CN113515871A - Object-oriented artificial intelligence inspection method for rainfall forecast - Google Patents

Object-oriented artificial intelligence inspection method for rainfall forecast Download PDF

Info

Publication number
CN113515871A
CN113515871A CN202111048855.8A CN202111048855A CN113515871A CN 113515871 A CN113515871 A CN 113515871A CN 202111048855 A CN202111048855 A CN 202111048855A CN 113515871 A CN113515871 A CN 113515871A
Authority
CN
China
Prior art keywords
precipitation
fcst
obs
error
area
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.)
Granted
Application number
CN202111048855.8A
Other languages
Chinese (zh)
Other versions
CN113515871B (en
Inventor
郭洪涛
宋金杰
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.)
Jiangsu Quan Quan Information Technology Co ltd
Original Assignee
Jiangsu Quan Quan Information 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 Jiangsu Quan Quan Information Technology Co ltd filed Critical Jiangsu Quan Quan Information Technology Co ltd
Priority to CN202111048855.8A priority Critical patent/CN113515871B/en
Publication of CN113515871A publication Critical patent/CN113515871A/en
Application granted granted Critical
Publication of CN113515871B publication Critical patent/CN113515871B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F30/00—Computer-aided design [CAD]
    • G06F30/20—Design optimisation, verification or simulation
    • 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
    • G06F2113/00—Details relating to the application field
    • G06F2113/08—Fluids
    • 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)
  • Physics & Mathematics (AREA)
  • Environmental & Geological Engineering (AREA)
  • Theoretical Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • Geometry (AREA)
  • Evolutionary Computation (AREA)
  • Computer Hardware Design (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Atmospheric Sciences (AREA)
  • Biodiversity & Conservation Biology (AREA)
  • Ecology (AREA)
  • Environmental Sciences (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

The invention provides an object-oriented artificial intelligence inspection method for rainfall forecast, which comprises the following specific steps: the method comprises the following steps: reading the precipitation observation field (R obs) And precipitation forecasting field: (R fcst) The data of (a); step two: are respectively pairedR obsAndR fcstcarrying out nine-point Gaussian filtering smoothing; step three: setting precipitation thresholdR 0Retention of not less thanR 0Is/are as followsR obsAndR fcstis less thanR 0All the precipitation is zero; step four: are respectively atR obsAndR fcstidentifying all continuous precipitation areas; step five: based onR obsIn a region of falling water andR fcstcenter of gravity distance of middle precipitation area for precipitation areaMatching of domains; step six: and carrying out error decomposition on each matched pair of the precipitation areas. The inspection accuracy can be improved by dividing different inspection regions by an objective method and then performing precipitation error decomposition in each region.

Description

Object-oriented artificial intelligence inspection method for rainfall forecast
Technical Field
The invention relates to the field of rainfall forecast product inspection, in particular to an object-oriented rainfall forecast artificial intelligence inspection method.
Background
Precipitation is an important physical quantity in meteorology and hydrology, and precipitation forecast is always a key and difficult problem in the fields of scientific research and business. Currently, the precipitation forecast test indexes commonly used by forecasters mainly include ts (score), root mean square error and the like. These indexes strictly follow the principle of point-to-point inspection, and although the mathematical understanding is easier and the connection with the business is tighter, the indexes can only provide the overall evaluation of the forecast performance, and other valuable rainfall space characteristic information, such as the falling area, the shape and the like of a rain strip, is often missed. In order to obtain the information missed by the traditional indexes, a forecaster can only carry out subjective and qualitative analysis by a weather detection method, the workload is large, and the error is uncontrollable. In view of this drawback, Object-oriented precipitation inspection methods, such as cra (continuous Rain area), SAL (Structure, Amplitude, Location), and MODE (Method for Object-based precipitation Evaluation), have been recently developed. On one hand, the MODE calculates the similarity through a fuzzy logic algorithm based on a plurality of characteristic parameters of rainfall, matches the rainfall areas in the forecast field and the observation field one by one, and then calculates the traditional inspection score in sequence. On the other hand, both the CRA and SAL perform tests on a single precipitation region, and quantitatively decompose the prediction error into a position error, an intensity error, a structural error and the like. Although the above test methods have been put into practice, they all have disadvantages. The MODE can only give the matching relation between different observation precipitation areas and different forecast precipitation areas, the CRA and SAL need to subjectively select precipitation areas, and the detection result is very sensitive to the range of the areas. In order to solve the problems, the invention provides an object-oriented quantitative detection technology for the rainfall forecast, which comprises the steps of dividing different detection areas by using an objective method, and then carrying out rainfall error decomposition in each area.
Disclosure of Invention
In order to solve the problems, the invention provides an artificial intelligence inspection method for the object-oriented rainfall forecast, which comprises the steps of dividing different inspection areas by using an objective method, and then carrying out rainfall error decomposition in each area, so that the inspection accuracy can be improved.
An artificial intelligence inspection method for object-oriented rainfall forecast comprises the following specific steps:
the method comprises the following steps: reading precipitation observation fieldR obsAnd precipitation forecast fieldR fcstThe data of (a);
step two: are respectively pairedR obsAndR fcstcarrying out nine-point Gaussian filtering smoothing;
step three: setting precipitation thresholdR 0Retention of not less thanR 0Is/are as followsR obsAndR fcstis less thanR 0All the precipitation is zero;
step four: are respectively atR obsAndR fcstidentifying all continuous precipitation areas;
wherein each successive precipitation zone is identified by a method whereinR obsIn the precipitation area are respectively usedA、B、CAnd the reference numeral … … denotes a reference numeral,R fcstin the precipitation area are respectively useda、b、c… … denotes:
selecting any non-zero value precipitation lattice pointx, yAdding into the grid point queue of the falling water areaZ；
② sequentially judgingx, yWhether the precipitation of the surrounding 8 lattice points is zero or not; if not, addZ(ii) a If the number of the areas is zero, the area identification is finished;
③ pairZThe other grid points in the second step are repeatedly judged until the second step is finishedZJudging all the grid points in the grid;
④Zall the grid points in the area are the coverage range of the continuous precipitation area;
step five: based onR obsIn a region of falling water andR fcstmatching the center of gravity distance of the middle precipitation area with the precipitation area;
wherein the water reducing area matching process is as follows:
(ii) calculating the center of gravity of a single precipitation zonex c , y c )：
Figure RE-DEST_PATH_IMAGE001
（1）
wherein ,r i,j 、x i,j 、y i,j are respectively asi, jPrecipitation, warp coordinates and weft coordinates on grid points;
calculating the distance between the gravity centers of two precipitation areasL：
Figure RE-DEST_PATH_IMAGE002
（2）
Wherein the superscripts obs and fcst respectively representR obsAndR fcsta middle dewatering area;
making two-dimensional matrix by using gravity center distance between every two water falling areas
Figure RE-DEST_PATH_IMAGE003
, wherein n obsAndn fcstare respectively asR obsAndR fcstif the number of the identified continuous precipitation areas is less than the number of the identified continuous precipitation areas
Figure RE-DEST_PATH_IMAGE004
Both for the minimum value of each row and for the minimum value of each column, thenI, JThe two corresponding water reducing areas are a matched pair;
step six: and carrying out error decomposition on each matched pair of the precipitation areas.
As a further improvement of the present invention, the error decomposition process in the step six is as follows:
due to the areaAAnd areacCoverage is generally not the same and so willR obsMiddle zoneAOuter andR fcstmiddle zonecAll the other precipitation is zero to obtain the precipitation field
Figure RE-DEST_PATH_IMAGE005
And
Figure RE-DEST_PATH_IMAGE006
；
② calculating the total errorE total：
Figure RE-DEST_PATH_IMAGE007
（3）
wherein ,Nis the total number of grid points;
③ Retention
Figure RE-63961DEST_PATH_IMAGE005
Without change, first use the regioncThe center of gravity of the pair is a midpoint and the angle increment of the pair is 1 DEG
Figure RE-636894DEST_PATH_IMAGE006
Rotating, and increasing distance by 1 time of grid distance
Figure RE-823156DEST_PATH_IMAGE006
Carrying out translation; after each rotation and translation, recalculation occurs
Figure RE-599789DEST_PATH_IMAGE005
And
Figure RE-7636DEST_PATH_IMAGE006
the overall error between; through traversal, the corresponding rotation angle when the total error is minimum is foundθAnd a translation vector Δx,Δy；
Retention
Figure RE-485891DEST_PATH_IMAGE005
Unchanged, will be original
Figure RE-205454DEST_PATH_IMAGE006
RotateθAngle is obtained
Figure RE-DEST_PATH_IMAGE008
First, calculate
Figure RE-204022DEST_PATH_IMAGE008
And
Figure RE-325431DEST_PATH_IMAGE005
total error ofE 1：
Figure RE-DEST_PATH_IMAGE009
（4）
Recalculating orientation errorE rotation：
Figure RE-DEST_PATH_IMAGE010
（5）
Hold
Figure RE-302483DEST_PATH_IMAGE005
Without change, will
Figure RE-260074DEST_PATH_IMAGE008
Translation deltax,ΔyVector derivation
Figure RE-DEST_PATH_IMAGE011
First, calculate
Figure RE-311600DEST_PATH_IMAGE011
And
Figure RE-975931DEST_PATH_IMAGE005
total error ofE 2：
Figure RE-DEST_PATH_IMAGE012
（6）
Recalculating position errorsE shift：
Figure RE-DEST_PATH_IMAGE013
（7）
wherein E 1Was obtained from (iv).
Strength of calculationError in degreeE amplitude：
Figure RE-DEST_PATH_IMAGE014
（8）
Seventhly, calculating the shape errorE pattern：
Figure RE-DEST_PATH_IMAGE015
（9）
wherein E 2AndE amplituderespectively obtained by fifthly and sixthly;
by the above calculation, the overall error is decomposed into a linear combination of the orientation error, the position error, the intensity error, and the shape error, that is:
Figure RE-DEST_PATH_IMAGE016
（10）。
the invention provides an object-oriented artificial intelligence inspection method for rainfall forecast, which comprises the following specific steps: the method comprises the following steps: reading the precipitation observation field (R obs) And precipitation forecasting field: (R fcst) The data of (a); step two: are respectively pairedR obsAndR fcstcarrying out nine-point Gaussian filtering smoothing; step three: setting precipitation thresholdR 0Retention of not less thanR 0Is/are as followsR obsAndR fcstis less thanR 0All the precipitation is zero; step four: are respectively atR obsAndR fcstidentifying all continuous precipitation areas; step five: based onR obsIn a region of falling water andR fcstmatching the center of gravity distance of the middle precipitation area with the precipitation area; step six: and carrying out error decomposition on each matched pair of the precipitation areas. The inspection accuracy can be improved by dividing different inspection regions by an objective method and then performing precipitation error decomposition in each region.
Drawings
FIG. 1 is a technical architecture and execution flow diagram;
FIG. 2 is a flow chart of dewatering area matching;
FIG. 3a is a view of an example of a raw precipitation observation field in mm;
FIG. 3b shows the original precipitation forecast field in mm of the example;
FIG. 4a is a graph of the embodiment after the precipitation observation field is smoothed, in mm;
FIG. 4b is the embodiment of the precipitation forecast field after smoothing, in mm;
FIG. 5a shows the case where the precipitation is extracted at a precipitation observation field of 2mm or more in the example;
FIG. 5b shows the case where the precipitation forecast field of the embodiment extracts precipitation of 2mm or more;
FIG. 6a shows the identification and numbering of the precipitation area of the continuous observation field in mm according to the embodiment;
FIG. 6b is a diagram of the identification and numbering of the rainfall areas of the continuous forecasting field in units of mm according to the embodiment;
FIG. 7a is a view of a precipitation area a of the precipitation observation field according to the embodiment;
FIG. 7b illustrates the forecast rainfall area A of the rainfall forecast farm of the embodiment;
FIG. 7c shows the precipitation observation field in case of precipitation zone b;
FIG. 7d illustrates the forecast rainfall area B of the rainfall forecast farm of the embodiment;
FIG. 7e illustrates the precipitation observation field in case of precipitation zone b;
FIG. 7f shows the forecast rainfall area B of the rainfall forecast site of the embodiment.
Detailed Description
The invention is described in further detail below with reference to the following detailed description and accompanying drawings:
the work flow chart of the invention is shown in figure 1, and the invention provides an artificial intelligence inspection method for object-oriented precipitation forecast, which specifically comprises the following steps:
first, reading the precipitation observation field (R obs) And precipitation forecasting field: (R fcst) The data of (1). Both are two-dimensional spatial lattice fields, warp: (xDirection), weft direction (yDirection) the number of grid points is the same, and the grid intervals are consistent;
second, respectively toR obsAndR fcstand carrying out nine-point Gaussian filtering smoothing processing. Not only isolated and sporadic precipitation points are removed, but also random disturbance errors are reduced. Wherein the nine-point Gaussian filter smoothing operator is:
Figure RE-DEST_PATH_IMAGE017
（11）；
third, a precipitation threshold is setR 0Retention of not less thanR 0Is/are as followsR obsAndR fcstis less thanR 0All the precipitation of (a) is taken to be zero. From the angle of the correlation with meteorological and hydrological disasters, stronger rainfall is preferentially considered, and weaker rainfall is ignored;
the fourth step is respectively atR obsAndR fcstidentifies all successive precipitation zones.R obsIn the precipitation area are respectively usedA、B、CAnd the reference numeral … … denotes a reference numeral,R fcstin the precipitation area are respectively useda、b、cAnd … …. The identification method of each successive precipitation zone is shown in fig. 2:
selecting any non-zero value precipitation lattice points (x, y) Adding a grid queue of falling water areas (Z）；
(II) sequentially judgingx, y) Whether the precipitation of the surrounding 8 lattice points is zero or not; if not, addZ(ii) a If the number of the areas is zero, the area identification is finished;
③ pairZThe other grid points in the second step are repeatedly judged until the second step is finishedZJudging all the grid points in the grid;
④Zall the grid points in the area are the coverage range of the continuous precipitation area;
the fifth step is based onR obsIn a region of falling water andR fcstand matching the center of gravity distance of the middle water falling area with the water falling area. The matching process is as follows:
(ii) calculating the center of gravity of a single precipitation zonex c , y c )：
Figure RE-890929DEST_PATH_IMAGE001
,（12）
wherein ,r i,j 、x i,j 、y i,j respectively is (i, j) Precipitation, warp coordinates and weft coordinates on grid points;
calculating the distance between the centers of gravity of two precipitation areas (L）：
Figure RE-273500DEST_PATH_IMAGE002
,（13）
Wherein the superscripts obs and fcst respectively representR obsAndR fcsta middle dewatering area;
making two-dimensional matrix by using gravity center distance between every two water falling areas
Figure RE-1153DEST_PATH_IMAGE003
, wherein n obsAndn fcstare respectively asR obsAndR fcstif the number of the identified continuous precipitation areas is less than the number of the identified continuous precipitation areas
Figure RE-49485DEST_PATH_IMAGE004
Both for the minimum value of each row and for the minimum value of each column, thenI, JThe two corresponding water reducing areas are a matched pair;
and sixthly, carrying out error decomposition on each matched pair of the water reducing areas. To match with the pairA↔cFor example, the specific decomposition process is as follows:
due to the areaAAnd areacCoverage is generally not the same and so willR obsMiddle zoneAOuter andR fcstmiddle zonecAll the other precipitation is zero to obtain the precipitation field
Figure RE-587914DEST_PATH_IMAGE005
And
Figure RE-503786DEST_PATH_IMAGE006
；
(ii) calculating the total error: (E total）：
Figure RE-989125DEST_PATH_IMAGE007
,（14）
wherein ,Nis the total number of grid points;
③ Retention
Figure RE-346157DEST_PATH_IMAGE005
Without change, first use the regioncThe center of gravity of the pair is a midpoint and the angle increment of the pair is 1 DEG
Figure RE-570334DEST_PATH_IMAGE006
Rotating, and increasing distance by 1 time of grid distance
Figure RE-458656DEST_PATH_IMAGE006
Carrying out translation; after each rotation and translation, recalculation occurs
Figure RE-734304DEST_PATH_IMAGE005
And
Figure RE-539318DEST_PATH_IMAGE006
the overall error between; through traversal, the corresponding rotation angle when the total error is minimum is foundθAnd a translation vector (Δ)x,Δy)；
Retention
Figure RE-622811DEST_PATH_IMAGE005
Unchanged, will be original
Figure RE-513276DEST_PATH_IMAGE006
RotateθAngle is obtained
Figure RE-324106DEST_PATH_IMAGE008
First, calculate
Figure RE-999938DEST_PATH_IMAGE008
And
Figure RE-506530DEST_PATH_IMAGE005
total error of (1: (E 1）：
Figure RE-635023DEST_PATH_IMAGE009
，（15）
Recalculating orientation error (E rotation）：
Figure RE-515123DEST_PATH_IMAGE010
。（16）
Hold
Figure RE-763571DEST_PATH_IMAGE005
Without change, will
Figure RE-375818DEST_PATH_IMAGE008
Translation (Delta)x,Δy) Vector derivation
Figure RE-991607DEST_PATH_IMAGE011
First, calculate
Figure RE-412749DEST_PATH_IMAGE011
And
Figure RE-818498DEST_PATH_IMAGE005
total error of (1: (E 2）：
Figure RE-663963DEST_PATH_IMAGE012
，（17）
Recalculating the position error (E shift）：
Figure RE-953999DEST_PATH_IMAGE013
，（18）
wherein E 1Was obtained from (iv).
Calculating the intensity errorE amplitude）：
Figure RE-175902DEST_PATH_IMAGE014
。（19）
Calculating the shape error (c)E pattern）：
Figure RE-680833DEST_PATH_IMAGE015
，（20）
wherein E 2AndE amplituderespectively obtained by fifthly and sixthly;
by the above calculation, the overall error is decomposed into a linear combination of the orientation error, the position error, the intensity error, and the shape error, that is:
Figure RE-434549DEST_PATH_IMAGE016
。（21）
as a specific embodiment of the invention:
for 7 months and 27 days 00 in 2021: 00 universal time ERA5 is used for collecting average 24-hour rainfall forecast products for inspection (inspection area: 25-45 degrees N, 110-150 degrees E);
firstly, reading a rainfall observation field and a rainfall forecast field (figure 3a is the condition of an original rainfall observation field of the embodiment in unit mm; figure 3b is the condition of the original rainfall forecast field of the embodiment in unit mm);
secondly, nine-point Gaussian filtering smoothing is respectively carried out on the rainfall observation field and the rainfall forecast field (fig. 4a shows the smooth condition of the rainfall observation field in unit mm; fig. 4b shows the smooth condition of the rainfall forecast field in unit mm);
thirdly, setting a precipitation threshold (2 mm), and only keeping the precipitation which is larger than or equal to the threshold (fig. 5a shows that the precipitation extracted by the precipitation observation field is larger than or equal to 2 mm; fig. 5b shows that the precipitation extracted by the precipitation forecast field is larger than or equal to 2 mm; particularly, the area surrounded by a solid line);
fourthly, all the continuous precipitation areas are identified (FIG. 6a shows the identification and numbering condition of the precipitation areas of the continuous observation field in unit mm, and FIG. 6b shows the identification and numbering condition of the precipitation areas of the continuous forecast field in unit mm, and the observation precipitation areasA~EForecasting precipitation areaa~c）；
The fifth step, the result of the precipitation area matching based on the gravity center distance (as shown in the following table;A↔a，B↔b，E↔c(ii) a C and D have no matching pair, which means that the two water-reducing areas are not forecasted);
Figure RE-DEST_PATH_IMAGE018
sixthly, performing error decomposition on each matched pair of the rainfall areas to obtain a specific error value of each matched pair, wherein a figure 7a shows the rainfall area a condition of the rainfall observation field of the embodiment; FIG. 7b illustrates the forecast rainfall area A of the rainfall forecast farm of the embodiment; FIG. 7c shows the precipitation observation field in case of precipitation zone b; FIG. 7d illustrates the forecast rainfall area B of the rainfall forecast farm of the embodiment; FIG. 7e illustrates the precipitation observation field in case of precipitation zone b; FIG. 7f shows the forecast rainfall area B of the rainfall forecast field of the embodiment, and it can be seen from the corresponding diagramA↔a，B↔b，E↔cThe smaller error corresponds to the result of the table of the fifth step.
The above description is only a preferred embodiment of the present invention, and is not intended to limit the present invention in any way, but any modifications or equivalent variations made according to the technical spirit of the present invention are within the scope of the present invention as claimed.

Claims (2)

1. An artificial intelligence inspection method for object-oriented rainfall forecast comprises the following specific steps:
the method comprises the following steps: reading precipitation observation fieldR obsAnd precipitation forecast fieldR fcstThe data of (a);
step two: are respectively pairedR obsAndR fcstcarrying out nine-point Gaussian filtering smoothing;
step three: setting precipitation thresholdR 0Retention of not less thanR 0Is/are as followsR obsAndR fcstis less thanR 0All the precipitation is zero;
step four: are respectively atR obsAndR fcstidentifying all continuous precipitation areas;
wherein each successive precipitation zone is identified by a method whereinR obsIn the precipitation area are respectively usedA、B、CAnd the reference numeral … … denotes a reference numeral,R fcstin the precipitation area are respectively useda、b、c… … denotes:
selecting any non-zero value precipitation lattice pointx, yAdding into the grid point queue of the falling water areaZ；
② sequentially judgingx, yWhether the precipitation of the surrounding 8 lattice points is zero or not; if not, addZ(ii) a If the number of the areas is zero, the area identification is finished;
③ pairZThe other grid points in the second step are repeatedly judged until the second step is finishedZJudging all the grid points in the grid;
④Zall the grid points in the area are the coverage range of the continuous precipitation area;
step five: based onR obsIn a region of falling water andR fcstmatching the center of gravity distance of the middle precipitation area with the precipitation area;
wherein the water reducing area matching process is as follows:
(ii) calculating the center of gravity of a single precipitation zonex c , y c )：
Figure 220989DEST_PATH_IMAGE001
（1）
wherein ,r i,j 、x i,j 、y i,j are respectively asi, jLattice pointsPrecipitation, warp-wise coordinates and weft-wise coordinates;
calculating the distance between the gravity centers of two precipitation areasL：
Figure 388665DEST_PATH_IMAGE002
（2）
Wherein the superscripts obs and fcst respectively representR obsAndR fcsta middle dewatering area;
making two-dimensional matrix by using gravity center distance between every two water falling areas
Figure 134904DEST_PATH_IMAGE003
, wherein n obsAndn fcstare respectively asR obsAndR fcstif the number of the identified continuous precipitation areas is less than the number of the identified continuous precipitation areas
Figure 958504DEST_PATH_IMAGE004
Both for the minimum value of each row and for the minimum value of each column, thenI, JThe two corresponding water reducing areas are a matched pair;
step six: and carrying out error decomposition on each matched pair of the precipitation areas.
2. The method of claim 1, wherein the method comprises:
the error decomposition process in the sixth step comprises the following steps:
due to the areaAAnd areacCoverage is generally not the same and so willR obsMiddle zoneAOuter andR fcstmiddle zonecAll the other precipitation is zero to obtain the precipitation field
Figure 877918DEST_PATH_IMAGE005
And
Figure 103363DEST_PATH_IMAGE006
；
② calculating the total errorE total：
Figure 82821DEST_PATH_IMAGE007
（3）
wherein ,Nis the total number of grid points;
③ Retention
Figure 393716DEST_PATH_IMAGE005
Without change, first use the regioncThe center of gravity of the pair is a midpoint and the angle increment of the pair is 1 DEG
Figure 119751DEST_PATH_IMAGE006
Rotating, and increasing distance by 1 time of grid distance
Figure 934123DEST_PATH_IMAGE006
Carrying out translation; after each rotation and translation, recalculation occurs
Figure 350061DEST_PATH_IMAGE005
And
Figure 882674DEST_PATH_IMAGE006
the overall error between; through traversal, the corresponding rotation angle when the total error is minimum is foundθAnd a translation vector Δx,Δy；
Retention
Figure 409470DEST_PATH_IMAGE005
Unchanged, will be original
Figure 78349DEST_PATH_IMAGE006
RotateθAngle is obtained
Figure 868450DEST_PATH_IMAGE008
First, firstComputing
Figure 950676DEST_PATH_IMAGE008
And
Figure 750005DEST_PATH_IMAGE005
total error ofE 1：
Figure 804548DEST_PATH_IMAGE009
（4）
Recalculating orientation errorE rotation：
Figure 499972DEST_PATH_IMAGE010
（5）
Hold
Figure 800984DEST_PATH_IMAGE005
Without change, will
Figure 607266DEST_PATH_IMAGE008
Translation deltax,ΔyVector derivation
Figure 578633DEST_PATH_IMAGE011
First, calculate
Figure 444958DEST_PATH_IMAGE011
And
Figure 439459DEST_PATH_IMAGE005
total error ofE 2：
Figure 846170DEST_PATH_IMAGE012
（6）
Recalculating position errorsE shift：
Figure 609726DEST_PATH_IMAGE013
（7）
wherein E 1Is obtained from the fourth step;
calculating the error of intensityE amplitude：
Figure 646953DEST_PATH_IMAGE014
（8）
Seventhly, calculating the shape errorE pattern：
Figure 191066DEST_PATH_IMAGE015
（9）
wherein E 2AndE amplituderespectively obtained by fifthly and sixthly;
by the above calculation, the overall error is decomposed into a linear combination of the orientation error, the position error, the intensity error, and the shape error, that is:
Figure 339151DEST_PATH_IMAGE016
（10）。
CN202111048855.8A 2021-09-08 2021-09-08 Object-oriented precipitation prediction artificial intelligence inspection method Active CN113515871B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202111048855.8A CN113515871B (en) 2021-09-08 2021-09-08 Object-oriented precipitation prediction artificial intelligence inspection method

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202111048855.8A CN113515871B (en) 2021-09-08 2021-09-08 Object-oriented precipitation prediction artificial intelligence inspection method

Publications (2)

Publication Number Publication Date
CN113515871A true CN113515871A (en) 2021-10-19
CN113515871B CN113515871B (en) 2023-10-20

Family

ID=78063028

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202111048855.8A Active CN113515871B (en) 2021-09-08 2021-09-08 Object-oriented precipitation prediction artificial intelligence inspection method

Country Status (1)

Country Link
CN (1) CN113515871B (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116224472A (en) * 2023-02-21 2023-06-06 南京气象科技创新研究院 Tropical cyclone rainfall forecast and inspection method based on regional tree

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JPH10268065A (en) * 1997-03-24 1998-10-09 Nippon Hoso Kyokai <Nhk> Rainfall prediction method and rainfall decrease alert device
CN111624680A (en) * 2020-05-13 2020-09-04 苏翔 Precipitation forecast displacement correction algorithm based on object diagnosis
CN113033957A (en) * 2021-02-26 2021-06-25 兰州中心气象台(兰州干旱生态环境监测预测中心) Multi-mode rainfall forecast and real-time dynamic inspection and evaluation system

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JPH10268065A (en) * 1997-03-24 1998-10-09 Nippon Hoso Kyokai <Nhk> Rainfall prediction method and rainfall decrease alert device
CN111624680A (en) * 2020-05-13 2020-09-04 苏翔 Precipitation forecast displacement correction algorithm based on object diagnosis
CN113033957A (en) * 2021-02-26 2021-06-25 兰州中心气象台(兰州干旱生态环境监测预测中心) Multi-mode rainfall forecast and real-time dynamic inspection and evaluation system

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116224472A (en) * 2023-02-21 2023-06-06 南京气象科技创新研究院 Tropical cyclone rainfall forecast and inspection method based on regional tree
CN116224472B (en) * 2023-02-21 2024-01-19 南京气象科技创新研究院 Tropical cyclone rainfall forecast and inspection method based on regional tree

Also Published As

Publication number Publication date
CN113515871B (en) 2023-10-20

Similar Documents

Publication Publication Date Title
Hatir et al. Deep learning-based weathering type recognition in historical stone monuments
Zeraatpisheh et al. Digital mapping of soil properties using multiple machine learning in a semi-arid region, central Iran
Nabiollahi et al. Assessing agricultural salt-affected land using digital soil mapping and hybridized random forests
Liu et al. Using t-distributed Stochastic Neighbor Embedding (t-SNE) for cluster analysis and spatial zone delineation of groundwater geochemistry data
CN101915764B (en) Road surface crack detection method based on dynamic programming
CN103810704B (en) Based on support vector machine and the SAR image change detection of discriminative random fields
Nagini et al. Agriculture yield prediction using predictive analytic techniques
Lorenz et al. Predicting the US drought monitor using precipitation, soil moisture, and evapotranspiration anomalies. Part II: Intraseasonal drought intensification forecasts
CN111428989A (en) TOPSIS model-based mining vestige comprehensive evaluation method
Jiao et al. Next-generation remote sensing and prediction of sand and dust storms: State-of-the-art and future trends
CN113515871A (en) Object-oriented artificial intelligence inspection method for rainfall forecast
CN112348264A (en) Carbon steel corrosion rate prediction method based on random forest algorithm
Ntale et al. Prediction of East African seasonal rainfall using simplex canonical correlation analysis
CN113537108A (en) Crop chlorophyll content inversion method based on unmanned aerial vehicle hyperspectral remote sensing
CN115293577A (en) Method for analyzing chemical control factors of underground water in alpine-cold flow region based on machine learning
Arnault et al. A tropical Atlantic dynamics analysis by combining machine learning and satellite data
CN117171533B (en) Real-time acquisition and processing method and system for geographical mapping operation data
Zhu et al. Prediction of deep soil water content (0–5 m) with in-situ and remote sensing data
Potvin et al. An iterative storm segmentation and classification algorithm for convection-allowing models and gridded radar analyses
Herho A univariate extreme value analysis and change point detection of monthly discharge in Kali Kupang, Central Java, Indonesia
El-Rawy et al. Assessing and segmenting salt-affected soils using in-situ EC measurements, remote sensing, and a modified deep learning MU-NET convolutional neural network
CN108694382B (en) Soil pH classification method based on ultra-wideband radar sensor
CN116340838A (en) Classifying method of inner Mongolian grassland based on multi-feature fusion self-attention mechanism
Karydis Environmental quality assessment based on the analysis of extreme values: a practical approach for evaluating eutrophication
CN109444055A (en) A kind of EO-1 hyperion calculation method of alkaline land soil salt content

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
CB02 Change of applicant information

Address after: Room 906, Building 4, No. 18 Jialing Jiangdong Street, Jianye District, Nanjing City, Jiangsu Province, 210000

Applicant after: Zhongke Xingtu Deep Sea Technology Co.,Ltd.

Address before: Room 905 and 906, 9 / F, building 4, No. 18, Jiangdong Street, Jialing, Jianye District, Nanjing City, Jiangsu Province, 210000

Applicant before: JIANGSU QUAN QUAN INFORMATION TECHNOLOGY CO.,LTD.

CB02 Change of applicant information
GR01 Patent grant
GR01 Patent grant