CN113205155A - Multi-source precipitation data fusion method based on partition self-adaptive weight - Google Patents
Multi-source precipitation data fusion method based on partition self-adaptive weight Download PDFInfo
- Publication number
- CN113205155A CN113205155A CN202110583497.4A CN202110583497A CN113205155A CN 113205155 A CN113205155 A CN 113205155A CN 202110583497 A CN202110583497 A CN 202110583497A CN 113205155 A CN113205155 A CN 113205155A
- Authority
- CN
- China
- Prior art keywords
- data
- precipitation data
- satellite
- precipitation
- weight
- 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
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/25—Fusion techniques
- G06F18/254—Fusion techniques of classification results, e.g. of results related to same input data
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01W—METEOROLOGY
- G01W1/00—Meteorology
- G01W1/10—Devices for predicting weather conditions
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/24—Querying
- G06F16/245—Query processing
- G06F16/2458—Special types of queries, e.g. statistical queries, fuzzy queries or distributed queries
- G06F16/2465—Query processing support for facilitating data mining operations in structured databases
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/12—Computing arrangements based on biological models using genetic models
- G06N3/126—Evolutionary algorithms, e.g. genetic algorithms or genetic programming
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Data Mining & Analysis (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Biophysics (AREA)
- Evolutionary Biology (AREA)
- Health & Medical Sciences (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Environmental & Geological Engineering (AREA)
- Software Systems (AREA)
- Databases & Information Systems (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Computation (AREA)
- Mathematical Physics (AREA)
- Computational Linguistics (AREA)
- Atmospheric Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Computing Systems (AREA)
- Biomedical Technology (AREA)
- Genetics & Genomics (AREA)
- Physiology (AREA)
- Molecular Biology (AREA)
- Biodiversity & Conservation Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Environmental Sciences (AREA)
- Ecology (AREA)
- Fuzzy Systems (AREA)
- Probability & Statistics with Applications (AREA)
- Radio Relay Systems (AREA)
Abstract
The invention discloses a multi-source precipitation data fusion method based on partition self-adaptive weight, which comprises the following steps of S1, respectively preprocessing satellite precipitation data and ground station observation precipitation data; s2, evaluating the precision of the preprocessed satellite precipitation data, and screening out the satellite precipitation data meeting the precision evaluation requirement; s3, performing fusion correction on the satellite precipitation data meeting the precision evaluation requirement and the ground station observation precipitation data based on a partition self-adaptive weight fusion method to obtain the optimal weight of each level of satellite precipitation data; and S4, making a weight self-lookup table based on the optimal weight of each level of satellite precipitation data. The method has the advantages that the method integrates the advantages of high spatial resolution of satellite precipitation data and high precision of ground station observation data, properly eliminates the disadvantages of the satellite precipitation data and the ground station observation data, and makes up for the deficiencies of the two data to a greater extent through a reverse solving mode, so that precipitation data with high accuracy and high applicability are obtained.
Description
Technical Field
The invention relates to the technical field of satellite precipitation inversion and fusion, in particular to a multi-source precipitation data fusion method based on partition self-adaptive weight.
Background
The single precipitation data cannot estimate the space-time precipitation with higher basin expansion precision, so that the satellite precipitation data with wider coverage range and larger error and the ground station observation data with sparse distribution and higher precision can be fused by making up for the deficiencies. Methods for fusing precipitation data of satellites and rainfall stations can be generally divided into two main categories of global correction and local correction. Wherein, the global correction method comprises an average deviation correction method, a linear regression method, a dual-core smoothing method and the like; the local correction method mainly comprises collaborative kriging, geographic weighted regression, Bayesian fusion and the like. However, the difference of precipitation in spatial distribution and time is not considered in the common global correction method, the traditional local correction method focuses on the spatial autocorrelation of precipitation and is suitable for the places with dense rainfall stations, and in the regions with sparse rainfall stations, the spatial autocorrelation is overestimated, so that the fusion result generates great uncertainty.
In most areas of China, the terrain is complex, rainfall stations are difficult to arrange, and therefore rainfall data are lacked, so that the method has limitations, how to further improve the precision of fused data, and the problem that the rainfall data with wide applicability and high space-time resolution becomes a great concern at present is formed. In addition, the satellite precipitation data is influenced by a time scale, how to fully consider the influence of time and space in fusion correction, and how to establish a set of method in a data-lacking area to obtain precipitation data with higher precision becomes a problem to be solved in the current research.
Disclosure of Invention
The invention aims to provide a multi-source precipitation data fusion method based on partition self-adaptive weight, so that the problems in the prior art are solved.
In order to achieve the purpose, the technical scheme adopted by the invention is as follows:
a multi-source precipitation data fusion method based on partition self-adaptive weight comprises the following steps,
s1, preprocessing the satellite precipitation data and the ground station observation precipitation data respectively;
s2, evaluating the precision of the preprocessed satellite precipitation data, and screening out the satellite precipitation data meeting the precision evaluation requirement;
s3, performing fusion correction on the satellite precipitation data meeting the precision evaluation requirement and the ground station observation precipitation data based on a partition self-adaptive weight fusion method to obtain the optimal weight of each level of satellite precipitation data;
and S4, making a weight self-lookup table based on the optimal weight of each level of satellite precipitation data.
Preferably, step S1 specifically includes the following steps,
preprocessing satellite precipitation data: reading various satellite precipitation data in the file in batches by using a programming language, carrying out format conversion, and unifying the time and space resolution of the various satellite precipitation data; acquiring daily rainfall of various satellite rainfall data in longitude and latitude;
preprocessing the observation precipitation data of the ground station: checking the missing condition of each ground station data, and if the data is seriously missing, removing the ground station data; if the data is slightly missing, the interpolation processing is performed on the ground point data.
Preferably, step S2 specifically includes the following steps,
s21, performing space-time characteristic analysis on the satellite precipitation data and the ground station observation data based on the DEM elevation data, and judging whether the analysis results of the satellite precipitation data and the ground station precipitation data are similar or not, wherein the closer the results of the satellite precipitation data and the ground station precipitation data are, the higher the precision of the corresponding satellite precipitation data is;
s22, respectively adopting correlation coefficients, relative deviation, root mean square error and average absolute error to evaluate the consistency of the satellite precipitation data, and simultaneously respectively using detection rate, false alarm rate and critical determination index to evaluate the detection capability of the satellite precipitation data so as to screen out the satellite precipitation data meeting the evaluation requirement; the consistency assessment and the detectability assessment each include daily, monthly, quarterly, and yearly scale assessments.
Preferably, the consistency evaluation specifically includes calculating each evaluation index, including a correlation coefficient, a relative deviation, a root mean square error and an average absolute error, respectively, and the closer the correlation coefficient is to 1, the closer the other three errors are to 0, the better the consistency evaluation result is represented; the calculation formula of each evaluation index is as follows
Wherein R is a correlation coefficient; BIAS is relative deviation; RMSE is root mean square error; MAE is the mean absolute error; siFor the ith satellite precipitation data, PiObserving precipitation data for the ith ground station; 1,2,3, …, n; and n is the total number of the satellite precipitation data.
Preferably, the detection capability assessment method comprises the steps of sequentially calculating a detection rate, a false alarm rate and a critical determination index, wherein the calculation formulas are respectively as follows,
h is the number of precipitation events accurately detected by the satellite precipitation data; f is the number of false precipitation events; m is the number of missed precipitation events; the POD is used for measuring the detection capability of the rainfall event, the value range is [0,1], and the higher the value of the POD is, the higher the successful detection degree of the satellite rainfall data to the rainfall event is; the FAR is used for measuring the probability of the precipitation event being mispredicted, the value range is [0,1], and the smaller the value of the FAR is, the smaller the false alarm degree of the satellite precipitation data to precipitation is; the CSI is the proportion of the total number of precipitation events correctly detected by the satellite precipitation data to the total number of the events, the characteristics of the satellite precipitation data set can be comprehensively reflected, the value range is [0,1], the optimal value is 1, and the higher the value of the CSI is, the better the detection capability evaluation effect is.
Preferably, step S3 specifically includes the following steps,
s31, classifying the daily precipitation data according to the provisions of the meteorological bureau in a grading way, wherein the daily precipitation data comprise light rain, medium rain, heavy rainstorm and extra heavy rainstorm;
s32, solving the optimal weight of each level of precipitation month by adopting a genetic algorithm;
s33, judging whether the optimal weights meet the error requirements, if so, entering the step S4; otherwise, return to step S32.
Preferably, step S32 specifically includes the following steps,
s321, for each level of monthly rainfall, determining the weight of the satellite rainfall data of the same level of the same month in the corresponding multi-source data fusion process to beA certain value is obtained; therefore, the space vector P is formed by observing precipitation data of any level of ground stations day by day0The space vectors formed by the corresponding satellite precipitation data and the ground station observation precipitation data are respectively P1，P2，…，PnThe weight of the corresponding satellite precipitation data is beta1，β2，…，βnThe error is epsilon; the following relationship is satisfied among the spatial vector, the weight and the error,
P0＝β1P1+β2P2+…+βnPn+ε
wherein, the premise of satisfying the optimal weight is that the error epsilon approaches to 0;
s322, setting a target function, solving by adopting a genetic algorithm, and searching for optimal weight beta 'of each level of satellite precipitation data'1，β′2，...，β′n(ii) a The objective function is as follows,
f＝min[P0-(β1P1+β2P2+…+βnPn)]。
preferably, in step S33, to avoid the situation that the genetic algorithm falls into the local optimal solution when solving the optimal weight of the satellite precipitation data, it is necessary to determine whether the optimal weight of the satellite precipitation data is the global optimal solution, and if yes, the process goes to step S4; if not, the process returns to step S32 until the optimal weight is the global optimal solution.
Preferably, step S5 is further included after step S4, and step S5 is to calculate the precipitation amount of the study area of the missing data by using a weight self-lookup table to supplement and improve the missing data of the study area; the specific process is that,
s51, performing basin partitioning on the research basin by using a known ground station as a base point and adopting a Thiessen polygon method; and the precipitation types of other ground stations in each subarea are specified to be the same as those of the known stations;
s52, sequentially extracting longitude and latitude coordinates of the satellite precipitation data according to the satellite precipitation data of the research basin, and simultaneously eliminating coordinates coincident with the positions of known ground stations;
s53, sequentially judging the partition of each coordinate;
s54, determining the precipitation level of each partition;
s55, sequentially comparing the weight self-lookup table, and sequentially calculating the precipitation amount at each defect position of the research area according to a data fusion formula, so as to supplement and perfect the defect data of the research area; the data fusion formula is as follows,
P0＝β′1P1+β′2P2+…+β′nPn。
preferably, in step S53, the coordinates of each satellite precipitation data in step S52 are sequentially compared with the boundary range of each partition, and the partition to which each coordinate belongs is determined, so that each satellite precipitation data is accurately partitioned.
The invention has the beneficial effects that: 1. the advantages of high spatial resolution of satellite precipitation data and high precision of ground station observation data are integrated, disadvantages of the satellite precipitation data and the ground station observation data are properly eliminated, and two kinds of data are made up for each other to a greater extent in a reverse solving mode, so that precipitation data with high accuracy and high applicability are obtained. 2. The advantages of the data mining technology in processing mass data are well exerted, the association rule analysis method is applied to the precipitation forecast of the area, and the accuracy of the forecast result can be further improved based on the hidden relation among the data.
Drawings
FIG. 1 is a schematic flow chart of a precipitation data fusion method according to an embodiment of the present invention;
FIG. 2 is a flow chart illustrating data preprocessing according to an embodiment of the present invention;
FIG. 3 is a schematic flow chart illustrating accuracy evaluation of satellite precipitation data according to an embodiment of the present invention;
FIG. 4 is a schematic flowchart illustrating fusion correction of precipitation data by the partition adaptive weight fusion method according to an embodiment of the present invention;
FIG. 5 is a schematic flow chart illustrating the calculation of precipitation using a weight self-lookup table according to an embodiment of the present invention;
fig. 6 is a schematic diagram of a result of performing basin partitioning on a lancanvasu basin by using the thiessen polygon method in 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 detailed description and specific examples, while indicating the invention, are intended for purposes of illustration only and are not intended to limit the scope of the invention.
Example one
In the embodiment, as shown in fig. 1, a multi-source precipitation data fusion method based on partition adaptive weights is provided, which includes the following steps,
s1, preprocessing the satellite precipitation data and the ground station observation precipitation data respectively;
s2, evaluating the precision of the preprocessed satellite precipitation data, and screening out the satellite precipitation data meeting the precision evaluation requirement;
s3, performing fusion correction on the satellite precipitation data meeting the precision evaluation requirement and the ground station observation precipitation data based on a partition self-adaptive weight fusion method to obtain the optimal weight of each level of satellite precipitation data;
and S4, making a weight self-lookup table based on the optimal weight of each level of satellite precipitation data.
In this embodiment, the fusion method provided by the present invention specifically includes five parts, which are respectively: data preprocessing, data precision evaluation, optimal weight solving, weight self-lookup table forming and weight self-lookup table application. The following is a detailed explanation of these five parts.
First, data preprocessing
This section corresponds to the content of step S1, and as shown in fig. 2, step S1 specifically includes the following content,
preprocessing satellite precipitation data: reading various satellite precipitation data in the file in batches by using a programming language, carrying out format conversion, and unifying the time and space resolution of the various satellite precipitation data; acquiring daily rainfall of various satellite rainfall data in longitude and latitude;
preprocessing the observation precipitation data of the ground station: checking the missing condition of each ground station data, and if the data is seriously missing, removing the ground station data; if the data is slightly missing, the interpolation processing is performed on the ground point data.
Second, data precision evaluation
This section corresponds to the content of step S2, and as shown in fig. 3, step S2 specifically includes the following content,
s21, performing space-time characteristic analysis on the satellite precipitation data and the ground station observation data based on the DEM elevation data, and judging whether the analysis results of the satellite precipitation data and the ground station precipitation data are similar or not, wherein the closer the results of the satellite precipitation data and the ground station precipitation data are, the higher the precision of the corresponding satellite precipitation data is; the space-time characteristic analysis comprises precipitation annual distribution characteristics, annual variation trend and space distribution characteristics;
s22, respectively adopting a correlation coefficient R, a relative deviation BIAS, a root mean square error RMSE and an average absolute error MAE to carry out consistency evaluation on the rainfall data of each satellite, and simultaneously respectively adopting a detection rate POD, a false alarm rate FAR and a critical determination index CSI to evaluate the detection capability of the rainfall data of each satellite so as to screen out the rainfall data of the satellite meeting the evaluation requirement; the consistency assessment and the detectability assessment each include daily, monthly, quarterly, and yearly scale assessments.
Specifically, the consistency evaluation is to calculate each evaluation index respectively, including a correlation coefficient, a relative deviation, a root mean square error and an average absolute error, wherein the closer the correlation coefficient is to 1, the closer the other three errors are to 0, the better the consistency evaluation result is represented; the calculation formula of each evaluation index is as follows
Wherein R is a correlation coefficient; BIAS is relative deviation; RMSE is root mean square error; MAE is the mean absolute error; siFor the ith satellite precipitation data, PiObserving precipitation data for the ith ground station; 1,2,3, …, n; and n is the total number of the satellite precipitation data.
The detection capability assessment specifically comprises sequentially calculating detection rate, false alarm rate and critical determination index, wherein the calculation formulas are respectively as follows,
h is the number of precipitation events accurately detected by the satellite precipitation data; f is the number of false precipitation events; m is the number of missed precipitation events; the POD is used for measuring the detection capability of the rainfall event, the value range is [0,1], and the higher the value of the POD is, the higher the successful detection degree of the satellite rainfall data to the rainfall event is; the FAR is used for measuring the probability of the precipitation event being mispredicted, the value range is [0,1], and the smaller the value of the FAR is, the smaller the false alarm degree of the satellite precipitation data to precipitation is; the CSI is the proportion of the total number of precipitation events correctly detected by the satellite precipitation data to the total number of the events, the characteristics of the satellite precipitation data set can be comprehensively reflected, the value range is [0,1], the optimal value is 1, and the higher the value of the CSI is, the better the detection capability evaluation effect is.
Solving for optimal weight
For each 0.1 x 0.1 grid in the data set, there is ground site observation data and satellite precipitation data that meets the accuracy assessment criteria, and the accuracy of the satellite precipitation data is typically affected by the precipitation intensity and season. The invention classifies the daily rainfall of the research area and solves the weight of the satellite rainfall data in the graded rainfall monthly.
This section corresponds to the content of step S3, and as shown in fig. 4, step S3 specifically includes the following content,
s31, classifying the precipitation data by grades, including light rain, medium rain, heavy rainstorm and extra heavy rainstorm, according to the regulations of the weather bureau (according to precipitation intensity grade division standards issued by the national weather bureau);
s32, solving the optimal weight of each level of precipitation month by adopting a genetic algorithm;
s33, judging whether the optimal weights meet the error requirements, if so, entering the step S4; otherwise, return to step S32.
The step S32 specifically includes the following contents,
s321, regarding every level of monthly rainfall, determining the weight of the satellite rainfall data of the same level of the same month in the corresponding multi-source data fusion process as a certain value; therefore, the space vector P is formed by observing precipitation data of any level of ground stations day by day0The space vectors formed by the corresponding satellite precipitation data and the ground station observation precipitation data are respectively P1，P2，…，PnThe weight of the corresponding satellite precipitation data is beta1，β2，…，βnThe error is epsilon; the following relationship is satisfied among the spatial vector, the weight and the error,
P0＝β1P1+β2P2+…+βnPn+ε
wherein, the premise of satisfying the optimal weight is that the error epsilon approaches to 0;
s322, setting a target function, solving by adopting a genetic algorithm, and searching for optimal weight beta 'of each level of satellite precipitation data'1，β′2，...，β′n(ii) a The objective function is as follows,
f＝min[P0-(β1P1+β2P2+…+βnPn)]。
the basic operation process of the genetic algorithm comprises initialization, individual evaluation, fitness calculation, selection operation, cross operation, mutation operation and termination condition judgment. The method fully exerts the main characteristics of genetic algorithm selection, intersection and variation when solving the objective function, and finds the global optimal solution.
Step S33 is specifically that, in order to avoid the situation that the genetic algorithm falls into a local optimal solution when solving the optimal weight of the satellite precipitation data, it needs to be determined whether the optimal weight of the satellite precipitation data is a global optimal solution, and if so, the process goes to step S4; if not, the process returns to step S32 until the optimal weight is the global optimal solution.
Weight self-checking table formation
This part corresponds to the content of step S4, and step S4 is to create a weight self-lookup table based on the optimal weight of each category of monthly precipitation. The optimal weight of each satellite precipitation data determined in the steps S32-S33 is filled in a table month by month according to each precipitation level to form a weight self-lookup table.
Fifth, application of weight self-checking table
Step S5 is further included after step S4, which corresponds to the content of step S5, and as shown in fig. 5, step S5 is to calculate the precipitation amount of the study area of the missing data by using a weight self-lookup table to supplement and perfect the missing data of the study area; the specific process is that,
s51, performing basin partitioning on the research basin by using a known ground station as a base point and adopting a Thiessen polygon method; and the precipitation types of other ground stations in each subarea are specified to be the same as those of the known stations;
s52, sequentially extracting longitude and latitude coordinates of the satellite precipitation data according to the satellite precipitation data of the research basin, and simultaneously eliminating coordinates coincident with the positions of known ground stations;
s53, sequentially judging the partition of each coordinate;
s54, determining the precipitation level of each partition;
s55, sequentially comparing the weight self-lookup table, and sequentially calculating the precipitation amount at each defect position of the research area according to a data fusion formula, so as to supplement and perfect the defect data of the research area; the data fusion formula is as follows,
P0＝β′1P1+β′2P2+…+β′nPn。
specifically, in step S53, coordinates of the satellite precipitation data in step S52 are successively compared with boundary ranges of the partitions, and the partition to which the coordinates belong is determined, so that the satellite precipitation data are accurately partitioned.
Example two
In this embodiment, the execution process of the fusion method provided by the present invention is described in detail with reference to specific examples. The example is described by taking 5 ground observation sites of Kanga, Miao stone sand field ditch, Miyao, Xiaojing valley and Kino mountain of the Yangtze river basin as examples.
First, data preprocessing
Firstly, data preprocessing is carried out on ground station data and satellite precipitation data, and the purpose is to unify data formats and facilitate subsequent operations.
1. And (3) checking the missing condition of each station data for the ground station observation data, removing the station if the station data are seriously missing, and performing interpolation processing if the station data are slightly missing. Longitude and latitude information of 5 ground observation stations of Kanga, Miao tailing sand field ditch, Miyao, Xiaojing valley and Kinuo mountain in the Langanjiang river basin is shown in table 1, and day-by-day observation precipitation data from 1 month and 1 day in 2010 to 12 months and 31 days in 2015 is shown in table 2.
TABLE 1 site latitude and longitude information
TABLE 2 ground station daily observation precipitation data (unit: mm)
2. And for the satellite precipitation data, the precipitation data in the file is read in batch by using a programming language, format conversion is carried out, and the time and space resolution (resampling) of various data is unified. The present invention takes TRMM 3B42RT, IMERG Early Run, IMERG Late Run, and IMERG Final Run precipitation data as an example, and the related information is shown in Table 3. The pretreatment is carried out to obtain the daily precipitation amount of each satellite precipitation product at the longitude and latitude positions of 98.6, 29.1, 99.2, 25.9, 99.7, 26.3, 100.6, 23.7, 101.0 and 22.0 respectively (the precipitation fusion part adopts data from 2010-01-01 to 2015-12-31), and the steps are shown in table 4, table 5, table 6 and table 7 sequentially.
TABLE 3 satellite precipitation data information
TABLE 4 TRMM 3B42 precipitation data (unit: mm)
TABLE 5 GPM IMERG Early Run precipitation data (units: mm)
TABLE 6 GPM IMERG Late Run precipitation data (unit: mm)
TABLE 7 GPM IMERG Final Run precipitation data (units: mm)
Second, data precision evaluation
And respectively performing space-time change characteristic analysis (precipitation annual distribution characteristic, annual change trend and spatial distribution characteristic) on the ground station observation data and the satellite precipitation data based on the DEM elevation data, so that the analysis result of each satellite data and the ground station observation data is compared, and the closer to the analysis result of the ground station, the higher the precision of the satellite precipitation data is. In addition, the accuracy evaluation work is carried out, which mainly comprises consistency evaluation and detection capability evaluation work, and the specific steps are as follows:
1. and (5) evaluating the consistency. And respectively adopting a correlation coefficient R, a relative deviation BIAS, a root mean square error RMSR and an average absolute error MAE to calculate, wherein the closer the correlation coefficient is to 1, the closer the other three indexes are to 0, the better the consistency evaluation result is. Wherein, each parameter calculation formula is as follows:
in the formula, SiRepresenting satellite precipitation data, PiPrecipitation data is observed on behalf of ground sites.
2. And (5) evaluating the detection capability. And calculating by sequentially applying a detection rate POD, a false alarm rate FAR and a critical determination index CSI, wherein the evaluation work comprises a daily scale, a monthly scale, a quarterly scale and a yearly scale. Wherein, each evaluation index calculation formula is as follows:
in the formula: h is the number of precipitation events accurately detected by the satellite precipitation product; f is the number of false precipitation events; m is the number of missed precipitation events; the POD is used for measuring the detection capability of the precipitation event, the value range is [0,1], and the larger the value is, the higher the successful detection degree of the satellite precipitation product on the precipitation event is; the FAR is used for measuring the probability of the precipitation event being mispredicted, the value range is [0,1], and the smaller the value is, the smaller the false alarm degree of the satellite precipitation product on precipitation is; the CSI represents the proportion of the total number of precipitation events correctly detected by the satellite precipitation product to the total number of the events, the characteristics of the satellite precipitation data set can be comprehensively reflected, the value range is [0,1], the optimal value is 1, and the larger the value is, the better the evaluation effect is.
Solving for optimal weight
And correcting the satellite precipitation data meeting the precision evaluation requirement screened out by the second part of contents and the ground station observation precipitation data based on a partition self-adaptive weight fusion method.
1. The daily precipitation data are graded according to precipitation intensity grade grading standards issued by the national weather bureau (as shown in table 8).
TABLE 8 precipitation strength grade division Standard (inland part)
2. And (4) solving the weight by adopting a genetic algorithm according to the precipitation level monthly.
For each 0.1 x 0.1 grid in the data set, there is a ground site observation and satellite precipitation data that meets the evaluation criteria, and the accuracy of the satellite precipitation data is typically affected by the intensity of the precipitation and the season. The method classifies daily rainfall in a research area and solves the weight of each satellite rainfall data in graded rainfall monthly.
For each level of monthly precipitation, the weight of the satellite precipitation data in the same level of the same month in the corresponding multi-source data fusion process is considered to be a certain value. Therefore, the space vector P is formed by observing precipitation data of any level of ground stations day by day0The space vectors formed by the corresponding satellite precipitation data are respectively P1，P2，…，PnThe weights are respectively beta1，β2，…，βnAnd the error is epsilon, the following formula is satisfied:
P0＝β1P1+β2P2+…+βnPn+ε
as can be seen from the above equation, the premise for satisfying the optimal weight is that the error e approaches 0. Therefore, numerical methods can be used to solve, and the objective function is as follows:
f＝min[P0-(β1P1+β2P2+…+βnPn)]
solving by adopting a genetic algorithm aiming at a target function, and searching for optimal weight beta'1，β′2，...，β′n. WhereinThe basic operation process of the genetic algorithm comprises initialization, individual evaluation, fitness calculation, selection operation, cross operation, mutation operation and termination condition judgment. The method gives full play to the main characteristics of genetic algorithm selection, intersection and variation when solving the objective function, and searches for the global optimal solution.
3. Obtaining optimal weight beta 'of each satellite rainfall data through genetic algorithm'1，β′2，...，β′nThen, in order to avoid the genetic algorithm from falling into the local optimal solution, whether the genetic algorithm is the global optimal solution or not is further judged, if yes, a fusion correction result is obtained through calculation, and the fourth step is carried out; otherwise, returning to the step 2 until the requirements are met, and performing data fusion.
Weight self-checking table formation
And making a weight self-lookup table based on the hierarchical optimal weight. Obtaining 3 precipitation data meeting the requirements of the embodiment by a satellite precipitation data precision evaluation method, wherein the precipitation data are GPM IMERG Early Run, Late Run and Final Run, and the corresponding optimal weights are recorded as beta'1，β′2，β′3The resulting weights are tabulated in table 9.
Table 9 weight self-lookup table
Fifth, application of weight self-checking table
And solving the rainfall of the area lacking the survey by using a weight self-checking table, wherein the method comprises the following specific steps of:
1. partitioning is carried out based on known ground observation stations by adopting a Thiessen polygon method, and precipitation types of other places in each partition are specified to be the same as those of the known stations. In this embodiment, the lanucangriver basin is divided into thiessens by taking 5 ground observation sites of tibetan gao (a), miao stone sand field ditch (B), maidsha (C), miniascape (D) and kenozan (E) as base points, and the division results are shown in fig. 6.
2. Satellite precipitation data of GPM IMERG Early Run, Late Run and Final Run of the Yangtze river basin, 9 and 30 days of 2020, are respectively shown in tables 10, 11 and 12, and the precipitation amount is observed by a ground station of the Tibetan Gao (A), the Miao stone sand field ditch (B), the Mitsubishi (C), the Xiaojing valley (D) and the Kinoshan (E) in the same day in Table 13. And (4) gradually extracting longitude and latitude coordinates based on the precipitation data of each satellite, simultaneously eliminating coordinates coincident with the positions of known ground observation stations, and entering the next step after the coordinates are eliminated.
TABLE 10 GPM IMERG Early Run satellite precipitation data (units: mm)
TABLE 11 GPM IMERG Late Run satellite precipitation data (unit: mm)
TABLE 12 GPM IMERG Final Run satellite precipitation data (unit: mm)
Watch 13 ground station precipitation data
3. The filtered coordinates were successively compared with the boundary range of each partition (which has A, B, C, D, E total 5 regions in this example) using a programming language.
Firstly, judging whether the coordinate belongs to the area A, if so, carrying out the next step 5, if not, judging whether the coordinate belongs to the area B, sequentially comparing the coordinate with the boundary ranges of A, B, C, D and E until determining the area to which the coordinate belongs, and then, entering the step 4.
4. The precipitation category at the coordinates is determined, and the patent provides that the precipitation level at each position in each partition is consistent with the conditions of the known ground stations in the area.
5. The precipitation at the selected coordinates is calculated against a weight self-lookup table according to the following equation.
P0＝β′1P1+β′2P2+…+β′nPn)
And (4) performing the operations of the steps (2) to (6) on each extracted coordinate until the precipitation of each lack-of-measurement position in the research area is calculated. The partition calculation is performed by taking the precipitation data of GPM IMERG Early Run, Late Run and Final Run of 30 days in 9 months in 2020 as an example, and the result is shown in Table 14.
TABLE 14 partition adaptive weighting method for calculating precipitation result (unit: mm)
By adopting the technical scheme disclosed by the invention, the following beneficial effects are obtained:
the invention discloses a multi-source precipitation data fusion method based on partition self-adaptive weight, which integrates the advantages of high spatial resolution of satellite precipitation data and high precision of ground station observation data, properly eliminates the disadvantages of the two, and makes up for the deficiencies of the two data to a greater extent by a reverse solving mode, thereby obtaining precipitation data with higher accuracy and stronger applicability. The advantages of the data mining technology in processing mass data are well exerted, the association rule analysis method is applied to the precipitation forecast of the area, and the accuracy of the forecast result can be further improved based on the hidden relation among the data.
The foregoing is only a preferred embodiment of the present invention, and it should be noted that, for those skilled in the art, various modifications and improvements can be made without departing from the principle of the present invention, and such modifications and improvements should also be considered within the scope of the present invention.
Claims (10)
1. A multi-source precipitation data fusion method based on partition self-adaptive weight is characterized by comprising the following steps: comprises the following steps of (a) carrying out,
s1, preprocessing the satellite precipitation data and the ground station observation precipitation data respectively;
s2, evaluating the precision of the preprocessed satellite precipitation data, and screening out the satellite precipitation data meeting the precision evaluation requirement;
s3, performing fusion correction on the satellite precipitation data meeting the precision evaluation requirement and the ground station observation precipitation data based on a partition self-adaptive weight fusion method to obtain the optimal weight of each level of satellite precipitation data;
and S4, making a weight self-lookup table based on the optimal weight of each level of satellite precipitation data.
2. The partition adaptive weight-based multi-source precipitation data fusion method according to claim 1, wherein: the step S1 specifically includes the following contents,
preprocessing satellite precipitation data: reading various satellite precipitation data in the file in batches by using a programming language, carrying out format conversion, and unifying the time and space resolution of the various satellite precipitation data; acquiring daily rainfall of various satellite rainfall data in longitude and latitude;
preprocessing the observation precipitation data of the ground station: checking the missing condition of each ground station data, and if the data is seriously missing, removing the ground station data; if the data is slightly missing, the interpolation processing is performed on the ground point data.
3. The partition adaptive weight-based multi-source precipitation data fusion method according to claim 1, wherein: the step S2 specifically includes the following contents,
s21, performing space-time characteristic analysis on the satellite precipitation data and the ground station observation data based on the DEM elevation data, and judging whether the analysis results of the satellite precipitation data and the ground station precipitation data are similar or not, wherein the closer the results of the satellite precipitation data and the ground station precipitation data are, the higher the precision of the corresponding satellite precipitation data is;
s22, respectively adopting correlation coefficients, relative deviation, root mean square error and average absolute error to evaluate the consistency of the satellite precipitation data, and simultaneously respectively using detection rate, false alarm rate and critical determination index to evaluate the detection capability of the satellite precipitation data so as to screen out the satellite precipitation data meeting the evaluation requirement; the consistency assessment and the detectability assessment each include daily, monthly, quarterly, and yearly scale assessments.
4. The partition adaptive weight-based multi-source precipitation data fusion method according to claim 3, wherein: specifically, the consistency evaluation is to calculate each evaluation index respectively, including a correlation coefficient, a relative deviation, a root mean square error and an average absolute error, wherein the closer the correlation coefficient is to 1, the closer the other three errors are to 0, the better the consistency evaluation result is represented; the calculation formula of each evaluation index is as follows
Wherein R is a correlation coefficient; BIAS is relative deviation; RMSE is root mean square error; MAE is the mean absolute error; siFor the ith satellite precipitation data, PiObserving precipitation data for the ith ground station; 1,2,3, …, n; and n is the total number of the satellite precipitation data.
5. The partition adaptive weight-based multi-source precipitation data fusion method according to claim 3, wherein: the detection capability assessment specifically comprises sequentially calculating detection rate, false alarm rate and critical determination index, wherein the calculation formulas are respectively as follows,
h is the number of precipitation events accurately detected by the satellite precipitation data; f is the number of false precipitation events; m is the number of missed precipitation events; the POD is used for measuring the detection capability of the rainfall event, the value range is [0,1], and the higher the value of the POD is, the higher the successful detection degree of the satellite rainfall data to the rainfall event is; the FAR is used for measuring the probability of the precipitation event being mispredicted, the value range is [0,1], and the smaller the value of the FAR is, the smaller the false alarm degree of the satellite precipitation data to precipitation is; the CSI is the proportion of the total number of precipitation events correctly detected by the satellite precipitation data to the total number of the events, the characteristics of the satellite precipitation data set can be comprehensively reflected, the value range is [0,1], the optimal value is 1, and the higher the value of the CSI is, the better the detection capability evaluation effect is.
6. The partition adaptive weight-based multi-source precipitation data fusion method according to claim 1, wherein: the step S3 specifically includes the following contents,
s31, classifying the daily precipitation data according to the provisions of the meteorological bureau in a grading way, wherein the daily precipitation data comprise light rain, medium rain, heavy rainstorm and extra heavy rainstorm;
s32, solving the optimal weight of each level of precipitation month by adopting a genetic algorithm;
s33, judging whether the optimal weights meet the error requirements, if so, entering the step S4; otherwise, return to step S32.
7. The partition adaptive weight-based multi-source precipitation data fusion method of claim 6, wherein: the step S32 specifically includes the following contents,
s321, regarding every level of monthly rainfall, determining the weight of the satellite rainfall data of the same level of the same month in the corresponding multi-source data fusion process as a certain value; therefore, the space vector P is formed by observing precipitation data of any level of ground stations day by day0The space vectors formed by the corresponding satellite precipitation data and the ground station observation precipitation data are respectively P1，P2，…，PnThe weight of the corresponding satellite precipitation data is beta1，β2，…，βnThe error is epsilon; the following relationship is satisfied among the spatial vector, the weight and the error,
P0＝β1P1+β2P2+…+βnPn+ε
wherein, the premise of satisfying the optimal weight is that the error epsilon approaches to 0;
s322, setting an objective function, and adoptingSolving by using a genetic algorithm, and searching optimal weight beta 'of rainfall data of each level of satellite'1，β′2，...，β′n(ii) a The objective function is as follows,
f＝min[P0-(β1P1+β2P2+…+βnPn)]。
8. the partition adaptive weight-based multi-source precipitation data fusion method of claim 6, wherein: step S33 is specifically that, in order to avoid the situation that the genetic algorithm falls into a local optimal solution when solving the optimal weight of the satellite precipitation data, it needs to be determined whether the optimal weight of the satellite precipitation data is a global optimal solution, and if so, the process goes to step S4; if not, the process returns to step S32 until the optimal weight is the global optimal solution.
9. The partition adaptive weight-based multi-source precipitation data fusion method according to claim 7, wherein: step S5 is further included after step S4, and step S5 is to calculate the precipitation of the study area of the lack of measurement data by using a weight self-lookup table so as to supplement and improve the lack of measurement data of the study area; the specific process is that,
s51, performing basin partitioning on the research basin by using a known ground station as a base point and adopting a Thiessen polygon method; and the precipitation types of other ground stations in each subarea are specified to be the same as those of the known stations;
s52, sequentially extracting longitude and latitude coordinates of the satellite precipitation data according to the satellite precipitation data of the research basin, and simultaneously eliminating coordinates coincident with the positions of known ground stations;
s53, sequentially judging the partition of each coordinate;
s54, determining the precipitation level of each partition;
s55, sequentially comparing the weight self-lookup table, and sequentially calculating the precipitation amount at each defect position of the research area according to a data fusion formula, so as to supplement and perfect the defect data of the research area; the data fusion formula is as follows,
P0＝β′1P1+β′2P2+…+β′nPn。
10. the partition adaptive weight-based multi-source precipitation data fusion method of claim 9, wherein: specifically, in step S53, coordinates of the satellite precipitation data in step S52 are successively compared with boundary ranges of the partitions, and the partition to which the coordinates belong is determined, so that the satellite precipitation data are accurately partitioned.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110583497.4A CN113205155A (en) | 2021-05-27 | 2021-05-27 | Multi-source precipitation data fusion method based on partition self-adaptive weight |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110583497.4A CN113205155A (en) | 2021-05-27 | 2021-05-27 | Multi-source precipitation data fusion method based on partition self-adaptive weight |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113205155A true CN113205155A (en) | 2021-08-03 |
Family
ID=77023422
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110583497.4A Pending CN113205155A (en) | 2021-05-27 | 2021-05-27 | Multi-source precipitation data fusion method based on partition self-adaptive weight |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113205155A (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114020725A (en) * | 2021-11-11 | 2022-02-08 | 中国水利水电科学研究院 | Window sliding GPM data correction method considering spatial distribution |
| CN115795399A (en) * | 2023-01-31 | 2023-03-14 | 中国科学院地理科学与资源研究所 | Self-adaptive fusion method and system for multi-source remote sensing precipitation data |
| CN117290675A (en) * | 2023-11-27 | 2023-12-26 | 中国电建集团西北勘测设计研究院有限公司 | Precipitation data processing method and device, storage medium and electronic equipment |
| CN117668750A (en) * | 2023-12-04 | 2024-03-08 | 中国水利水电科学研究院 | Fusion potential quantification method for multi-source precipitation product |
| CN117708113A (en) * | 2024-02-06 | 2024-03-15 | 中国电建集团西北勘测设计研究院有限公司 | Precipitation data construction method |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103713336A (en) * | 2013-12-24 | 2014-04-09 | 广西壮族自治区气象服务中心 | Hydropower station basin areal rainfall meteorology forecast method based on GIS subarea |
| CN107918165A (en) * | 2016-10-09 | 2018-04-17 | 清华大学 | More satellites fusion Prediction of Precipitation method and system based on space interpolation |
| CN108761574A (en) * | 2018-05-07 | 2018-11-06 | 中国电建集团北京勘测设计研究院有限公司 | Rainfall evaluation method based on Multi-source Information Fusion |
| CN110118982A (en) * | 2019-04-12 | 2019-08-13 | 大连理工大学 | A kind of satellite precipitation data bearing calibration based on space optimization interpolation |
| CN111078678A (en) * | 2019-12-18 | 2020-04-28 | 中国气象局乌鲁木齐沙漠气象研究所 | Satellite precipitation data correction method based on multi-source information fusion and scale reduction |
| CN112070286A (en) * | 2020-08-25 | 2020-12-11 | 贵州黔源电力股份有限公司 | Rainfall forecast early warning system for complex terrain watershed |
| CN112800634A (en) * | 2021-04-07 | 2021-05-14 | 水利部交通运输部国家能源局南京水利科学研究院 | Rainfall estimation method and system coupling dry-wet state identification and multi-source information fusion |
-
2021
- 2021-05-27 CN CN202110583497.4A patent/CN113205155A/en active Pending
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103713336A (en) * | 2013-12-24 | 2014-04-09 | 广西壮族自治区气象服务中心 | Hydropower station basin areal rainfall meteorology forecast method based on GIS subarea |
| CN107918165A (en) * | 2016-10-09 | 2018-04-17 | 清华大学 | More satellites fusion Prediction of Precipitation method and system based on space interpolation |
| CN108761574A (en) * | 2018-05-07 | 2018-11-06 | 中国电建集团北京勘测设计研究院有限公司 | Rainfall evaluation method based on Multi-source Information Fusion |
| CN110118982A (en) * | 2019-04-12 | 2019-08-13 | 大连理工大学 | A kind of satellite precipitation data bearing calibration based on space optimization interpolation |
| CN111078678A (en) * | 2019-12-18 | 2020-04-28 | 中国气象局乌鲁木齐沙漠气象研究所 | Satellite precipitation data correction method based on multi-source information fusion and scale reduction |
| CN112070286A (en) * | 2020-08-25 | 2020-12-11 | 贵州黔源电力股份有限公司 | Rainfall forecast early warning system for complex terrain watershed |
| CN112800634A (en) * | 2021-04-07 | 2021-05-14 | 水利部交通运输部国家能源局南京水利科学研究院 | Rainfall estimation method and system coupling dry-wet state identification and multi-source information fusion |
Non-Patent Citations (1)
| Title |
|---|
| 吴晨晨: ""多源降雨信息评估及在洪水预报中的耦合利用研究——以诺敏河小二沟以上流域为例"", 《中国优秀博硕士学位论文全文数据库(硕士)基础科学辑》 * |
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114020725A (en) * | 2021-11-11 | 2022-02-08 | 中国水利水电科学研究院 | Window sliding GPM data correction method considering spatial distribution |
| CN114020725B (en) * | 2021-11-11 | 2022-04-22 | 中国水利水电科学研究院 | Window sliding GPM data correction method considering spatial distribution |
| CN115795399A (en) * | 2023-01-31 | 2023-03-14 | 中国科学院地理科学与资源研究所 | Self-adaptive fusion method and system for multi-source remote sensing precipitation data |
| CN117290675A (en) * | 2023-11-27 | 2023-12-26 | 中国电建集团西北勘测设计研究院有限公司 | Precipitation data processing method and device, storage medium and electronic equipment |
| CN117290675B (en) * | 2023-11-27 | 2024-02-27 | 中国电建集团西北勘测设计研究院有限公司 | Precipitation data processing method and device, storage medium and electronic equipment |
| CN117668750A (en) * | 2023-12-04 | 2024-03-08 | 中国水利水电科学研究院 | Fusion potential quantification method for multi-source precipitation product |
| CN117668750B (en) * | 2023-12-04 | 2024-09-24 | 中国水利水电科学研究院 | Fusion potential quantification method for multi-source precipitation product |
| CN117708113A (en) * | 2024-02-06 | 2024-03-15 | 中国电建集团西北勘测设计研究院有限公司 | Precipitation data construction method |
| CN117708113B (en) * | 2024-02-06 | 2024-05-17 | 中国电建集团西北勘测设计研究院有限公司 | Precipitation data construction method |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113205155A (en) | Multi-source precipitation data fusion method based on partition self-adaptive weight | |
| Boughton et al. | Estimating runoff in ungauged catchments from rainfall, PET and the AWBM model | |
| Merz et al. | Regionalisation of catchment model parameters | |
| Rauthe et al. | A Central European precipitation climatology–Part I: Generation and validation of a high-resolution gridded daily data set (HYRAS) | |
| Roemmich et al. | The 2004–2008 mean and annual cycle of temperature, salinity, and steric height in the global ocean from the Argo Program | |
| Tveito et al. | A GIS-based agro-ecological decision system based on gridded climatology | |
| Sexton et al. | Using NEXRAD and rain gauge precipitation data for hydrologic calibration of SWAT in a northeastern watershed | |
| CN110334446A (en) | The mountain torrents Critical Rainfall calculation method of NO emissions reduction processing based on satellite precipitation data | |
| CN111401602A (en) | Assimilation method for satellite and ground rainfall measurement values based on neural network | |
| CN114254802B (en) | Prediction method for vegetation coverage space-time change under climate change drive | |
| CN111652443A (en) | Method for predicting ocean fishery resource abundance by comprehensive multi-source satellite remote sensing and application thereof | |
| Zhu et al. | Uncertainty assessment in baseflow nonpoint source pollution prediction: The impacts of hydrographic separation methods, data sources and baseflow period assumptions | |
| CN105069295A (en) | Assimilation method for satellite and ground rainfall measured values based on Kalman filtering | |
| Chen et al. | Improving the practicability of remote sensing data-assimilation-based crop yield estimations over a large area using a spatial assimilation algorithm and ensemble assimilation strategies | |
| CN116910041B (en) | Daily correction method for remote sensing precipitation product based on scale analysis | |
| CN110874454A (en) | Method for accurately measuring and calculating regional scale moso bamboo carbon reserves based on mixed probability density | |
| McGehee et al. | An updated isoerodent map of the conterminous United States | |
| Golbeck et al. | Uncertainty estimation for operational ocean forecast products—a multi-model ensemble for the North Sea and the Baltic Sea | |
| Ndomba | Validation of PSIAC model for sediment yields estimation in ungauged catchments of Tanzania | |
| Huynh et al. | An optimal rain-gauge network using a GIS-based approach with spatial interpolation techniques for the Mekong River Basin | |
| CN113627465A (en) | Rainfall data space-time dynamic fusion method based on convolution long-short term memory neural network | |
| Zhu et al. | Comprehensive quantitative assessment of the performance of fourteen satellite precipitation products over Chinese mainland | |
| Al Kloub | An optimal method for high-resolution population geo-spatial data | |
| CN116630676B (en) | Large-scale-range field classification processing method and device and electronic equipment | |
| Gyasi-Agyei et al. | Optimal selection of daily satellite precipitation product based on structural similarity index at 1 km resolution for the Pra catchment, Ghana |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20210803 |
|
| RJ01 | Rejection of invention patent application after publication |