K samples according to the weight and distributing the prediction results to new test samples as prediction values, namely, the basin rainfall runoff process.
The invention has the advantages that: because a large amount of hydrologic data is deficient in a part of areas due to the fact that ground stations are not established or management is poor, accuracy and efficiency of forecasting the hydrologic of the river basin by adopting a traditional method are difficult to meet the requirements. Based on the technical problem, the modeling data required by the two-dimensional hydrologic hydrodynamic model is obtained by carrying out refinement treatment on the obtained public data by utilizing the ArcGIS technology, the problem of no data is solved, and compared with other data processing platforms, the method adopts the ArcGIS technology for processing, and has remarkable effects on map data precision, data integrity, logic consistency, attribute precision and accessory quality. And then, constructing a watershed two-dimensional hydrologic hydrodynamic model according to the obtained modeling data, and simulating by utilizing the hydrodynamic model to obtain a high-precision watershed rainfall runoff full hydrodynamic physical process. Because the calculation amount of the rainfall runoff process of the river basin level is large, and the simulation efficiency of the two-dimensional hydrologic model is far lower than that of the hydrologic model, the practical application requirement cannot be met; therefore, a large number of simulation is carried out by utilizing a two-dimensional hydrologic hydrodynamic model, so that drainage basin rainfall production and convergence processes under different rainfall situations are obtained, a drainage basin rainfall-runoff process database is established, and then a rainfall characteristic parameter-runoff process database is established; then taking the model as a learning object of a machine learning model, and adopting a K Nearest Neighbor (KNN) machine learning algorithm to establish the machine learning model; the K Nearest Neighbor (KNN) machine learning algorithm does not need to build a model in a training stage, so that the training speed of the model is effectively improved, new data points can be adapted in real time, the requirement of model forecasting efficiency can be fully met, and the purpose of efficient and high-precision data-free hydrologic forecasting is achieved. Finally, data support is provided for reasonable water resource utilization, disaster prevention and control, ecological environment protection, agricultural production and urban planning, scientific basis is provided for decision makers, sustainable development of society is promoted, and life safety and economic stability of people are ensured.
In order to verify the beneficial effects of the present invention, the inventors performed the following experimental verification:
The method is verified by adopting a reservoir in Jilin city, wherein the normal reservoir capacity of the reservoir is 700 kilocubic meters, and the rain collecting area is 69 square kilometers. And establishing a two-dimensional hydrodynamic model by collecting elevation data of a water collecting area, infiltration, land utilization, actual measurement of warehouse-in flow and corresponding rainfall, and verifying. And then setting 81 groups of rainfall scenes according to the method, respectively inputting the calibrated and verified two-dimensional hydrologic hydrodynamic model to obtain corresponding reservoir storage flow results, and establishing a drainage basin rainfall-runoff process database. And then taking the model as a learning object of a K Nearest Neighbor (KNN) machine learning model to establish the machine learning model. The method is verified by selecting local rainfall of 7 th month 6 years and 15 th month 2022 and corresponding actual measurement and warehousing flow, and under the condition of 2022 rainfall of 7 th month 6 days, the actual measurement flood peak flow of the reservoir is 58m 3/s, and the predicted flood peak flow is 67m 3/s; under the condition of rainfall of 2022, 7 and 15 days, the measured flood peak flow rate of the reservoir is 43m 3/s, and the predicted flood peak flow rate is 49m 3/s; the prediction time of the model under the condition of two rainfall fields is less than 1s, which shows that the method has extremely high prediction efficiency while ensuring the precision, and is specifically shown in the table 1.
TABLE 1 comparison of typical rainfall event flood peak flow, peak time and total incoming water forecast and measured values
Drawings
Fig. 1 is a flow chart of the present invention.
Detailed Description
The invention is further illustrated by the following figures and examples, which are not intended to be limiting.
Example 1. Referring to fig. 1, firstly, the disclosed DEM elevation data and image data are acquired, the DEM elevation data are refined by adopting a spatial interpolation and terrain analysis technology, and the ArcGIS software is utilized to divide the land utilization types of the river basin by taking the image data as a reference. Secondly, combining hydrologic characteristic data such as drainage basin rainfall, infiltration, evaporation and the like, constructing a two-dimensional hydrologic hydrodynamic model in the drainage basin production and convergence process, and verifying the accuracy of the model by utilizing actually measured drainage basin rainfall-runoff data. Then, according to the machine learning training characteristics, extracting rainfall characteristic parameters which mainly influence the basin production and collection process; and setting different rainfall situations according to the rainfall characteristics of the research area, simulating the high-precision drainage basin production and convergence process under different rainfall situations by using the constructed drainage basin production and convergence process two-dimensional hydrokinetic model, and establishing a drainage basin rainfall-runoff process database under different rainfall situations of the drainage basin. Finally, extracting rainfall characteristic parameters, establishing a rainfall characteristic parameter-runoff process database, training and learning the rainfall characteristic parameter-runoff process database by adopting a K Nearest Neighbor (KNN) machine learning algorithm, establishing a data-free regional river basin hydrologic high-efficiency high-precision forecasting model based on a machine learning technology, and inputting actual measurement or forecasting rainfall to efficiently and high-precision forecasting river basin runoff process.
The method is implemented according to the following steps:
Step 1, obtaining network-published DEM terrain and image data, refining the DEM elevation data by adopting a spatial interpolation and terrain analysis technology, and dividing the basin land utilization type by using ArcGIS software with the image data as a reference.
And 2, obtaining DEM data and land utilization data according to the step 1, and constructing a drainage basin two-dimensional hydrologic hydrodynamic model by combining hydrologic characteristic data such as drainage basin rainfall, infiltration, evaporation and the like.
And 3, performing parameter calibration and accuracy verification on the two-dimensional hydrologic hydrodynamic model of the river basin constructed in the step 2 by using the actually measured rainfall-runoff data of the river basin.
And step 4, determining rainfall characteristic parameters which mainly influence the basin production and confluence process according to training characteristics of a KNN machine learning algorithm.
And 5, setting different rainfall situations according to the rainfall characteristics of the research area and the rainfall characteristic parameters determined in the step 4, simulating a high-precision drainage basin production and convergence process under different rainfall situations by using the constructed drainage basin production and convergence process two-dimensional hydrokinetic model, and establishing a drainage basin rainfall-runoff process database under different rainfall situations.
And 6, automatically extracting rainfall characteristic parameters by using python programming software programming, converting the drainage basin rainfall-runoff process data into rainfall characteristic parameters-runoff process data, training and learning the established drainage basin hydrologic high-efficiency high-precision forecasting model based on a machine learning technology by adopting a KNN machine learning algorithm.
And 7, inputting the actually measured rainfall or weather forecast rainfall data of the river basin into a model for efficiently and accurately forecasting the hydrologic state of the river basin in the non-data area based on a machine learning technology, and rapidly outputting a corresponding rainfall runoff process of the river basin by the model.
The step 1 is specifically as follows:
And 1.1, acquiring the disclosed DEM data and image data from the Internet. Here, if the range is large, the data with the highest accuracy cannot be directly acquired, and then the data needs to be downloaded in blocks, and then fusion processing is performed by using a mosaicto NEW RASTER or a mosaictool in the ArcGIS.
And step 1.2, performing spatial interpolation on the obtained public DEM data. Importing DEM data into ArcGIS, performing spatial interpolation by using INVERSE DISTANCE WEIGHTED interpolation tools, selecting an interpolation method, and setting the size of an output grid, an interpolation range and interpolation weights; and operating an interpolation tool, and automatically interpolating the DEM data by the ArcGIS to generate high-resolution DEM data.
And step 1.3, performing terrain analysis on the high-resolution DEM data generated in the step 1.2. Referring to the obtained public image data, establishing a surface file by using an ArcGIS, starting an editing function, manually drawing areas with complex topography in a flow domain, such as cities, mountains, rivers, valleys and the like by using a Create Features tool, cutting and manually correcting the high-resolution DEM data generated in the step 1.2 by using a Clip or Extract by Mask tool in the ArcGIS, and fusing the cut and corrected areas with the high-resolution DEM data generated in the step 1.2 to obtain the refined DEM data required by the two-dimensional hydrological hydrodynamic model.
And 1.4, dividing the drainage basin land utilization types. Referring to the public image data obtained in the step 1.1, establishing a surface file by using an ArcGIS, starting an editing function, manually drawing different land utilization types by using a Create Features tool, and obtaining land utilization data required by a two-dimensional hydrologic hydrodynamic model.
The step 2 is specifically as follows:
And (3) obtaining refined DEM data and land utilization data required by the two-dimensional hydrologic model according to the step (1), and constructing the two-dimensional hydrologic model by combining hydrologic characteristic data such as rainfall, infiltration, evaporation and the like of the watershed. It is noted here in particular that if there is a lack or insufficiency of hydrologic characteristics such as rainfall, infiltration, evaporation, etc. in the basin, it is possible to replace it with hydrologic characteristics of nearby or similar basins, and then to rate the parameters by means of actual measurement data.
The step 3 is specifically as follows:
and (3) calibrating model parameters and checking accuracy by using the actually measured drainage basin rainfall-runoff process data, wherein the actually measured drainage basin rainfall-runoff process data is not lower than two groups, one group carries out parameter correction, and the other group carries out model verification so as to ensure the accuracy of the constructed drainage basin two-dimensional hydrologic hydrodynamic model.
The step 4 is specifically as follows:
because the training mode of the KNN machine learning algorithm is that one or more characteristic values correspond to one or a group of results, it is necessary to extract characteristic parameters of the rainfall process, and the whole rainfall process is represented as comprehensively as possible by using several characteristic parameters. According to the method, three parameters with higher correlation among a rainpeak coefficient, the accumulated rainfall within 3 hours at maximum and the accumulated total rainfall are finally selected as rainfall characteristic parameters by taking the correlation between characteristic parameters and the influence of the runoff process as a principle.
The step 5 is specifically as follows:
Sufficient training data is the basis for constructing a watershed hydrologic efficient high-precision forecasting model of a non-material area based on a machine learning technology, and the training data must contain watershed rainfall-runoff process data under all conditions of the watershed as much as possible.
And 5.1, setting the reproduction period to be 1 year, 2 years, 3 years, 5 years, 10 years, 20 years, 30 years, 50 years and 100 years according to a river basin rainfall intensity formula or a 24-hour rainfall distribution coefficient, and setting different conditions in 3 according to the rainpeak coefficient, the maximum 3-hour accumulated rainfall and the accumulated total rainfall. Wherein the rain peak coefficient is divided into a front peak, a middle peak and a rear peak, and the accumulated rainfall for 3 hours is set according to the size; the accumulated total rainfall is also divided into 1 year first, 2 years first, 3 years first, 5 years first, 10 years first, 20 years first, 30 years first, 50 years first, 100 years first and 9 minutes according to different rainfall reproduction periods. According to the situation, the total of 81 rainfall scenes are arranged and combined.
And 5.2, respectively inputting the 81 rainfall scenes determined in the step 5.1 into the two-dimensional hydrologic hydrodynamic model of the drainage basin production and convergence process constructed and calibrated in the step 2 and the step 3, simulating the drainage basin runoff process under each rainfall scene, and establishing a drainage basin rainfall-runoff process database.
The step 6 is specifically as follows:
Step 6.1, a KNN algorithm is adopted as a core algorithm for constructing a high-efficiency high-precision model for forecasting the hydrologic of a watershed in a non-data area based on a machine learning technology, and the specific algorithm principle is as follows:
The KNN algorithm is a learning algorithm developed on the basis of the theory of a vector space model, and has higher stability. The mechanism of the algorithm is concise, each sample is regarded as a vector or coordinate point in the R n space in the algorithm, K 'neighbors' with the smallest distance with the input sample, namely the most similar distance, are found through a distance formula, and then the new sample is predicted through information provided by the K 'neighbors'.
(1) Sorting the training samples into a format of (x, f (x)); where x is a characteristic parameter of the sample, and may have a plurality of characteristic parameters, i.e., (x 1,x2,x3,…,xn).
(2) And for the input samples to be predicted, calculating the distances between the samples and each sample in the training set one by one through a distance formula, and screening K samples closest to the x distance from the distances. The distance formula is shown as formula (1).
Wherein:
x i,xj is two samples;
x il,xjl is the first eigenvalue of samples x i and x j, respectively;
L p(xi,xj) is the distance between samples x i and x j.
The distance measurement adopts a p function, wherein p is a super parameter in a KNN algorithm, p is more than or equal to 1, when p=1, L 1(xi,xj) is Manhattan distance, and the formula is shown as formula (2):
when p=2, L 2(xi,xj) is the euclidean distance, the formula is shown in formula (3):
When p= infinity, L ∞(xi,xj) is the chebyshev distance, as shown in formula (4):
(3) Selecting K samples closest to the input sample in distance according to a distance formula, and determining by using a mean value when the regression problem is processed, namely taking the mean value of the K target values closest to the sample in the training set as a target result of a new sample; in response to this, the model may also assign different weights to each "neighbor" based on the proximity between the input sample and the "neighbor", and determine the target result of the new sample by means of weighted averaging.
And 6.2, training the model by taking the basin rainfall-runoff process database obtained in the step 5 as a sample based on a machine learning theory KNN model, and specifically implementing the method according to the following steps:
Step 6.2.1, extracting characteristic parameters of rainfall data of the existing samples; and extracting three characteristic parameters (a rain peak coefficient, a maximum 3-hour accumulated rainfall and an accumulated total rainfall) of 81 rainfall and two actually measured rainfall by using python programming, and converting the river basin rainfall-runoff data into rainfall characteristic parameter-runoff process data.
And 6.2.2, inputting the 81 rainfall characteristic parameters-runoff process data obtained in the step 6.2.1 into a KNN machine learning model for training, and completing sample instantiation to obtain an (. Obj) file.
And 6.3, testing the (.obj) file trained in the step 6.2.2 by utilizing the rainfall characteristic parameters of the two actually measured waterbasins and the runoff process data in the step 6.2.1, continuously and manually adjusting the K value until the accuracy is optimal, and finally successfully constructing a model for efficiently and accurately forecasting the waterbasins hydrologic in the non-data area based on the machine learning technology.
The step 7 is specifically as follows:
And 7.1, inputting actual measurement rainfall or weather forecast rainfall data of the river basin into a water efficient high-precision forecasting model of the river basin in the non-data area based on a machine learning technology, respectively calculating the distance between a predicted sample and each original sample in a training set by using a distance formula, and screening K samples closest to the predicted sample. And finally, selecting a Euclidean distance formula which is good in performance on both the training set and the testing set, namely formula (3), through comprehensive comparison of fitting effects.
And 7.2, distributing the predicted results of the K samples according to the weight and distributing the predicted results to new test samples as predicted values, namely, the basin rainfall runoff process.
Claims (10)
1. A watershed hydrologic high-efficiency high-precision forecasting method for a non-data area is characterized in that open data are processed based on ArcGIS to obtain modeling data; constructing a watershed two-dimensional hydrological hydrodynamic model based on the modeling data; simulating by using a watershed two-dimensional hydrological hydrodynamic model to obtain a watershed rainfall production converging process under each rainfall scene, establishing a watershed rainfall-runoff process database, and establishing a rainfall characteristic parameter-runoff process database based on the watershed rainfall-runoff process database; constructing a machine learning-based prediction model by taking the rainfall characteristic parameter-runoff process database as a learning object; inputting actual measurement or forecast rainfall to a forecast model, and forecasting the runoff process of the river basin.
2. The efficient and high-precision watershed hydrologic forecasting method for a non-material area according to claim 1, wherein the modeling material acquiring process is as follows: and carrying out refinement treatment on the disclosed DEM elevation data by adopting a spatial interpolation and topography analysis technology, and dividing the basin land utilization type by taking the disclosed image data as a reference.
3. The method for efficient and high-precision forecasting of watershed hydrology for non-material areas according to claim 2, wherein the method for obtaining wide-range and public DEM elevation data is as follows: firstly, downloading data in blocks, and then, fusing the data downloaded in the blocks by utilizing a mobile to NEW RASTER or a mobile tool in the ArcGIS.
4. The method for efficient and high-precision watershed hydrologic forecasting of a non-material area according to claim 2, characterized in that the method of spatial interpolation is as follows:
Importing the publicly downloaded DEM elevation data into an ArcGIS, and performing spatial interpolation by using INVERSE DISTANCE WEIGHTED interpolation tools in the ArcGIS; the method comprises the following steps: firstly, selecting an interpolation method, and setting the size of an output grid, an interpolation range and interpolation weights; and secondly, operating an interpolation tool, and automatically interpolating the imported DEM elevation data by the ArcGIS to generate high-resolution DEM elevation data.
5. The method for efficient and high-precision forecasting of watershed hydrology for a non-informative area according to claim 2, wherein the topography analysis technique is implemented as follows: referring to the disclosed image data, establishing a surface file by using an ArcGIS, starting an editing function, manually drawing a region with complex topography in a region by using a Create Features tool in the ArcGIS, cutting and manually correcting the DEM elevation data subjected to spatial interpolation processing by using a Clip or Extract by Mask tool in the ArcGIS, and fusing the cut and corrected region with the DEM elevation data subjected to spatial interpolation processing.
6. The method for efficient and high-precision watershed hydrologic forecasting of a non-data area according to claim 2, characterized in that the division of the watershed land utilization types is as follows: referring to the disclosed image data, an arcGIS is utilized to establish a surface file, an editing function is started, and various land utilization types are manually drawn by utilizing a Create Features tool in the arcGIS.
7. The method for efficiently and accurately forecasting the watershed hydrology in the non-data area according to claim 2, wherein the watershed two-dimensional hydrology hydrodynamic model is constructed according to the finely processed DEM elevation data and the watershed land utilization type and by combining the watershed hydrology characteristic data; the watershed hydrologic characteristic data comprise rainfall, infiltration and evaporation.
8. The efficient and high-precision watershed hydrologic forecasting method for the non-data area according to claim 1, wherein the constructed watershed two-dimensional hydrologic hydrodynamic model performs precision verification by using the actually measured watershed rainfall-runoff data.
9. The efficient high-precision watershed hydrologic forecasting method for a non-material area according to claim 1, wherein the machine learning model is constructed as follows:
Extracting rainfall characteristic parameters which mainly influence the drainage basin production and collection process, and setting various rainfall situations by combining the rainfall characteristics of the research area; simulating a basin production and convergence process under each rainfall scene by using the basin two-dimensional hydrologic hydrodynamic model, and establishing a basin rainfall-runoff process database under each rainfall scene; extracting rainfall characteristic parameters, and establishing a rainfall characteristic parameter-runoff process database; and training and learning a rainfall characteristic parameter-runoff process database by adopting a K neighbor machine learning algorithm, and establishing a machine learning-based watershed hydrologic forecasting model in the non-data area.
10. The method for efficient and high-precision forecasting of watershed hydrology for a non-informative area according to claim 1, wherein the forecasting of the watershed runoff process is as follows: inputting measured rainfall or weather forecast rainfall data of the river basin into a forecast model; respectively calculating the distance between the predicted sample and each original sample in the training set by using a Euclidean distance formula, and screening K samples closest to the predicted sample from the distances; and distributing the prediction results of the K samples according to the weight and distributing the prediction results to new test samples as prediction values, namely, the basin rainfall runoff process.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311361992.6A CN118112685A (en) | 2023-10-19 | 2023-10-19 | Watershed hydrologic efficient high-precision forecasting method for non-data area |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311361992.6A CN118112685A (en) | 2023-10-19 | 2023-10-19 | Watershed hydrologic efficient high-precision forecasting method for non-data area |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN118112685A true CN118112685A (en) | 2024-05-31 |
Family
ID=91213072
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311361992.6A Pending CN118112685A (en) | 2023-10-19 | 2023-10-19 | Watershed hydrologic efficient high-precision forecasting method for non-data area |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN118112685A (en) |
-
2023
- 2023-10-19 CN CN202311361992.6A patent/CN118112685A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN114372685B (en) | Urban storm waterlogging risk assessment method based on SWMM model | |
| CN113673765B (en) | Small drainage basin flood forecasting and early warning method and terminal | |
| CN112149314B (en) | Multi-sand reservoir capacity silt flushing simulation method based on virtual reservoir capacity correction | |
| CN108614915B (en) | Hydrological model free construction strategy method based on scene driving | |
| CN109815611B (en) | Basin boundary generating method based on digital basin | |
| CN115130396A (en) | Distributed hydrological model modeling method for riverway type reservoir area | |
| CN111143498B (en) | Small river flood forecasting method | |
| CN114648617A (en) | Water system extraction method based on digital elevation model DEM | |
| CN105279317A (en) | DEM based flat ground river network water flow direction estimation method | |
| CN117332291B (en) | Distributed photovoltaic-oriented regional resource monitoring layout method and system | |
| CN112700349A (en) | Anemometer tower site selection method and device | |
| CN113987969A (en) | Land-water dynamic simulation model of multi-elevation scale water flow network | |
| Wang et al. | A statistical hydrological model for Yangtze river watershed based on stepwise cluster analysis | |
| CN113609683B (en) | Rainwater pipe network sub-catchment area dividing method based on basin analysis and Thiessen polygons | |
| CN116416395A (en) | Digital earth surface based on point cloud compression flood inundation rapid assessment method of model | |
| CN118112685A (en) | Watershed hydrologic efficient high-precision forecasting method for non-data area | |
| CN115965243A (en) | Model chain for dynamic evaluation of flood risk under land use change | |
| Sekac et al. | Identifying potential sites for hydropower plant development in Busu catchment: Papua New Guinea | |
| CN111369102B (en) | Method and device for extracting waterlogging risk points | |
| Xiaohui et al. | General catchment delineation method and its application into the Middle Route Project of China’s south-to-north water diversion | |
| Ioannou et al. | Geoinformatics as a tool for the application of energy policy | |
| KR102341554B1 (en) | System and method for analysis of dam site suitability | |
| Jacob et al. | Hydraulic Structures Design Data Provided by GIS Tools and Hydrologic Modelling–the Case of Aljezur Basin | |
| Faouzi | Hydropower forecast in Ethiopia using the HBV model | |
| Faouzi | Hydropower forecast in Ethiopia using the HBV model-A case study in Omo-Gibe river basin |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination |