CN112363168B - Assimilation fusion method based on radar extrapolation and mode prediction - Google Patents

Assimilation fusion method based on radar extrapolation and mode prediction Download PDF

Info

Publication number
CN112363168B
CN112363168B CN202110042124.6A CN202110042124A CN112363168B CN 112363168 B CN112363168 B CN 112363168B CN 202110042124 A CN202110042124 A CN 202110042124A CN 112363168 B CN112363168 B CN 112363168B
Authority
CN
China
Prior art keywords
precipitation
radar
time
extrapolation
estimation field
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
Application number
CN202110042124.6A
Other languages
Chinese (zh)
Other versions
CN112363168A (en
Inventor
史冰颖
闵锦忠
张录军
叶万恒
张琦蓓
高语秋
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Nanjing Manxing Data Technology Co ltd
Original Assignee
Nanjing Manxing Data Technology Co ltd
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Nanjing Manxing Data Technology Co ltd filed Critical Nanjing Manxing Data Technology Co ltd
Priority to CN202110042124.6A priority Critical patent/CN112363168B/en
Publication of CN112363168A publication Critical patent/CN112363168A/en
Application granted granted Critical
Publication of CN112363168B publication Critical patent/CN112363168B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
    • G01S13/00—Systems using the reflection or reradiation of radio waves, e.g. radar systems; Analogous systems using reflection or reradiation of waves whose nature or wavelength is irrelevant or unspecified
    • G01S13/88—Radar or analogous systems specially adapted for specific applications
    • G01S13/95—Radar or analogous systems specially adapted for specific applications for meteorological use
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
    • G01S13/00—Systems using the reflection or reradiation of radio waves, e.g. radar systems; Analogous systems using reflection or reradiation of waves whose nature or wavelength is irrelevant or unspecified
    • G01S13/88—Radar or analogous systems specially adapted for specific applications
    • G01S13/95—Radar or analogous systems specially adapted for specific applications for meteorological use
    • G01S13/958—Theoretical aspects
    • 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
    • Y02A90/00—Technologies having an indirect contribution to adaptation to climate change
    • Y02A90/10—Information and communication technologies [ICT] supporting adaptation to climate change, e.g. for weather forecasting or climate simulation

Landscapes

  • Engineering & Computer Science (AREA)
  • Radar, Positioning & Navigation (AREA)
  • Remote Sensing (AREA)
  • Physics & Mathematics (AREA)
  • Electromagnetism (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • General Physics & Mathematics (AREA)
  • Radar Systems Or Details Thereof (AREA)

Abstract

The invention discloses an assimilation fusion method based on radar extrapolation and mode prediction, which is characterized in that a three-dimensional variational assimilation method is adopted to fuse radar extrapolation and numerical prediction to construct an appropriate short-term assimilation fusion precipitation prediction method, the problem of development of error distribution characteristics of extrapolation estimation and mode prediction before objective estimation and assimilation fusion is solved, the method has the advantages of conveniently and objectively distributing mode prediction and radar extrapolation fusion weights to achieve advantage complementation, and a new solution is provided for reasonably distributing the fusion weights of the two prediction methods.

Description

Assimilation fusion method based on radar extrapolation and mode prediction
Technical Field
The invention belongs to the technical field of weather, and particularly relates to an assimilation fusion method based on radar extrapolation and mode prediction.
Background
China is in an east Asia monsoon region, covers high, medium and low latitudes, has fluctuant and variable topography and complex natural conditions, so that the meteorological disasters in China have more types, high occurrence frequency, wide distribution region and long duration, and are seriously influenced by the meteorological disasters; the meteorological disasters caused by strong rainfall have strong paroxysmal and social hazards, so the early warning is accurately carried out, and the significance for preventing and reducing disasters is great.
At present, 0-6 hour approach precipitation forecasting methods mainly include extrapolation forecasting, numerical forecasting, concept model forecasting and the like. The mesoscale numerical mode prediction is limited by a mode Spin-up, an ideal prediction result is difficult to output in the first 2 hours, but the accuracy is higher in 2-6 hours; the accuracy of the proximity prediction technology based on radar observation and echo recognition, tracking and extrapolation is obviously superior to that of a high-resolution numerical prediction mode within 0-2 hours, but high-quality prediction for more than 2 hours is difficult to provide. Therefore, the project group adopts a three-dimensional variation method to fuse radar extrapolation nowcasting and numerical prediction, combines the advantages of the two methods to make up the mutual deficiency, and provides a reasonable scheme for effective prediction of convection scale weather systems, particularly convection strong precipitation for 0-6 hours.
The nowcasting is a forecasting service project for preventing disastrous weather such as an emergency local strong storm and the like, and aims to meet the special weather service requirements of various economic departments as much as possible and provide high-quality weather guarantee service for various social public activities; the short-time approach rainfall forecast of 0-6 h mainly comprises extrapolation forecast, numerical forecast, concept model forecast and the like by using real-time radar data; the current mainstream forecasting method usually adopts extrapolation forecasting based on radar observation data to forecast rainfall within 2h, and selects rainfall forecasting based on a numerical weather mode to forecast for hours to days.
However, although there are many alternative forecasting methods, how to improve the accuracy of 0-6 h short-term precipitation forecasting is still a complicated worldwide problem, and long-term research is required.
In the field of radar extrapolation prediction, a cross correlation method is a method commonly used for radar extrapolation rainfall approach prediction, and the main idea is as follows: the method comprises the steps of obtaining movement vector characteristics of a convection system in different directions by utilizing the space optimal correlation of data such as radar echoes and the like in adjacent time, predicting the future position and strength of the echoes according to the directions and the strength of the radar echoes, and estimating the precipitation at the future time by utilizing a Z-R relation; the algorithm starts from the pioneering work of Rinehart et al at first, and then, with the improvement of technologies such as radar data resolution and the like, numerous scholars further improve and develop on the basis of the TREC algorithm; comprehensive analysis shows that the TREC and the development method thereof only need radar echo data of two times, have quick running time, are simple and convenient methods for forecasting the position of the rainfall system at the future moment, but cannot forecast the evolution trend of the system, have the forecasting accuracy rapidly reduced along with the forecasting time efficiency, and generally have the forecasting time efficiency not more than 2 h.
In contrast, for a numerical weather forecast mode, the method can well simulate the process of the water-falling field evolving along with time, and has the advantages that the radar extrapolation cannot achieve in forecast of hours or even days in the future.
However, because the input factors such as the initial water material field and the like are difficult to reflect the real atmospheric state, and in addition, other technical problems, the numerical mode is difficult to achieve the ideal forecasting effect 2h before the forecasting.
Therefore, according to the characteristic that the extrapolation prediction technology based on radar observation and the rainfall prediction technology based on numerical mode are respectively long, the prediction results of the extrapolation prediction technology and the rainfall prediction technology based on numerical mode are subjected to advantage complementation through the fusion technology to form a 0-6 h quantitative rainfall prediction scheme, and the method is a feasible scheme for solving the accuracy rate of short-time approaching rainfall prediction.
In the 2011 world weather research plan nowcasting working group (WMO-WWRP WGNR) meeting of the world weather organization, experts of various countries discuss and emphasize that the fusion of a radar extrapolation technology and a numerical prediction technology is a basic technical way for prolonging the nowcasting time to more than 2h and is promoted as an important technical development direction.
At present, technologies for fusing radar extrapolation prediction and numerical mode prediction mainly fall into three categories: weighted average method, trend adjustment method, AMOR method.
The core idea of the first kind of method is to determine the weight of the extrapolation prediction and the numerical model prediction in the fusion system according to the time interval, the weight of the extrapolation prediction in the fusion system is reduced along with the time advance, the weight of the numerical prediction in the fusion system is increased along with the time advance, and the representative system of the method is Nimrod and the like developed by gold.
The second method has the central idea of correcting the extrapolation prediction according to the variation trend of the pattern prediction result on the falling area and the intensity.
The third category of AMOR methods has the main idea that firstly, the error of the mode forecast at the current moment on the falling area or the intensity is calculated, then the change trend of the error on the time is estimated, and finally the mode forecast at the future moment is corrected on the basis.
Nowadays, a fusion rainfall forecast model based on radar nowcasting and numerical models is rapidly developing: countries and regions such as the united kingdom, the united states, canada, australia, hong kong have developed numerous quantitative precipitation forecast studies based on fusion techniques, developing a number of short-term forecasting systems that have achieved preliminary success in business.
However, in continental areas of china, the existing work achievement is limited due to relatively late fusion forecast research. However, with the increasing importance of the society in recent years, favorable results have been obtained. The AMOR method is specifically a short-time quantitative rainfall forecast fusion technical scheme developed on the basis of a Bj-C system and a j-RUC system based on technologies such as Cressman objective, analysis Fast Fourier Transform (FFT), a multi-scale optical flow method, a Weber function, a tangent dynamic weight fusion method and the like. Meanwhile, the method is used for testing 7 typical strong precipitation cases in 2011 and 2012 summer in Jingjin Ji areas, and the method shows that the method has remarkable advantages in prediction skills of more-than-5 mm-strong precipitation time-by-time accumulated quantitative precipitation prediction compared with adjacent prediction and numerical prediction in 0-6 h time period. Schumann and Wangxing propose a rainfall probability fusion forecasting method based on dynamic weight in 2017. The method comprises the steps of establishing a scoring model suitable for weight distribution, dynamically distributing weights of two forecasting methods in different forecasting timeliness according to scores of the two forecasting methods in different forecasting timeliness, and showing that forecasting of each forecasting timelines after fusion shows a technical score which is similar to or even higher than a radar extrapolation or numerical mode through scoring verification of Brier and the like.
Wilson and Roberts et al, in 2010, verified the forecasting ability of various forecasting systems with unified standards, and their results show that: in general, the prediction accuracy of the system adopting the fusion prediction technology is superior to that of the system which simply depends on the radar echo extrapolation prediction. Meanwhile, Wilson also points out that the key for improving the short-term forecasting capability lies in continuously optimizing an extrapolation and forecasting technology on one hand, and also needs to rely on the progress of an extrapolation algorithm and a numerical mode fusion technology on the other hand.
Disclosure of Invention
The invention aims to provide an assimilation fusion method based on radar extrapolation and mode prediction, and aims to solve the technical problems in the prior art.
In order to achieve the purpose, the invention provides the following technical scheme: an assimilation fusion method based on radar extrapolation and mode prediction specifically comprises the following steps:
step 1, selecting K precipitation cases in 0-6 h of the same target area, and reading quantitative precipitation data fields observed by automatic stations of each precipitation case
Figure 207300DEST_PATH_IMAGE001
；
Forecasting each precipitation case by respectively using radar extrapolation and mode forecasting to correspondingly obtain radar extrapolated precipitation estimation fields at various times
Figure 268797DEST_PATH_IMAGE002
Sum-mode forecast precipitation estimation field
Figure 490831DEST_PATH_IMAGE003
；
Step 2, extrapolating the radar to a precipitation estimation field
Figure 309882DEST_PATH_IMAGE002
Interpolating to the station of the automatic station where the target area is located to obtain the interpolated precipitation estimation field
Figure 72302DEST_PATH_IMAGE004
And then calculating the deviation increment of each grid point at each time in each precipitation case by using a Cressman analysis method:
Figure 191568DEST_PATH_IMAGE005
=
Figure 581573DEST_PATH_IMAGE006
(
Figure 153500DEST_PATH_IMAGE007
)；
using said offset increments
Figure 454031DEST_PATH_IMAGE008
Modified radar extrapolated precipitation estimation field
Figure 427804DEST_PATH_IMAGE002
And obtaining the quantitative precipitation estimation field corrected at each time in each precipitation case:
Figure 726061DEST_PATH_IMAGE009
=
Figure 785284DEST_PATH_IMAGE010
and will be
Figure 92768DEST_PATH_IMAGE011
As the true value of each time descending water field in each precipitation case;
step 3, forecasting precipitation estimation field according to the mode
Figure 983364DEST_PATH_IMAGE003
Radar extrapolation precipitation estimation field
Figure 718102DEST_PATH_IMAGE002
And the corrected quantitative precipitation estimation field
Figure 261691DEST_PATH_IMAGE009
The variance of the forecast precipitation estimation value of each time-down mode is calculated according to the following formula
Figure 841708DEST_PATH_IMAGE012
And variance of radar extrapolated precipitation estimates
Figure 586810DEST_PATH_IMAGE013
：
Figure 492449DEST_PATH_IMAGE014
；
Step 4, forecasting precipitation estimation field according to the mode
Figure 526264DEST_PATH_IMAGE003
And radar extrapolation precipitation estimation field
Figure 909972DEST_PATH_IMAGE002
Respectively calculating the correlation coefficient of the distance between any grid point m and any grid point n in all grid points at each time
Figure 978423DEST_PATH_IMAGE015
The calculation formula is as follows:
Figure 851701DEST_PATH_IMAGE016
；
step 5, taking the correlation coefficient of the distance between any grid point m and any grid point n in all grid points at each time
Figure 107233DEST_PATH_IMAGE017
The ordinate and the distance r between the grid point m and the grid point n are the abscissa, and the time-series data are respectively drawn
Figure 557281DEST_PATH_IMAGE017
A scatter plot;
step 6, the times are divided into
Figure 683500DEST_PATH_IMAGE017
Dividing all points in the scatter diagram into X groups, obtaining the average position of each group of points, and performing polynomial fitting on the average position of each group of points to obtain an approximate fitting curve under each time
Figure 930942DEST_PATH_IMAGE018
；
Step 7, utilizing approximate fitting curve
Figure 673770DEST_PATH_IMAGE018
Corrected approximate fitted curve of standard deviation
Figure 461597DEST_PATH_IMAGE018
To obtain an approximate fitCurve line
Figure 973481DEST_PATH_IMAGE018
Is averaged to fit the curve
Figure 391824DEST_PATH_IMAGE019
；
Step 8, forecasting the variance of the rainfall estimation field according to the mode
Figure 684265DEST_PATH_IMAGE012
Variance of radar extrapolation precipitation estimation field
Figure 210537DEST_PATH_IMAGE020
、
Figure 108086DEST_PATH_IMAGE017
Is averaged to fit the curve
Figure 697330DEST_PATH_IMAGE019
Respectively calculating the error correlation function of each time-lower mode forecast rainfall estimation field
Figure 414750DEST_PATH_IMAGE021
The concrete formula is as follows:
Figure 482064DEST_PATH_IMAGE023
；
step 9, forecasting the variance of the rainfall estimation field according to the mode
Figure 234119DEST_PATH_IMAGE012
Variance of radar extrapolation precipitation estimation field
Figure 994264DEST_PATH_IMAGE020
Error correlation function of mode forecast precipitation estimation field
Figure 995719DEST_PATH_IMAGE021
Respectively calculating the error covariance matrix of each time lower mode forecast
Figure 132302DEST_PATH_IMAGE024
Error covariance matrix for sum radar extrapolation prediction
Figure 759371DEST_PATH_IMAGE025
The concrete formula is as follows:
Figure 424839DEST_PATH_IMAGE026
、
Figure 116852DEST_PATH_IMAGE027
，
wherein, the error covariance matrix of any lattice point m
Figure 853863DEST_PATH_IMAGE028
Is marked as
Figure 314932DEST_PATH_IMAGE029
；
Step 10, reading a model forecast rainfall estimation field to be fused in a target area
Figure 416880DEST_PATH_IMAGE030
And radar extrapolation precipitation estimation field
Figure 330609DEST_PATH_IMAGE031
Error covariance matrix based on mode prediction
Figure 808995DEST_PATH_IMAGE032
Error covariance matrix of radar extrapolation prediction
Figure 121640DEST_PATH_IMAGE033
Separately constructing each time-next-related precipitation field
Figure 660069DEST_PATH_IMAGE034
Three-dimensional variational objective function of
Figure 857832DEST_PATH_IMAGE035
The concrete formula is as follows:
Figure 139909DEST_PATH_IMAGE036
；
step 11, solving the three-dimensional variational objective function at each time by using a method of gradually iterating to obtain a minimum value
Figure 309990DEST_PATH_IMAGE037
Solutions of the precipitation field when the minimum value is reached
Figure 550479DEST_PATH_IMAGE038
Namely:
Figure 438800DEST_PATH_IMAGE039
。
preferably, in step 2: spacing of site numbered st from grid point coordinate (i, j)
Figure 258989DEST_PATH_IMAGE040
And (4) the influence radius is smaller than R, and R is 10 times of grid distance.
Preferably, the correction method in step 7 includes: to approximate a fitted curve
Figure 549156DEST_PATH_IMAGE041
Removing the point with the error larger than three times of standard deviation sigma from the mean value, and redrawing each time next
Figure 160878DEST_PATH_IMAGE017
Is averaged to fit the curve
Figure 536496DEST_PATH_IMAGE019
。
Preferably, in step 1:
k is the number of precipitation cases, K =1,2, …, K;
ST is the automatic station number, ST =1,2, …, ST;
t is time, t =0,1,2, …, 60;
i is a lattice point number in the x direction, I =1,2, …, I;
j is a grid point number in the y direction, J =1,2, …, J;
i is the total number of lattice points in the x direction;
j is the total number of grid points in the y-direction.
Preferably, in step 2:
Figure 957113DEST_PATH_IMAGE042
the distance between the site numbered st and the grid point with coordinates (i, j).
Preferably, in the step 4:
r is the distance between any grid point m and any grid point n;
Figure 836207DEST_PATH_IMAGE043
is the coordinate of any grid point m;
Figure 684078DEST_PATH_IMAGE044
is the coordinate of any grid point n;
Figure 546991DEST_PATH_IMAGE045
，
Figure 443403DEST_PATH_IMAGE046
，
Figure 504900DEST_PATH_IMAGE047
，
Figure 195776DEST_PATH_IMAGE048
。
preferably, in step 11:
Figure 342723DEST_PATH_IMAGE049
t =0,1,2, …,60 is the precipitation estimation field after each time-next fusion.
The invention has the technical effects and advantages that: the assimilation fusion method based on radar extrapolation and mode prediction adopts a three-dimensional variation assimilation method to fuse radar extrapolation and numerical prediction to construct an appropriate short-term assimilation fusion precipitation prediction method, solves the problem of research on error distribution characteristics of extrapolation estimation and mode prediction before objective estimation and assimilation fusion, has the advantages of conveniently and objectively distributing mode prediction and radar extrapolation fusion weights to achieve advantage complementation, and provides a new solution for reasonably distributing the fusion weights of the two prediction methods.
Drawings
FIG. 1 shows step 6 of the second embodiment of the present invention
Figure 508738DEST_PATH_IMAGE050
Approximate curve fitting diagram in time.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to fig. 1 in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
The first embodiment is as follows:
the assimilation fusion method based on radar extrapolation and mode prediction of the embodiment specifically comprises the following steps:
step 1, selecting K precipitation cases in 0-6 h of the same target area, and reading quantitative precipitation data fields observed by automatic stations of each precipitation case
Figure 424741DEST_PATH_IMAGE001
(ii) a Wherein K is the number of precipitation cases, K =1,2, …, K; ST is the automatic station number, ST =1,2, …, ST; t is time, t =0,1,2, …, 60;
Figure 552097DEST_PATH_IMAGE001
the cumulative precipitation output is once every 6 minutes, namely the cumulative precipitation of 0, 6, 12, … and 360 minutes;
forecasting each precipitation case by respectively using radar extrapolation and mode forecasting to correspondingly obtain radar extrapolated precipitation estimation fields at various times
Figure 124024DEST_PATH_IMAGE002
Sum-mode forecast precipitation estimation field
Figure 424555DEST_PATH_IMAGE003
(ii) a Wherein I is a lattice point number in the x direction, I =1,2, …, I; j is a grid point number in the y direction, J =1,2, …, J; i is the total number of lattice points in the x direction; j is the total number of grid points in the y direction, and similarly, the results of the radar extrapolation and pattern prediction are output every 6 minutes, t =0,1,2, …, 60.
Step 2, extrapolating the radar to a precipitation estimation field
Figure 132748DEST_PATH_IMAGE002
Interpolating to the station of the automatic station where the target area is located to obtain the interpolated precipitation estimation field
Figure 165426DEST_PATH_IMAGE004
And then calculating the deviation increment of each grid point at each time in each precipitation case by using a Cressman analysis method:
Figure 224649DEST_PATH_IMAGE051
=
Figure 529204DEST_PATH_IMAGE052
(
Figure 623062DEST_PATH_IMAGE007
) Wherein:
Figure 561062DEST_PATH_IMAGE042
the distance between the station numbered st and the grid point with coordinates (i, j), in this step
Figure 107581DEST_PATH_IMAGE042
Selecting 10 times grid spacing for the influence radius R to be smaller than the influence radius R;
using said offset increments
Figure 953177DEST_PATH_IMAGE051
Modified radar extrapolated precipitation estimation field
Figure 698280DEST_PATH_IMAGE002
And obtaining the quantitative precipitation estimation field corrected at each time in each precipitation case:
Figure 603919DEST_PATH_IMAGE009
=
Figure 372155DEST_PATH_IMAGE010
and will be
Figure 284091DEST_PATH_IMAGE011
As a true value for each time-descending water field in each precipitation case.
Step 3, forecasting precipitation estimation field according to the mode
Figure 86962DEST_PATH_IMAGE003
Radar extrapolation precipitation estimation field
Figure 897923DEST_PATH_IMAGE002
And the corrected quantitative precipitation estimation field
Figure 419035DEST_PATH_IMAGE009
The variance of the forecast precipitation estimation value of each time-down mode is calculated according to the following formula
Figure 668750DEST_PATH_IMAGE012
And variance of radar extrapolated precipitation estimates
Figure 326128DEST_PATH_IMAGE013
：
Figure 573569DEST_PATH_IMAGE014
。
Step 4, forecasting precipitation estimation field according to the mode
Figure 316397DEST_PATH_IMAGE003
And radar extrapolation precipitation estimation field
Figure 304558DEST_PATH_IMAGE002
Respectively calculating the correlation coefficient of the distance between any grid point m and any grid point n in all grid points at each time
Figure 878758DEST_PATH_IMAGE015
The calculation formula is as follows:
Figure 297101DEST_PATH_IMAGE016
；
wherein r is the distance between any grid point m and any grid point n;
Figure 527226DEST_PATH_IMAGE053
is the coordinate of any grid point m;
Figure 56427DEST_PATH_IMAGE054
is the coordinate of any grid point n;
Figure 688397DEST_PATH_IMAGE055
，
Figure 543220DEST_PATH_IMAGE056
，
Figure 57378DEST_PATH_IMAGE057
，
Figure 390270DEST_PATH_IMAGE058
。
step 5, any grid point m in all grid points at each timeCorrelation coefficient with distance between arbitrary lattice points n
Figure 342658DEST_PATH_IMAGE059
The ordinate and the distance r between the grid point m and the grid point n are the abscissa, and the time-series data are respectively drawn
Figure 102804DEST_PATH_IMAGE059
And (6) a scatter diagram.
Step 6, the times are divided into
Figure 104258DEST_PATH_IMAGE059
Dividing all points in the scatter diagram into X groups, obtaining the average position of each group of points, and performing polynomial fitting on the average position of each group of points to obtain an approximate fitting curve under each time
Figure 975262DEST_PATH_IMAGE060
。
Step 7, utilizing approximate fitting curve
Figure 581824DEST_PATH_IMAGE060
Corrected approximate fitted curve of standard deviation
Figure 309609DEST_PATH_IMAGE060
To obtain an approximate fitting curve
Figure 736042DEST_PATH_IMAGE060
Is averaged to fit the curve
Figure 410737DEST_PATH_IMAGE061
；
The correction method comprises the following steps: to approximate a fitted curve
Figure 871805DEST_PATH_IMAGE041
Removing the point with the error larger than three times of standard deviation sigma from the mean value, and redrawing each time next
Figure 259840DEST_PATH_IMAGE062
Mean fitted curve ofThread
Figure 439149DEST_PATH_IMAGE063
Therefore, the error caused by the point with the overlarge error is reduced, and a stable result which is more in line with the reality is obtained.
Step 8, forecasting the variance of the rainfall estimation field according to the mode
Figure 448693DEST_PATH_IMAGE012
Variance of radar extrapolation precipitation estimation field
Figure 29847DEST_PATH_IMAGE020
、
Figure 37118DEST_PATH_IMAGE062
Is averaged to fit the curve
Figure 703722DEST_PATH_IMAGE063
Respectively calculating the error correlation function of each time-lower mode forecast rainfall estimation field
Figure 516958DEST_PATH_IMAGE021
The concrete formula is as follows:
Figure 952618DEST_PATH_IMAGE064
。
step 9, forecasting the variance of the rainfall estimation field according to the mode
Figure 661948DEST_PATH_IMAGE012
Variance of radar extrapolation precipitation estimation field
Figure 547340DEST_PATH_IMAGE020
Error correlation function of mode forecast precipitation estimation field
Figure 101949DEST_PATH_IMAGE021
Respectively calculating the error covariance matrix of each time lower mode forecast
Figure 188854DEST_PATH_IMAGE065
Error covariance matrix for sum radar extrapolation prediction
Figure 69085DEST_PATH_IMAGE025
The concrete formula is as follows:
Figure 179124DEST_PATH_IMAGE026
、
Figure 537424DEST_PATH_IMAGE027
，
wherein, the error covariance matrix of any lattice point m
Figure 947677DEST_PATH_IMAGE024
Is marked as
Figure 795547DEST_PATH_IMAGE066
。
Step 10, reading a model forecast rainfall estimation field to be fused in a target area
Figure 658461DEST_PATH_IMAGE030
And radar extrapolation precipitation estimation field
Figure 551943DEST_PATH_IMAGE067
Error covariance matrix based on mode prediction
Figure 551123DEST_PATH_IMAGE024
Error covariance matrix of radar extrapolation prediction
Figure 569894DEST_PATH_IMAGE025
Separately constructing each time-next-related precipitation field
Figure 920104DEST_PATH_IMAGE068
Three-dimensional variational objective function of
Figure 620207DEST_PATH_IMAGE069
The concrete formula is as follows:
Figure 739473DEST_PATH_IMAGE070
。
step 11, solving the three-dimensional variational objective function at each time by using a method of gradually iterating to obtain a minimum value
Figure 663566DEST_PATH_IMAGE071
Solutions of the precipitation field when the minimum value is reached
Figure 501072DEST_PATH_IMAGE072
Namely:
Figure 739287DEST_PATH_IMAGE039
；
Figure 241288DEST_PATH_IMAGE072
t =0,1,2, …,60 is the precipitation estimation field after each time-next fusion.
Example two:
the further design of this embodiment lies in: as shown in FIG. 1, this example will be in step 6
Figure 539545DEST_PATH_IMAGE050
All the points under the time are equally divided into 10 groups, and polynomial fitting is carried out according to the average position of each group to obtain
Figure DEST_PATH_IMAGE073
Approximate curve fitting diagram in time.
Finally, it should be noted that: although the present invention has been described in detail with reference to the foregoing embodiments, it will be apparent to those skilled in the art that modifications may be made to the embodiments or portions thereof without departing from the spirit and scope of the invention.

Claims (7)

1. An assimilation fusion method based on radar extrapolation and mode prediction is characterized by comprising the following steps:
step 1, selecting K precipitation cases in 0-6 h of the same target area, and reading quantitative precipitation data fields observed by automatic stations of each precipitation case
Figure 65389DEST_PATH_IMAGE001
；
Forecasting each precipitation case by respectively using radar extrapolation and mode forecasting to correspondingly obtain radar extrapolated precipitation estimation fields at various times
Figure 423689DEST_PATH_IMAGE002
Sum-mode forecast precipitation estimation field
Figure 568362DEST_PATH_IMAGE003
；
Step 2, extrapolating the radar to a precipitation estimation field
Figure 353916DEST_PATH_IMAGE002
Interpolating to the station of the automatic station where the target area is located to obtain the interpolated precipitation estimation field
Figure 213900DEST_PATH_IMAGE004
And then calculating the deviation increment of each grid point at each time in each precipitation case by using a Cressman analysis method:
Figure 172628DEST_PATH_IMAGE005
=
Figure 437388DEST_PATH_IMAGE006
(
Figure 393842DEST_PATH_IMAGE007
)；
using said offset increments
Figure 540790DEST_PATH_IMAGE008
Modified radar extrapolated precipitation estimation field
Figure 444155DEST_PATH_IMAGE002
And obtaining the quantitative precipitation estimation field corrected at each time in each precipitation case:
Figure 563421DEST_PATH_IMAGE009
=
Figure 753094DEST_PATH_IMAGE010
and will be
Figure 56511DEST_PATH_IMAGE011
As the true value of each time descending water field in each precipitation case;
step 3, forecasting precipitation estimation field according to the mode
Figure 763567DEST_PATH_IMAGE003
Radar extrapolation precipitation estimation field
Figure 534077DEST_PATH_IMAGE002
And the corrected quantitative precipitation estimation field
Figure 832335DEST_PATH_IMAGE009
The variance of the forecast precipitation estimation value of each time-down mode is calculated according to the following formula
Figure 953874DEST_PATH_IMAGE012
And variance of radar extrapolated precipitation estimates
Figure 730201DEST_PATH_IMAGE013
：
Figure 824058DEST_PATH_IMAGE014
；
Step 4, forecasting precipitation estimation field according to the mode
Figure 355534DEST_PATH_IMAGE003
And radar extrapolation precipitation estimation field
Figure 902053DEST_PATH_IMAGE002
Respectively calculating the correlation coefficient of the distance between any grid point m and any grid point n in all grid points at each time
Figure 744720DEST_PATH_IMAGE015
The calculation formula is as follows:
Figure 693084DEST_PATH_IMAGE016
；
step 5, taking the correlation coefficient of the distance between any grid point m and any grid point n in all grid points at each time
Figure 598723DEST_PATH_IMAGE017
The ordinate and the distance r between the grid point m and the grid point n are the abscissa, and the time-series data are respectively drawn
Figure 429276DEST_PATH_IMAGE017
A scatter plot;
step 6, the times are divided into
Figure 812984DEST_PATH_IMAGE017
Dividing all points in the scatter diagram into X groups, obtaining the average position of each group of points, and performing polynomial fitting on the average position of each group of points to obtain an approximate fitting curve under each time
Figure 881434DEST_PATH_IMAGE018
；
Step 7, utilizing approximate fitting curve
Figure 692395DEST_PATH_IMAGE018
Corrected approximate fitted curve of standard deviation
Figure 479086DEST_PATH_IMAGE018
To obtain an approximate fitting curve
Figure 666484DEST_PATH_IMAGE018
Is averaged to fit the curve
Figure 609949DEST_PATH_IMAGE019
；
Step 8, forecasting the variance of the rainfall estimation field according to the mode
Figure 591811DEST_PATH_IMAGE012
Variance of radar extrapolation precipitation estimation field
Figure 396956DEST_PATH_IMAGE020
、
Figure 122467DEST_PATH_IMAGE017
Is averaged to fit the curve
Figure 696668DEST_PATH_IMAGE019
Respectively calculating the error correlation function of each time-lower mode forecast rainfall estimation field
Figure 849431DEST_PATH_IMAGE021
The concrete formula is as follows:
Figure 79556DEST_PATH_IMAGE022
；
step 9, forecasting the variance of the rainfall estimation field according to the mode
Figure 671074DEST_PATH_IMAGE012
Variance of radar extrapolation precipitation estimation field
Figure 37464DEST_PATH_IMAGE020
Error correlation function of mode forecast precipitation estimation field
Figure 423446DEST_PATH_IMAGE023
Respectively calculating the error covariance matrix of each time lower mode forecast
Figure 137937DEST_PATH_IMAGE024
Error covariance matrix for sum radar extrapolation prediction
Figure 470829DEST_PATH_IMAGE025
The concrete formula is as follows:
Figure 754043DEST_PATH_IMAGE026
、
Figure 779768DEST_PATH_IMAGE027
，
wherein, the error covariance matrix of any lattice point m
Figure 718905DEST_PATH_IMAGE028
Is marked as
Figure 386647DEST_PATH_IMAGE029
；
Step 10, reading a model forecast rainfall estimation field to be fused in a target area
Figure 993209DEST_PATH_IMAGE030
Mine and mineReach the outer water level estimation field of extrapolation
Figure 720993DEST_PATH_IMAGE031
Error covariance matrix based on mode prediction
Figure 881847DEST_PATH_IMAGE028
Error covariance matrix of radar extrapolation prediction
Figure 819192DEST_PATH_IMAGE032
Separately constructing each time-next-related precipitation field
Figure 545839DEST_PATH_IMAGE033
Three-dimensional variational objective function of
Figure 585471DEST_PATH_IMAGE034
The concrete formula is as follows:
Figure 764779DEST_PATH_IMAGE035
；
step 11, solving the three-dimensional variational objective function at each time by using a method of gradually iterating to obtain a minimum value
Figure 446427DEST_PATH_IMAGE036
Solutions of the precipitation field when the minimum value is reached
Figure 27581DEST_PATH_IMAGE037
Namely:
Figure 766343DEST_PATH_IMAGE038
。
2. the assimilation fusion method based on radar extrapolation and pattern prediction as claimed in claim 1, wherein in step 2: the station numbered st is associated with a station with coordinates (i,j) pitch of the grid points
Figure 432947DEST_PATH_IMAGE039
And (4) the influence radius is smaller than R, and R is 10 times of grid distance.
3. The assimilation fusion method based on radar extrapolation and pattern prediction as claimed in claim 1, wherein the modification in step 7 comprises: to approximate a fitted curve
Figure 715024DEST_PATH_IMAGE018
Removing the point with the error larger than three times of standard deviation sigma from the mean value, and redrawing each time next
Figure 150685DEST_PATH_IMAGE017
Is averaged to fit the curve
Figure 328856DEST_PATH_IMAGE019
。
4. The assimilation fusion method based on radar extrapolation and pattern prediction as claimed in claim 1, wherein in step 1:
k is the number of precipitation cases, K =1,2, …, K;
ST is the automatic station number, ST =1,2, …, ST;
t is time, t =0,1,2, …, 60;
Figure 482757DEST_PATH_IMAGE001
the cumulative precipitation output is once every 6 minutes, namely the cumulative precipitation of 0, 6, 12, … and 360 minutes;
i is a lattice point number in the x direction, I =1,2, …, I;
j is a grid point number in the y direction, J =1,2, …, J;
i is the total number of lattice points in the x direction;
j is the total number of grid points in the y-direction.
5. The assimilation fusion method based on radar extrapolation and pattern prediction as claimed in claim 1, wherein in step 2:
Figure 99683DEST_PATH_IMAGE040
the distance between the site numbered st and the grid point with coordinates (i, j).
6. The assimilation fusion method based on radar extrapolation and pattern prediction as claimed in claim 1, wherein in step 4:
r is the distance between any grid point m and any grid point n;
Figure 124271DEST_PATH_IMAGE041
is the coordinate of any grid point m;
Figure 535661DEST_PATH_IMAGE042
is the coordinate of any grid point n;
Figure 642770DEST_PATH_IMAGE043
，
Figure 266649DEST_PATH_IMAGE044
，
Figure 208060DEST_PATH_IMAGE045
，
Figure 993614DEST_PATH_IMAGE046
。
7. the assimilation fusion method based on radar extrapolation and pattern prediction as claimed in claim 1, wherein in step 11:
Figure 653265DEST_PATH_IMAGE047
t =0,1,2, …,60 is the precipitation estimation field after each time-next fusion.
CN202110042124.6A 2021-01-13 2021-01-13 Assimilation fusion method based on radar extrapolation and mode prediction Active CN112363168B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202110042124.6A CN112363168B (en) 2021-01-13 2021-01-13 Assimilation fusion method based on radar extrapolation and mode prediction

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202110042124.6A CN112363168B (en) 2021-01-13 2021-01-13 Assimilation fusion method based on radar extrapolation and mode prediction

Publications (2)

Publication Number Publication Date
CN112363168A CN112363168A (en) 2021-02-12
CN112363168B true CN112363168B (en) 2021-03-26

Family

ID=74534935

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202110042124.6A Active CN112363168B (en) 2021-01-13 2021-01-13 Assimilation fusion method based on radar extrapolation and mode prediction

Country Status (1)

Country Link
CN (1) CN112363168B (en)

Families Citing this family (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113640803B (en) * 2021-09-01 2022-07-22 江西师范大学 Short-time quantitative rainfall forecasting method based on echo intensity and echo top height extrapolation
CN113642813B (en) * 2021-10-18 2022-02-11 江苏铨铨信息科技有限公司 Rainfall extrapolation forecasting method based on physical equation
CN114205195B (en) * 2021-12-10 2024-01-26 东南大学 Cross-frequency-band MIMO space domain statistical CSI estimation method

Family Cites Families (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US7542852B1 (en) * 2005-01-25 2009-06-02 Weather Channel Inc Derivation and production of high-resolution, very short-term weather forecasts
US10203410B2 (en) * 2017-05-02 2019-02-12 Honeywell International Inc. Probabilistic weather severity estimation system
CN108549116A (en) * 2018-03-27 2018-09-18 南京恩瑞特实业有限公司 NRIET blending merge forecasting procedure
CN109283505B (en) * 2018-09-03 2022-06-07 南京信息工程大学 Method for correcting divergence phenomenon of radar echo extrapolated image
FR3090894B1 (en) * 2018-12-20 2021-03-05 Thales Sa METHOD AND SYSTEM FOR NEUTRALIZING THE EFFECT OF VIBRATIONS OF A ROTATING AIRCRAFT AIRCRAFT FOR AIRPORT DOPPLER RADAR
CN111428676B (en) * 2020-04-01 2023-04-07 南京信息工程大学 Short-term rainfall prediction method based on sparse correspondence and deep neural network
CN112070286B (en) * 2020-08-25 2023-11-24 贵州黔源电力股份有限公司 Precipitation forecast and early warning system for complex terrain river basin

Also Published As

Publication number Publication date
CN112363168A (en) 2021-02-12

Similar Documents

Publication Publication Date Title
CN112363168B (en) Assimilation fusion method based on radar extrapolation and mode prediction
CN112070286B (en) Precipitation forecast and early warning system for complex terrain river basin
CN107316095B (en) Regional weather drought level prediction method coupled with multi-source data
Mohanty et al. Role of sea surface temperature in modulating life cycle of tropical cyclones over Bay of Bengal
CN106780104A (en) A kind of mean wind direction computational methods based on probability statistics
Kong et al. Assessment of temperature extremes in China using RegCM4 and WRF
Lee et al. Evidence of specific MJO phase occurrence with summertime California Central Valley extreme hot weather
Kang et al. Evaluation of farmland losses from sea level rise and storm surges in the Pearl River Delta region under global climate change
CN105844427A (en) Calculating method for refined assessment on storm tide disaster risks
Park et al. Future changes in precipitation for identified sub‐regions in East Asia using bias‐corrected multi‐RCMs
CN107918713A (en) Flood bivariate based on Copula functions improves joint return period calculation method
JP2004069478A (en) Lightning prediction method
CN116187752A (en) Refined risk assessment method in typhoon disaster process
Rezacova et al. An estimation of the probable maximum precipitation for river basins in the Czech Republic
Chenchen et al. Improvement in the forecasting of heavy rainfall over South China in the DSAEF_LTP model by introducing the intensity of the tropical cyclone
Yu et al. Extreme temperature change of the last 110 years in Changchun, Northeast China
CN110097223B (en) Early warning method for damage of power transmission line under typhoon disaster
CN110619433A (en) Rapid selection method and system for power grid rainstorm numerical mode parameterization scheme
Tian et al. New downscaling prediction models for spring drought in China
CN105426668A (en) Tropical cyclone potential impact evaluation method based on comprehensive intensity index
CN107944466A (en) A kind of rainfall bias correction method based on segmentation thought
Hughes et al. An evaluation of the potential use of satellite rainfall data for input to water resource estimation models in southern Africa
CN115630337A (en) Quantitative evaluation method and system for extreme rainfall attribution based on large-scale climate remote correlation
CN113376711B (en) Method for forecasting lightning frequency based on convolutional neural network
Ivančan-Picek et al. Overview of the first HyMeX special observation period over Croatia

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