CN111898831B - Real-time flood probability forecasting practical method - Google Patents
Real-time flood probability forecasting practical method Download PDFInfo
- Publication number
- CN111898831B CN111898831B CN202010781843.5A CN202010781843A CN111898831B CN 111898831 B CN111898831 B CN 111898831B CN 202010781843 A CN202010781843 A CN 202010781843A CN 111898831 B CN111898831 B CN 111898831B
- Authority
- CN
- China
- Prior art keywords
- forecasting
- forecast
- flood
- probability
- error
- 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.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
- Y02A10/00—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE at coastal zones; at river basins
- Y02A10/40—Controlling or monitoring, e.g. of flood or hurricane; Forecasting, e.g. risk assessment or mapping
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Human Resources & Organizations (AREA)
- Economics (AREA)
- Strategic Management (AREA)
- Marketing (AREA)
- Game Theory and Decision Science (AREA)
- Entrepreneurship & Innovation (AREA)
- Development Economics (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Tourism & Hospitality (AREA)
- Physics & Mathematics (AREA)
- General Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention provides a practical method for forecasting real-time flood probability, which establishes a historical sample space by collecting forecast section hydrological element actual condition and certainty forecast data, and constructs a flood probability forecasting model base based on multi-method comparison of a nonparametric estimation method, an error distribution dynamic parameter inference method and the like, can consider forecast error differences of classification information such as different forecast periods, flow magnitude, rainfall distribution and the like, improves forecasting precision, is simple and intuitive, and is convenient to popularize and apply in production practice; the deterministic forecasting process manufactured according to the same forecasting time is used as input, a probability forecasting model corresponding to classification conditions is selected, an initial probability forecasting value of each time in the future flood process is calculated, a water quantity correction coefficient method is provided to correct the probability forecasting value of each time in the flood process line, the continuity and the correlation of hydrological factors in the same flood process are fully considered, and the performance of a flood probability forecasting interval can be effectively improved.
Description
Technical Field
The invention relates to the technical field of hydrologic forecasting, in particular to a practical method for forecasting real-time flood probability.
Background
At present, flood forecasting widely manufactured and used in the fields of domestic drainage basin water resource management, flood control and disaster reduction, flood resource utilization and the like is mostly deterministic, is output to users in a form of determining a single numerical value, is easy to understand formally, ignores uncertainty existing in a hydrologic forecasting process, and cannot provide more risk information. The probability forecasting method describes uncertainty of forecasting results by depicting forecasting quantities by using probability distribution, so that a decision maker can better evaluate risks, and the probability forecasting method becomes a research hotspot in the current hydrologic forecasting field.
Foreign probability forecasting services are developed based on an ensemble probability forecasting method, the calculated amount is huge, the timeliness is not high, the product form is only an objectively calculated probability forecasting interval, on one hand, the experience of a forecaster is not utilized, and on the other hand, the main uncertainty source of the faced time is not fully reflected. The domestic research on probability forecasting still stays at the scientific research stage, no examples of putting into production practice exist, and a real-time flood probability forecasting business complete method system based on the existing deterministic forecasting is lacked.
Disclosure of Invention
The invention aims to provide a practical method for forecasting the real-time flood probability aiming at the defects of the prior art, and converts scientific research into operation forecasting practice.
In order to achieve the purpose, the invention adopts the following technical scheme:
a real-time flood probability forecasting practical method comprises the following steps:
s1, collecting historical data and establishing a field flood historical sample space;
s2, according to the data of the historical samples, constructing and calibrating a flood probability forecasting model based on a forecasting error analysis method;
calculating a deterministic flood forecast result;
selecting the corresponding probability forecasting model according to the deterministic forecasting result and taking the deterministic forecasting result as input, and calculating an initial probability forecasting value of the element to be forecasted at each moment in the future flood process;
s3, correcting the initial probability forecast value to obtain a probability forecast interval of the flood process;
s4, calculating to obtain a typical forecasting process line group considering the main uncertainty source of the moment;
and S5, integrating the calculated deterministic flood forecast result and the various forecast information in the step S3 and the step S4, and finishing the making and issuing of the flood early warning information according to flood early warning issuing standards of different levels.
Further, the step S1 includes the steps of:
s11, collecting historical data, wherein the historical data comprises forecast section hydrological element live and certainty forecast data; the forecast section hydrological factor actual condition comprises actual condition data of flow, hydrology and other factors to be forecasted; the deterministic forecast data comprises forecast results of different forecast periods obtained using different forecasting methods;
s12, establishing a field flood historical sample space, selecting a relative error or an absolute error as a processing mode of forecasting errors according to error characteristics of forecasting hydrological elements, and then calculating to obtain a forecasting error sequence X by taking a live forecasting value sequence and a deterministic forecasting value sequence as a basis:
χ t ＝m t -y t or
In the formula: y is t And m t Respectively representing live and deterministic forecast values of the elements to be forecasted, then Y t0 And M t0 The method can respectively represent a live series and a deterministic forecast series of hydrological elements in a historical space; n is the sample length.
Further, the S2 adopts a flood probability forecasting model construction and calibration based on a non-parameter estimation method (NPEFM) or an error distribution dynamic parameter inference method (EDDPFM) to realize different forecast periods, flow levels and forecast rainfall classifications, wherein:
the probability prediction method of the non-parametric estimation method can be described as: inputting a historical error sample, respectively calculating various probability density functions of the prediction error by adopting a nonparametric estimation method based on a histogram, rosenblatt estimation, parzen kernel estimation or nearest neighbor estimation, drawing a probability density curve graph, preferably fitting with a sample scattered point series by adopting a mesh estimation method to determine an optimal nonparametric estimation method, calculating an accumulated probability curve of the nonparametric estimation method, and finally obtaining a prediction error value array R corresponding to each probability;
the probability prediction method based on the error distribution dynamic parameter estimation can be described as follows: establishing a probability density function and a distribution function model base of common distribution in multiple hydrological analyses of indexes, gamma, generalized normal, generalized logic, generalized extreme values, generalized pareto, gunn Bell, pearson type III, kappa, wecker and Logistic; constructing a parameter estimation method library based on L-moment and maximum likelihood estimation; establishing a distribution function goodness-of-fit evaluation method library consisting of a K-S inspection method, a deviation square sum minimum criterion and a Chichi information criterion; the optimal distribution form of the forecasting errors is worked out, the posterior probability density of forecasting elements is deduced on the basis of the optimal distribution, the covariant relation between distribution parameters and a deterministic forecasting series is analyzed by using historical samples, and finally a flood probability forecasting model is constructed;
the method specifically comprises the following steps:
s201, calculating the optimal distribution of the prediction error series: inputting historical error samples, and calculating an optimal distribution function F (x) obeyed by an error sample series of elements to be forecasted through hypothesis testing based on different distributions; the hypothesis testing process comprises parameter estimation of different distributions and goodness-of-fit evaluation of different distribution functions, and an optimal distribution form is selected;
s202, deducing the posterior probability density of the forecast elements: based on the probability density calculation formula of the probability density function f (x) and the random variable function phi (x) of the distribution, a posterior probability density function g (y) of the live hydrological elements is obtained through derivation t |m t ,Y t0 ,M t0 )：
y t ＝Φ(x)
g(y t |m t ,Y t0 ,M t0 )＝f(Φ -1 (y t )|Y t0 ,M t0 )·|[Φ -1 (y t )]'|
S203, constructing a flood probability forecasting model: analyzing a function relation set theta = Ψ (X) of a distribution function parameter theta and a determined forecasting value X in a historical forecasting sample space, and substituting the function relation set theta = Ψ (X) into the formula to construct a flood probability forecasting model;
s204: quantifying different classification conditions of a forecast period, a flow level and rainfall, and selecting an error distribution dynamic parameter inference method to construct a flood probability forecast model of the forecast section under different classification conditions;
taking the flow as a forecasting element, a forecasting error form as a relative error, and an optimal distribution function of the flow as Logistic distribution, the distribution function expression is as follows:
x～L(μ,δ)
in the formula: mu is a position parameter, and delta is a scale parameter; the probability density function is:
if it is
And
the mean and variance of the error sample X, respectively, can be estimated to have expected values of:
when flood forecasting is carried out in different forecasting periods, the statistical laws of forecasting errors are different, and the following assumptions are made:
σ x ＝h 2 (M t0 )
in the formula: function h 1 (M t0 ) And h 2 (M t0 ) Not limited to a particular form, it may follow M t0 May or may not be changed;
making a deterministic prediction of value m at time t t Rear, flow rate y t Can be viewed as a function of the relative error random variable x:
the inverse function can be found:
and (4) calculating by using the formula to obtain a probability forecasting model by combining historical sample data.
Further, the step S2 of calculating a deterministic flood forecast result includes the steps of:
s211: collecting and organizing live water and rain condition monitoring information, flood control situation analysis and reservoir pre-dispatching information;
s212: completing the rainfall production in the forecast period;
s213: flood forecasting calculation of a river system where the forecasting section is located is completed;
s214: correcting flood forecast calculation results through interval interaction, reservoir flood diversion calculation and river calculation;
s215: and (4) confirming the flood forecasting calculation result by combining artificial experience correction, thereby obtaining a deterministic flood forecasting result.
Further, the initial probability prediction value in S2 is obtained by taking the deterministic prediction process made according to the same prediction time as an input, selecting a probability prediction model corresponding to the classification condition, and calculating for each prediction period time of the future flood process.
Further, the S3 is characterized in that the initial probability prediction value at each time is corrected by using a correction coefficient method to obtain a probability prediction interval result of the flood process line, and the specific steps are as follows:
s31: calculating the series of accumulated forecast errors in different forecast periods (depending on the maximum forecast period and the length of the flood-fighting process of a forecast object) of elements to be forecasted from a historical sample space according to the flow magnitude and rainfall classification;
s32: obtaining forecast error ratio r under each guarantee rate i by adopting nonparametric estimation i Setting different correction proportions p according to forecast error characteristics of different forecast periods j j When the confidence coefficient alpha is set, the correction coefficient k of each forecast time in the flood process can be calculated i,j ，
Wherein i =1,2,3, \8230;, n; n is the length of the guaranteed rate array; j =1,2,3, \8230;, m; m is the length of the forecast period.
S33: and multiplying the upper probability forecast line and the lower probability forecast line of the confidence interval by a correction coefficient respectively to obtain a flood probability forecast correction interval.
Further, the S4 is specifically operated to: analyzing main uncertain sources existing in the period of time, taking different conditions possibly occurring in the sources as different scenes of an input boundary, calculating according to the steps of making deterministic flood forecasting results to obtain typical forecasting process lines corresponding to the different scenes, quantifying the forecasting risk information of the typical forecasting process lines, and comparing with a probability forecasting interval and the deterministic forecasting process lines when a consultation makes a decision.
Further, in S5, if the probability that a certain element to be forecasted of the forecast section exceeds the early warning threshold value in the comprehensive study and judgment forecast period is greater than or equal to a certain set value, the probability flood early warning information of the corresponding level is issued according to the flood early warning issuing standards of different levels.
The invention provides a flood probability forecasting model base based on nonparametric estimation and error distribution dynamic parameter calculation multi-method comparison, which can consider forecasting error differences of different forecasting periods, flow magnitude, rainfall and other classifications, improve forecasting precision, is simple and intuitive and is convenient to popularize in production practice;
the water quantity correction coefficient method is provided for correcting the probability forecast value of each time of the flood process line, the continuity and the correlation of hydrologic elements in the same flood process are considered, and the performance of a flood probability forecast interval can be effectively improved;
the method designs a business process and a product form of real-time flood probability forecasting, integrates a complete technical system capable of comprehensively studying and judging operation forecasting of various forecasting information such as deterministic forecasting, probability forecasting intervals, typical forecasting process line groups and the like, and provides a practical method for converting flood probability forecasting from scientific research to practical application.
Drawings
Fig. 1 is a flowchart of a real-time flood probability forecasting practical method according to an embodiment of the present invention;
FIG. 2 is a schematic diagram of a real-time flood probability forecasting product form according to the present invention;
fig. 3 (a) is a schematic diagram of a real-time flood probability forecasting result corresponding to different forecasting moments (7 months and 9 days) according to the embodiment of the present invention;
fig. 3 (b) is a schematic diagram of a real-time flood probability forecasting result corresponding to different forecasting moments (7 months and 10 days) according to the embodiment of the present invention;
fig. 3 (c) is a schematic diagram of a real-time flood probability forecasting result corresponding to different forecasting moments (7 months and 11 days) according to the embodiment of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is further described in detail below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are merely illustrative of the invention and do not limit the invention.
< example >
As shown in fig. 1, the real-time flood probability forecasting business method provided by this embodiment specifically includes the following steps;
step 1, collecting historical data to establish a historical sample space;
this step includes two substeps:
1.1 Collection of live and deterministic forecast data
The live data collected in the embodiment is data of flood forecasting of the warehousing flow of the three gorges reservoir, and is obtained from the water condition flood forecasting database.
The collected deterministic forecast data in the embodiment refers to 1-5 d forecast period three gorges reservoir warehousing flow deterministic forecast data issued by the ministry of hydrology of Yangtze river, and is acquired from a water regime forecast database of the ministry of hydrology of Yangtze river.
1.2 building historical sample space
And calculating relative errors according to the live flow and the deterministic forecast flow data to obtain a forecast error sample series. In the present embodiment, a classification prediction error historical sample space is established according to different forecast periods (1 d, 2d, 3d, 4d, 5 d).
Step 2, according to the historical sample data in the step 1, adopting a probability forecasting method based on error distribution dynamic parameter calculation to construct a flood probability forecasting model, wherein the step comprises three substeps:
2.1 optimal distribution of prediction error series
Historical error samples are input, and an optimal distribution function F (x) obeyed by the error sample series is obtained through hypothesis testing.
In the embodiment, 11 common distributions (including EXP, GAM, GNO, NOR, GLO, GEV, GPA, GUM, pliiii, KAP, and LOG) are selected for hypothesis testing, an L-moment method is adopted to estimate parameters of the distributions, a least squares of deviations (OLS) criterion and an akachi-pool information criterion (AIC) are adopted to evaluate the goodness of fit of the distribution functions, and the Logistic distribution function with the least OLS and AIC is selected as the optimal distribution form obeyed by the error sequence X.
TABLE 1 fitting test result of forecast error distribution function for three gorges reservoir warehousing
2.2 construction of flood probability forecasting model
The prior distribution of the three gorges reservoir warehousing flow relative error sequence X can be approximate to Logistic distribution:
x～L(μ,δ)
in the formula: mu is a position parameter and delta is a scale parameter. The probability density function is:
if it is
And
the mean and variance of the error sample X, respectively, can be estimated to have expected values of:
when flood forecasting is carried out in different forecasting periods, the statistical laws of forecasting errors are different, and the following assumptions are made:
σ x ＝h 2 (M t0 )
in the formula: function h 1 (M t0 ) And h 2 (M t0 ) Not limited to a particular form, it may follow M t0 The number of the terminal may be changed or not.
Making a deterministic prediction of value m at time t t Rear, flow rate y t Can be viewed as a function of the relative error random variable x:
the inverse function can be found:
obtaining a probability forecasting model of the 1-5 d forecast period three gorges reservoir warehousing flow by using the historical sample data in the step 1 and establishing the above formula, wherein a function h is used 1 (M t0 ) And h 2 (M t0 ) The results are shown in Table 2:
TABLE 2 mean and variance under different forecast periods
Step 3, making a deterministic flood forecasting result;
in the specific implementation, by collecting and mastering live rainfall monitoring, flood control situation analysis, reservoir pre-scheduling information and the like, and sequentially realizing the steps of forecast period rainfall making, river system automatic calculation, interval interaction and reservoir flood control calculation, river calculation, artificial experience correction confirmation and the like, a deterministic flood forecasting process of a 5d forecast period is made by taking a Yangtze river flood control forecasting and scheduling system as a tool and taking 7-month 9 days, 7-month 10 days and 7-month 11 days in 2018.7 flood process as forecasting basis time respectively, and the deterministic flood forecasting process is shown in fig. 3 (a) to (c).
And 4, taking the deterministic forecasting result of the step 3 as input, selecting 1-5 d forecasting periods corresponding to the three gorges reservoir warehousing flow probability forecasting model result of the step 2, and calculating an initial probability forecasting flow value at each forecasting period time in the future flood process.
Step 5, correcting the initial probability forecasting result of the step 4 to obtain a probability forecasting interval of the flood process;
in the specific implementation, the initial probability prediction value at each moment is corrected by adopting a water volume correction coefficient method, so that the probability prediction interval result of the flood process line is obtained. The specific description is as follows:
calculating the 5-day accumulated water quantity forecast error series in the warehousing flood process of the three gorges reservoir from the historical sample space, and obtaining the water quantity error ratio r under each guarantee rate i by adopting non-parameter estimation i Setting different correction proportions p according to forecast error characteristics of different forecast periods j j When the confidence coefficient alpha is set, the correction coefficient k of each forecast time in the flood process can be calculated i,j ，
Wherein i =1,2,3, \8230;, n; n is the length of the guaranteed rate array; j =1,2,3, \8230;, m.
And (4) multiplying the upper probability forecast line and the lower probability forecast line of the confidence interval by a correction coefficient respectively to obtain a flood probability forecast correction interval, which is shown in the steps (a) to (c) of fig. 3.
Step 6, calculating to obtain a typical forecasting process line group;
in the forecasting example, the main uncertainty source existing in the time interval is whether the pavilion mouth water reservoir is flooded or not, possible pavilion mouth water reservoir flooding conditions are constructed to serve as input boundary scenes, a typical forecasting process line (see figure 3) considering the pavilion mouth flooding is obtained through calculation according to the deterministic forecasting making steps, and the typical forecasting process line is compared with a probability forecasting interval and the deterministic forecasting process line when a consultation is made to decide.
Step 7, integrating various forecast information in the steps 3, 5 and 6, predicting that the probability of 90% of the warehousing flow of the three gorges reservoir exceeds 50000m in 5 days in the future in 7-month and 9-2018-7-month and 10-2018-7-month 3 And/s, issuing blue early warning of flood in three gorges reservoir area at the upstream of the Yangtze river according to the 'Yangtze river water condition early warning issuing management method (trial)'; in 2018, 7, 11, the forecast shows that the probability of 90% of the warehousing flow of the three gorges reservoir exceeds 56700m in the 5 days in the future 3 And/s, issuing flood yellow early warning in the three gorges reservoir area at the upper part of the Yangtze river.
In addition, for different drainage basins and different forecasting cross sections, the error distribution of the forecasting object may be more similar to other distribution forms, such as exponential distribution (EXP), gamma distribution (GAM), normal distribution (NOR), generalized extreme value distribution (GEV), gunbell distribution (GUM), generalized pareto distribution (GPA), pearson type iii distribution (ppii), logistic distribution (LOG), etc., which are commonly used in hydrology, and when the method is applied specifically, the distribution form most suitable for the probability distribution of the forecasting error sequence X may be preferably selected by comparison, and the distribution parameters and the applicability are shown in table 3.
TABLE 3 hydrologic statistics common distribution function
In summary, the invention provides a flood probability forecasting model base based on nonparametric estimation and error distribution dynamic parameter calculation multi-method, which can consider forecasting error differences of different forecasting periods, flow magnitude, rainfall and other classifications, improve forecasting precision, is simple and intuitive, and is convenient to popularize in production practice; the provided water volume correction coefficient method corrects probability forecast values of flood process lines at all times, considers the continuity and the correlation of hydrological elements in the same flood process, and can improve the performance of an uncertain flood forecast interval. In a word, the business process and the product form of real-time flood probability forecasting are designed originally, a complete technical system capable of comprehensively researching and judging operation forecasting of various forecasting information such as deterministic forecasting, probability forecasting intervals, typical forecasting process line sets and the like is integrated, and a practical method for converting flood probability forecasting from scientific research to practical application is provided.
Establishing a historical sample space by collecting forecast section hydrological element actual condition and certainty forecast data; establishing probability forecasting models of different forecast periods, flow magnitudes and forecast rainfall classifications by adopting a non-parameter estimation method or an error distribution dynamic parameter deduction method; obtaining a deterministic flood forecasting result according to a conventional process; taking a deterministic forecasting process made at the same forecasting moment as input, selecting a probability forecasting model corresponding to classification conditions, calculating an initial probability forecasting value at each moment in the future flood process, and correcting by adopting a water yield correction coefficient method to obtain a probability forecasting interval result; analyzing main uncertainty sources existing in an faced time interval, and calculating to obtain a typical forecasting process line group comprising independent risk information; comprehensively studying and judging various forecast information such as deterministic forecast, probability forecast intervals, typical forecast process line groups and the like, and issuing corresponding probability flood early warning information if the probability that the flow rate or the water level of a certain main section exceeds an early warning threshold value in a forecast period is more than or equal to a certain set value according to various flood early warning issuing standards.
The above-mentioned embodiments only express the embodiments of the present invention, and the description thereof is more specific and detailed, but not construed as limiting the scope of the present invention. It should be noted that various changes and modifications can be made by those skilled in the art without departing from the spirit of the invention, and these changes and modifications are all within the scope of the invention. Therefore, the protection scope of the present patent shall be subject to the appended claims.
Claims (8)
1. A real-time flood probability forecasting practical method is characterized by comprising the following steps:
s1, collecting historical data and establishing a field flood historical sample space;
s2, according to the data of the historical samples, constructing and calibrating a flood probability forecasting model based on a forecasting error analysis method; calculating a deterministic flood forecast result;
selecting the corresponding probability forecasting model according to the deterministic flood forecasting result by taking the deterministic flood forecasting result as input, and calculating an initial probability forecasting value of the elements to be forecasted at each moment in the future flood process;
a flood probability forecasting model base based on non-parameter estimation and error distribution dynamic parameter calculation multi-method comparison is adopted, forecasting error differences of different forecasting periods, flow magnitude and rainfall classification are considered, and forecasting precision is improved;
s3, correcting the initial probability forecast value to obtain a probability forecast interval of the flood process;
s4, calculating to obtain a typical forecasting process line group considering the main uncertainty source of the moment;
and S5, integrating the calculated deterministic flood forecast result and the various forecast information in the step S3 and the step S4, and finishing the making and issuing of the flood early warning information according to flood early warning issuing standards of different levels.
2. The method of claim 1, wherein the step of collecting historical data and establishing a historical sample space for flood sessions comprises the steps of:
s11, collecting historical data, wherein the historical data comprises forecast section hydrological element live and certainty forecast data; the forecast section hydrological element live condition comprises live condition data of flow, hydrology and other elements to be forecasted; the deterministic forecast data comprises forecast results of different forecast periods obtained using different forecasting methods;
s12, establishing a scene flood historical sample space, selecting a relative error or an absolute error as a processing mode of forecasting errors according to the error characteristics of forecasting hydrological elements, and then calculating to obtain a forecasting error sequence X by taking a live forecast value sequence and a deterministic forecast value sequence as a basis:
χ t ＝m t -y t or
In the formula: y is t And m t Respectively representing live and deterministic forecast values of the elements to be forecasted, then Y t0 And M t0 The method can respectively represent a live series and a deterministic forecast series of hydrological elements in a historical space; n is the sample length.
3. The method of claim 2, wherein the method comprises: the construction and calibration of the flood probability forecasting model adopt a non-parameter estimation method or an error distribution dynamic parameter inference method to realize different forecast periods, flow levels and forecast rainfall classification, wherein,
the probability prediction method of the non-parametric estimation method can be described as: inputting a historical error sample, respectively calculating various probability density functions of the prediction error by adopting a nonparametric estimation method based on a histogram, rosenblatt estimation, parzen kernel estimation or nearest neighbor estimation, drawing a probability density curve graph, preferably fitting with a sample scattered point series by adopting a mesh estimation method to determine an optimal nonparametric estimation method, calculating an accumulated probability curve of the nonparametric estimation method, and finally obtaining a prediction error value array R corresponding to each probability;
the probability prediction method based on the error distribution dynamic parameter estimation can be described as follows: establishing a probability density function and a distribution function model base of common distribution in various hydrological analyses of indexes, gamma, generalized normality, generalized logic, generalized extreme values, generalized pareto, gunn Bell, pearson type III, kappa, wecker and Logistic; constructing a parameter estimation method library based on L-moment and maximum likelihood estimation; establishing a distribution function goodness-of-fit evaluation method library consisting of a K-S inspection method, a deviation square sum minimum criterion and a Chichi information criterion; the optimal distribution form of the forecasting errors is calculated, the posterior probability density of the forecasting elements is deduced based on the optimal distribution, the covariant relation between the distribution parameters and the deterministic forecasting series is analyzed by using historical samples, and finally a flood probability forecasting model is constructed;
the method specifically comprises the following steps:
s201, calculating the optimal distribution of the prediction error series: inputting historical error samples, and calculating an optimal distribution function F (x) obeyed by an error sample series of elements to be forecasted through hypothesis testing based on different distributions; the hypothesis testing process comprises parameter estimation of different distributions and goodness-of-fit evaluation of different distribution functions, and an optimal distribution form is selected;
s202, deducing the posterior probability density of the forecast elements: based on the probability density calculation formula of the probability density function f (x) and the random variable function phi (x) of the distribution, a posterior probability density function g (y) of the live hydrological elements is obtained through derivation t |m t ,Y t0 ,M t0 )：
y t ＝Φ(x)
g(y t |m t ,Y t0 ,M t0 )＝f(Φ -1 (y t )|Y t0 ,M t0 )·|[Φ -1 (y t )]'|
S203, constructing a flood probability forecasting model: analyzing a function relation set theta = Ψ (X) of a distribution function parameter theta and a determined forecasting value X in a historical forecasting sample space, and substituting the function relation set theta = Ψ (X) into the formula to construct a flood probability forecasting model;
s204: and quantifying different classification conditions of a forecast period, a flow level and rainfall, and selecting an error distribution dynamic parameter inference method to construct a flood probability forecast model of the forecast section under different classification conditions.
4. The method of claim 1, wherein the step of calculating the deterministic flood forecast result comprises the steps of:
s211: collecting and organizing live water and rain condition monitoring information, flood control situation analysis and reservoir pre-scheduling information;
s212: completing the rainfall manufacturing in the forecast period;
s213: flood forecasting calculation of a river system where the forecasting section is located is completed;
s214: correcting flood forecast calculation results through interval interaction, reservoir flood diversion calculation and river calculation;
s215: and confirming the flood forecast calculation result by combining artificial experience correction, thereby obtaining a deterministic flood forecast result.
5. The method of claim 1, wherein the method comprises: the initial probability prediction value is obtained by taking a deterministic prediction process made according to the same prediction time as input, selecting a probability prediction model corresponding to classification conditions and calculating each prediction period moment of the future flood process.
6. The method of claim 1, wherein the method comprises: correcting the initial probability forecast value at each moment by adopting a correction coefficient method to obtain the probability forecast interval result of the flood process line, and specifically comprising the following steps of:
s31: calculating the series of accumulated forecast errors of elements to be forecasted in different forecast periods from a historical sample space by flow magnitude and rainfall classification;
s32: obtaining forecast error ratio r under each guarantee rate i by adopting nonparametric estimation i Setting different correction proportions p according to forecast error characteristics of different forecast periods j j When the confidence coefficient alpha is set, the correction coefficient k of each forecast time in the flood process can be calculated i,j ，
Wherein i =1,2,3, \8230;, n; n is the length of the guaranteed rate array; j =1,2,3, \ 8230;, m; m is the length of a forecast period;
s33: and multiplying the upper probability forecasting line and the lower probability forecasting line of the confidence interval by a correction coefficient respectively to obtain a flood probability forecasting correction interval.
7. The method of claim 1, wherein the step S4 comprises the following steps: analyzing main uncertain sources existing in the period of time, taking different conditions possibly occurring in the sources as different scenes of an input boundary, calculating typical forecasting process lines corresponding to the different scenes according to the steps of making deterministic flood forecasting results, quantifying forecasting risk information of the typical forecasting process lines, and comparing the forecasting process lines with a probability forecasting interval and the deterministic forecasting process lines when a consultation makes a decision.
8. The method as claimed in claim 1, wherein in step S5, if the probability that the forecast threshold value of a forecast section exceeds a preset value is greater than or equal to a preset value, then the method issues the probability flood early warning information of the corresponding level according to the flood early warning issue criteria of different levels.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010781843.5A CN111898831B (en) | 2020-08-06 | 2020-08-06 | Real-time flood probability forecasting practical method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010781843.5A CN111898831B (en) | 2020-08-06 | 2020-08-06 | Real-time flood probability forecasting practical method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111898831A CN111898831A (en) | 2020-11-06 |
| CN111898831B true CN111898831B (en) | 2022-11-08 |
Family
ID=73247217
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010781843.5A Active CN111898831B (en) | 2020-08-06 | 2020-08-06 | Real-time flood probability forecasting practical method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111898831B (en) |
Families Citing this family (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112415635B (en) * | 2020-11-19 | 2022-03-29 | 中山大学 | Gamma and Gaussian distribution coupled monthly scale rainfall forecast correction method |
| CN112732390A (en) * | 2021-01-19 | 2021-04-30 | 广东省水文局韶关水文分局 | Man-machine cooperative interaction method for flood process line forecast |
| CN114169590A (en) * | 2021-11-22 | 2022-03-11 | 华中科技大学 | Reservoir warehousing runoff forecasting and correcting method and system based on multi-scenario division |
| CN115795919B (en) * | 2023-01-31 | 2023-06-27 | 中国水利水电科学研究院 | Method for generating and adapting flood frequency curve |
| CN116011687B (en) * | 2023-03-30 | 2023-08-11 | 山东锋士信息技术有限公司 | Flood forecasting method, system and medium based on Copula function |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101864750A (en) * | 2010-06-29 | 2010-10-20 | 西安理工大学 | Multi-model meta-synthesis flood forecasting system and forecasting method thereof |
| CN108416468A (en) * | 2018-02-08 | 2018-08-17 | 河海大学 | A kind of flood advanced early warning forecasting procedure |
| CN109993372A (en) * | 2019-04-12 | 2019-07-09 | 淮河水利委员会水文局(信息中心) | One kind being based on the probabilistic flood probability forecasting procedure of multi-source |
| CN110610256A (en) * | 2019-08-02 | 2019-12-24 | 华中科技大学 | Runoff forecast level evaluation method considering forecast difficulty under different forecast situations |
| CN111104750A (en) * | 2019-12-24 | 2020-05-05 | 西安理工大学 | Reservoir dispatching diagram drawing method of coupled ensemble prediction information |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20140156232A1 (en) * | 2012-11-30 | 2014-06-05 | International Business Machines Corporation | Methods, systems and computer program storage devices for generating a flooding forecast |
-
2020
- 2020-08-06 CN CN202010781843.5A patent/CN111898831B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101864750A (en) * | 2010-06-29 | 2010-10-20 | 西安理工大学 | Multi-model meta-synthesis flood forecasting system and forecasting method thereof |
| CN108416468A (en) * | 2018-02-08 | 2018-08-17 | 河海大学 | A kind of flood advanced early warning forecasting procedure |
| CN109993372A (en) * | 2019-04-12 | 2019-07-09 | 淮河水利委员会水文局(信息中心) | One kind being based on the probabilistic flood probability forecasting procedure of multi-source |
| CN110610256A (en) * | 2019-08-02 | 2019-12-24 | 华中科技大学 | Runoff forecast level evaluation method considering forecast difficulty under different forecast situations |
| CN111104750A (en) * | 2019-12-24 | 2020-05-05 | 西安理工大学 | Reservoir dispatching diagram drawing method of coupled ensemble prediction information |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111898831A (en) | 2020-11-06 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111898831B (en) | Real-time flood probability forecasting practical method | |
| CN109272146B (en) | Flood prediction method based on deep learning model and BP neural network correction | |
| CN109472403B (en) | Medium-and-long-term runoff forecasting method integrating empirical mode decomposition and remote correlation | |
| CN109543147B (en) | Method for nonlinear rapid diagnosis and attribution analysis of basin rainfall runoff relation | |
| CN107886160B (en) | BP neural network interval water demand prediction method | |
| CN110910004A (en) | Reservoir dispatching rule extraction method and system with multiple uncertainties | |
| CN112506990A (en) | Hydrological data anomaly detection method based on spatiotemporal information | |
| LU501097B1 (en) | An estimation method of regional available water supply amount with considering the joint probability distributions of multiple water sources | |
| Mediero et al. | Probabilistic calibration of a distributed hydrological model for flood forecasting | |
| CN113095694B (en) | Rainfall sand transportation model construction method suitable for multiple landform type areas | |
| CN115497272A (en) | Construction period intelligent early warning system and method based on digital construction | |
| CN109657287B (en) | Hydrological model precision identification method based on comprehensive scoring method | |
| CN109376937B (en) | Self-adaptive scheduling end-of-term water level prediction method based on ensemble empirical mode decomposition | |
| CN115907286A (en) | Social and economic drought assessment method | |
| CN109299853B (en) | Reservoir dispatching function extraction method based on joint probability distribution | |
| CN114548493A (en) | Method and system for predicting current overload of electric energy meter | |
| CN112711896B (en) | Complex reservoir group optimal scheduling method considering multi-source forecast error uncertainty | |
| CN113538063A (en) | Electricity charge abnormal data analysis method, device, equipment and medium based on decision tree | |
| CN111898097B (en) | Ecological flow determination method combining probability density and guarantee rate | |
| Maposa et al. | Modelling extreme flood heights in the lower Limpopo River basin of Mozambique using a time-heterogeneous generalised Pareto distribution | |
| CN117171128A (en) | Aquatic organism protection threshold identification method based on four-water coupling model | |
| Madsen | Automatic calibration and uncertainty assessment in rainfall-runoff modelling | |
| CN110009132A (en) | A kind of short-term electric load fining prediction technique based on LSTM deep neural network | |
| CN106557614B (en) | Flood frequency analysis method based on Halphen IB distribution | |
| Liu et al. | Parameter optimization and uncertainty assessment for rainfall frequency modeling using an adaptive Metropolis–Hastings algorithm |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |