CN112488385B - Precipitation prediction correction method and device based on multi-mode fusion - Google Patents
Precipitation prediction correction method and device based on multi-mode fusion Download PDFInfo
- Publication number
- CN112488385B CN112488385B CN202011362926.7A CN202011362926A CN112488385B CN 112488385 B CN112488385 B CN 112488385B CN 202011362926 A CN202011362926 A CN 202011362926A CN 112488385 B CN112488385 B CN 112488385B
- Authority
- CN
- China
- Prior art keywords
- precipitation
- mode
- prediction
- threshold
- forecasting
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000001556 precipitation Methods 0.000 title claims abstract description 246
- 230000004927 fusion Effects 0.000 title claims abstract description 70
- 238000000034 method Methods 0.000 title claims abstract description 67
- 238000012937 correction Methods 0.000 title claims abstract description 66
- 238000007689 inspection Methods 0.000 claims abstract description 27
- 238000004364 calculation method Methods 0.000 claims description 23
- 238000012360 testing method Methods 0.000 claims description 16
- 230000008033 biological extinction Effects 0.000 claims description 13
- 238000012545 processing Methods 0.000 claims description 9
- 230000008569 process Effects 0.000 abstract description 11
- 230000010354 integration Effects 0.000 abstract description 8
- 238000005516 engineering process Methods 0.000 abstract description 7
- 230000000694 effects Effects 0.000 description 9
- 238000011160 research Methods 0.000 description 8
- 238000004458 analytical method Methods 0.000 description 7
- 238000001604 Rao's score test Methods 0.000 description 5
- 241000219094 Vitaceae Species 0.000 description 4
- 230000008901 benefit Effects 0.000 description 4
- 235000021021 grapes Nutrition 0.000 description 4
- 230000009471 action Effects 0.000 description 3
- 238000010586 diagram Methods 0.000 description 3
- 230000006872 improvement Effects 0.000 description 3
- 238000007500 overflow downdraw method Methods 0.000 description 3
- 230000008092 positive effect Effects 0.000 description 3
- 239000011800 void material Substances 0.000 description 3
- 230000009286 beneficial effect Effects 0.000 description 2
- 230000008030 elimination Effects 0.000 description 2
- 238000003379 elimination reaction Methods 0.000 description 2
- 238000009825 accumulation Methods 0.000 description 1
- 230000032683 aging Effects 0.000 description 1
- 230000003321 amplification Effects 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 230000001149 cognitive effect Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 230000002349 favourable effect Effects 0.000 description 1
- 230000005764 inhibitory process Effects 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 238000011835 investigation Methods 0.000 description 1
- 230000007774 longterm Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000003199 nucleic acid amplification method Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 230000002093 peripheral effect Effects 0.000 description 1
- 230000009897 systematic effect Effects 0.000 description 1
- 230000001550 time effect Effects 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0639—Performance analysis of employees; Performance analysis of enterprise or organisation operations
- G06Q10/06393—Score-carding, benchmarking or key performance indicator [KPI] analysis
Landscapes
- Business, Economics & Management (AREA)
- Human Resources & Organizations (AREA)
- Engineering & Computer Science (AREA)
- Strategic Management (AREA)
- Economics (AREA)
- Development Economics (AREA)
- Entrepreneurship & Innovation (AREA)
- Operations Research (AREA)
- Game Theory and Decision Science (AREA)
- Marketing (AREA)
- Educational Administration (AREA)
- Quality & Reliability (AREA)
- Tourism & Hospitality (AREA)
- Physics & Mathematics (AREA)
- General Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a precipitation prediction correction method and device based on multi-mode fusion, wherein the method comprises the following steps: reading historical data of a plurality of forecasting modes in a historical time period, wherein the historical data comprises historical forecasting data and historical observing data; performing scoring inspection on each forecasting mode according to the historical data to obtain the score of each forecasting mode; and on the basis of taking the prediction mode with the highest score as a background field, correcting the background field by adopting a mode of threshold fusion and threshold nulling. The invention is based on the technology for forecasting the fusion grid precipitation of multiple forecasting modes, reduces the missing report and blank report situations in intelligent grid precipitation forecasting as much as possible, improves the performance error in the multi-mode integration process, reduces the performance error of the modes, improves the missing report of the precipitation mode with good forecasting performance, and further improves the accuracy rate of precipitation forecasting under the condition that other precipitation modes hit.
Description
Technical Field
The invention relates to the technical field of weather forecast, in particular to a rainfall forecast correction method and device based on multi-mode fusion.
Background
The intelligent grid objective precipitation forecasting technology is a statistical post-treatment release technology for various numerical model products, mainly adopts single-mode precipitation release and integrated forecasting or multi-mode precipitation integration, and mainly adopts a multi-mode precipitation integration method in the prior art.
In the prior art, when integrating a plurality of numerical mode products, performance errors of the modes are also unavoidable. The detection shows that the rainfall forecasting performances of different modes have larger difference, so that the situation that the rainfall product with good overall forecasting performance is missed and other rainfall products hit exists, thereby causing performance errors in the multi-mode integration process and affecting the accuracy rate of rainfall forecasting.
Disclosure of Invention
In order to solve at least one of the above technical problems, the present disclosure provides a precipitation prediction correction method and device based on multimode fusion.
In a first aspect, the present invention provides a precipitation prediction correction method based on multimode fusion, the method comprising:
reading historical data of a plurality of forecasting modes in a historical time period, wherein the historical data comprises historical forecasting data and historical observing data;
performing scoring inspection on each forecasting mode according to the historical data to obtain the score of each forecasting mode;
And on the basis of taking the prediction mode with the highest score as a background field, correcting the background field by adopting a mode of threshold fusion and threshold nulling.
Optionally, performing a scoring test on each prediction mode according to the historical data to obtain a score of each prediction mode, including:
under the preset check limiting condition, calculating TS scores of all forecasting modes according to the historical data.
Optionally, on the basis of taking the prediction mode with the highest score as a background field, correcting the background field by adopting a mode of threshold fusion and threshold blanking, including:
taking the forecast mode with the highest score as a background field;
in one grid of the background field, when the predicted precipitation of the prediction mode with the highest score is smaller than a first precipitation threshold value and the predicted precipitation of other prediction modes except the prediction mode with the highest score is larger than or equal to the first precipitation threshold value, calculating the prediction failure rate of each prediction mode in the prediction modes inconsistent with the prediction mode with the highest score under the respective precipitation threshold value;
judging whether all the forecasting failure rates are lower than a preset fusion threshold value, if yes, fusing the grids, otherwise, not processing the grids.
Optionally, on the basis of taking the prediction mode with the highest score as a background field, correcting the background field by adopting a mode of threshold fusion and threshold blanking, including:
taking the forecast mode with the highest score as a background field;
in one grid of the background field, the forecast precipitation of the forecast mode with the highest score is larger than zero and smaller than a second precipitation threshold, and the forecast precipitation exists in other forecast modes except for the forecast mode with the highest score, and the forecast accuracy of the forecast modes corresponding to all the forecast precipitation under the respective precipitation thresholds is calculated;
judging whether the prediction accuracy rate higher than a preset extinction threshold exists in all the prediction accuracy rates, if so, carrying out extinction on the grids, otherwise, not processing the grids.
Optionally, the method further comprises:
and correcting the background field by adopting a threshold fusion and threshold blanking mode, and then carrying out grading inspection to obtain the corrected grading.
Optionally, performing scoring inspection after correcting the background field by adopting a mode of threshold fusion and threshold nulling, including:
and correcting the background field by adopting a mode of threshold fusion and threshold blanking, and then carrying out scoring inspection on a preset time length.
Optionally, the method further comprises:
judging whether the score after correction reaches a preset standard score, if so, carrying out precipitation prediction by adopting a corrected background field, otherwise, adjusting the first precipitation threshold value and/or the second precipitation threshold value and/or the precipitation high threshold value and/or the precipitation low threshold value and/or the preset fusion threshold value and/or the preset emptying threshold value, and correcting the background field by adopting a threshold fusion and threshold emptying mode.
In a second aspect, the present invention provides a precipitation prediction correction device based on multimode fusion, the device comprising: the system comprises a data acquisition module, a scoring calculation module and a precipitation correction module, wherein,
the data acquisition module is used for reading historical data of a plurality of forecasting modes in a historical time period, wherein the historical data comprises historical forecasting data and historical observation data;
the scoring calculation module is used for scoring and checking each forecasting mode according to the historical data acquired by the data acquisition module to acquire the score of each forecasting mode;
and the rainfall correction module is used for correcting the background field by adopting a mode of threshold fusion and threshold extinction on the basis of taking the prediction mode with the highest score as the background field.
Optionally, the score calculating module is specifically configured to calculate the TS score of each prediction mode according to the historical data under a preset check limit.
Optionally, the precipitation correction module comprises a background selection unit, a failure calculation unit and a fusion judgment unit, wherein,
the background selection unit is used for taking the forecast mode with the highest score as a background field;
the failure calculation unit is used for calculating the failure rate of prediction under the respective precipitation amount threshold value of each prediction mode in the prediction modes inconsistent with the prediction mode with the highest score when the prediction precipitation amount of the prediction mode with the highest score is smaller than the first precipitation amount threshold value and the prediction precipitation amount of other prediction modes except the prediction mode with the highest score is larger than or equal to the first precipitation amount threshold value in one grid of the background field;
and the fusion judging unit is used for judging whether all the forecasting failure rates are lower than a preset fusion threshold value, if yes, fusing the grids, and otherwise, not processing the grids.
Optionally, the precipitation correction module comprises a background selection unit, a success calculation unit and a void judgment unit,
The background selection unit is used for taking the forecast mode with the highest score as a background field;
the success calculation unit calculates the prediction accuracy of all the prediction modes corresponding to the non-prediction precipitation under each precipitation low threshold value, wherein the prediction precipitation of the prediction mode with the highest score is larger than zero and smaller than the second precipitation threshold value in one grid of the background field, and the non-prediction precipitation exists in other prediction modes except for the prediction mode with the highest score;
the blanking judging unit is used for judging whether the prediction accuracy is higher than a preset blanking threshold, if yes, blanking the grid, otherwise, not processing the grid
Optionally, the scoring calculation module is further configured to score and test the corrected forecast mode obtained by the precipitation correction module, so as to obtain a corrected score.
Optionally, the score calculating module is further configured to perform score inspection on the corrected prediction mode obtained by the precipitation correction module for a preset time length, so as to obtain a corrected score.
Optionally, the device further includes a correction improvement module, configured to determine whether the corrected score reaches a preset standard score, if yes, perform precipitation prediction in a corrected prediction mode, otherwise, adjust the first precipitation threshold and/or the second precipitation threshold and/or the precipitation high threshold and/or the precipitation low threshold and/or the preset fusion threshold and/or the preset emptying threshold, and correct the background field.
In a third aspect, the present invention provides a readable storage medium having executable instructions thereon that, when executed, cause a computer to perform any of the methods as included in the first aspect.
In a fourth aspect, the present invention provides a computing device comprising: one or more processors, a memory, and a program, wherein the one or more programs are stored in the memory and configured to perform any of the methods as included in the first aspect by the one or more processors.
Compared with the prior art, the invention has at least the following beneficial effects:
the invention is based on the technology for forecasting the fusion grid precipitation of a plurality of forecasting modes, reduces the missing report and blank report situations in the intelligent grid precipitation forecasting as much as possible, improves the performance error in the multi-mode integration process, reduces the performance error of the modes, improves the missing report of the mode with good forecasting performance, and further improves the accuracy rate of precipitation forecasting under the condition of hit of other modes.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings that are required in the embodiments or the description of the prior art will be briefly described, and it is obvious that the drawings in the following description are some embodiments of the present invention, and other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.
FIG. 1 is a schematic flow chart of a precipitation prediction correction method based on multimode fusion according to an embodiment of the invention;
FIG. 2 is a schematic diagram of fusion nulling provided by an embodiment of the present invention;
fig. 3 is a block diagram of a precipitation prediction correction device based on multimode fusion according to an embodiment of the invention.
Detailed Description
For the purpose of making the objects, technical solutions and advantages of the embodiments of the present invention more apparent, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention, and it is apparent that the described embodiments are some embodiments of the present invention, but not all embodiments, and all other embodiments obtained by those skilled in the art without making any inventive effort based on the embodiments of the present invention are within the scope of protection of the present invention.
As shown in fig. 1, an embodiment of the present invention provides a precipitation prediction correction method based on multimode fusion, where the method includes:
reading historical data of a plurality of forecasting modes in a historical time period, wherein the historical data comprises historical forecasting data and historical observing data;
Performing scoring inspection on each forecasting mode according to the historical data to obtain the score of each forecasting mode;
on the basis of taking the prediction mode with the highest score as a background field, correcting the background field by adopting a mode of threshold fusion and threshold nulling to obtain a corrected prediction mode.
The invention obtains the corrected prediction mode, namely the new prediction mode, based on the technology for predicting and releasing the grid precipitation by combining the multiple prediction modes, reduces the situations of missing report and empty report in the intelligent grid precipitation prediction as much as possible, improves the performance error in the multi-mode integration process, reduces the performance error of the mode, improves the situation of missing report of the mode with good prediction performance and hit of other modes, and further improves the accuracy rate of precipitation prediction.
The invention will be described with reference to specific cases, and an embodiment of the invention provides a precipitation prediction correction method based on multimode fusion, which comprises the following steps:
1. data selection
Historical data of a plurality of forecasting modes in a historical time period is read, wherein the historical data comprises historical forecasting data and historical observing data.
In the embodiment, the selected historical data is three-source precipitation grid point analysis data of the national weather information center, and the data can well represent the actual distribution of the observed precipitation. The data time period is from 2018, 5, 1, 00, to 2019, 10, 31, 00, the time resolution is hour by hour, and the spatial resolution is 0.05 degrees by 0.05 degrees. The forecast mode is the high-resolution precipitation forecast of ECMWF (European mid-term weather forecast center) and GRAPES-mesoo (national weather forecast center) and NCEP precipitation forecast of NCEP (national environmental forecast center) issued by 00UTC (national weather service) at the same time. Meanwhile, the product SCMOCG is guided by refined grid precipitation forecast of the national weather center. The time effect of forecast data selection is 72 hours before, and for the convenience of calculation, uniform interpolation is carried out to obtain the spatial resolution (0.05 degrees multiplied by 0.05 degrees) consistent with the three-source precipitation grid point analysis data. The research range is selected as Qinling mountain and peripheral area, and the longitude and latitude are 31-40 DEG N and 103-113 DEG E.
It should be noted that different mesh sizes may be set according to actual requirements. All times were universal time. The historical time period can be selected arbitrarily, and the time and the spatial resolution can be adjusted according to the requirements. The forecast mode may include other modes such as JMA mode in addition to ECMWF mode, NCEP mode and GRAPES-Meso mode used in this embodiment. And because of the habit of those skilled in the art, different modes also become different products, such as ECMWF mode may also be referred to as ECMWF products, or ECMWF mode products, etc.
2. Pattern verification
And carrying out scoring inspection on each forecasting mode according to the historical data to obtain the score of each forecasting mode.
In this embodiment, the user can learn the predictive performance or systematic law of the pattern through an application or test. In fact, testing is the most direct and effective method of cognitive modeling. Thus, given some of the ground truth of the pattern test, the use of test scores includes the forecast Accuracy ACC (Accuracy), TS (Threat Score) scores, formulated as follows:
further, the ratio of the number of empty reports to the number of observations exceeding the threshold is defined as the forecast failure rate FR (Fail Ratio):
In the above formula, A is the number of times that precipitation exceeds a threshold value; b is the number of times of missing report; c is the number of times of empty newspaper; d is the number of times that the precipitation is correctly predicted not to exceed the threshold.
Assuming that the rainfall with the missing report of more than or equal to 2.0mm/3h can have a great influence on the public, the strong rainfall threshold value can also be called as the rainfall high threshold value which is more than or equal to 2.0mm/3h, and the rainfall standard in the sunny and rainy inspection can also be called as the rainfall low threshold value which is more than or equal to 0.1mm/3h. Precipitation tests were reported at 00 hours per day from 5.1.in 2018 to 31.10.31.in 2019, with 3 hour forecast scores over 72 hours. The text is collectively used to represent the forecast with "F" and the observation with "O".
And (3) checking the predicted hit and miss times of ECMWF and NCEP modes, wherein the rainfall is more than or equal to 2.0mm/3h, and the predicted hit and miss times are defined as hit: f is more than or equal to 2.0mm/3h, and O is more than or equal to 2.0mm/3 h; missing report: f is less than or equal to 2.0mm/3h, and O is more than or equal to 2.0mm/3 h. From the point of view of hit number, the spatial distribution of ECMWF and NCEP hit large value area is basically consistent, but ECMWF hit number is far higher than NCEP, in the whole research area, ECMWF hit number and NCEP hit number are respectively 23.8 times and 13.1 times, and hit number difference on grid points is up to 52 times. Analyzing the condition that one mode hits another mode and reporting is missed, when ECMWF hits, NCEP has a large number of reporting missing, especially at the junction of southwest corner of research area- -Shanxi and Sichuan, the number of reporting missing of grid points is up to 87; in contrast, although the hit count of the NCEP mode shows relatively poor performance, when NCEP hits, ECMWF also has missed reports, the maximum missed report count exceeds 35 times, and the large value area of the missed report count is relatively scattered, and the distribution regularity is poor.
When ECMWF is more than or equal to 3.0mm and O is more than or equal to 2.0mm, the space average failure rate is found to be 0.841. When the threshold value is changed, when ECMWF is more than or equal to 5.0mm and O is more than or equal to 2.0mm, the failure rate is greatly reduced, the average value is 0.371, and compared with ECMWF which is more than or equal to 3.0mm, the failure rate is reduced by 55.9%. Analysis shows that the larger the precipitation amount exceeds the inspection threshold, the more obvious the precipitation void report is reduced, and the lower the failure rate is, which is not only in ECMWF mode, but also in NCEP mode, grapes-Meso mode and SCMOCG mode related to the embodiment.
Under the limited condition of inspection, the ECMWF mode is taken as a comparison object, the ECMWF mode has obvious blank report on weak precipitation, and the NCEP mode and the SCMOCG mode have obvious inhibition effect on blank report. When ECMWF is more than or equal to 0.1 and less than 2.0, the number of times of eliminating ECMWF null reports by NCEP is 156.2 times on average, and the maximum number of times on grid points is 621 times; the maximum number of times of the SCMOCG elimination ECMWF null report is 431 times on grid points, and the average number of times is 133.7. Compared with NCEP mode, the overall form of the large value area of the number of times of eliminating the empty newspaper is consistent, but in a specific certain precipitation process, the situation of reducing the empty newspaper has great difference. At NCEP or SCMOCG precipitation F <0.1mm, the number of null reports is reduced not to the maximum at the grid points in FIGS. 2a and 2b, but to a higher level. In fact, the number of times that both NCEP and SCMOCG simultaneously reduce ECMWF is only 93.7 times on the spatial average, which is much less than 196.2 times in the case where NCEP or SCMOCG precipitation is less than 0.1mm, because there are fewer overlapping grid points where both modes simultaneously forecast no precipitation, and forecast conclusions are inconsistent at other grid points.
NCEP or SCMOCG can reduce the number of times of original correct prediction of ECMWF mode while eliminating empty report of ECMWF mode, so that the ECMWF has no report. From the number of times of missing report, when NCEP or SCMOCG precipitation amount F is less than 0.1, ECMWF is less than or equal to 0.1, the number of times of missing report is spatially averaged to 61 times, which is far less than the number of times of eliminating blank report, and when NCEP and SCMOCG precipitation amount is simultaneously less than 0.1mm, the number of times of missing report is less. The whole shows that when the precipitation is predicted in the ECMWF mode between 0.1 and 2.0, if the precipitation is not predicted in other modes, no precipitation is determined, and the accuracy of weather prediction can be improved.
It should be noted that, besides the TS score test disclosed in this embodiment, other score test methods such as ETS score test and FSS score test, or a comprehensive score test method may be also used in the method of the present invention.
3. Fusion correction
On the basis of taking the prediction mode with the highest score as a background field, correcting the background field by adopting a mode of threshold fusion and threshold nulling to obtain a new prediction mode.
In this embodiment, based on the above test results, three basic facts can be found: the method has the advantages that the forecasting performance of different modes is different, and even if the mode with a better checking score is adopted, empty report or missing report can occur in some cases compared with other modes; and the same forecasting mode is adopted, a checking threshold value is given, and when the forecasting precipitation exceeds the threshold value, the probability of empty report is smaller, and the failure rate is lower. According to the method, the weak rainfall in the mode with the better early-stage inspection score is eliminated by utilizing other modes, so that the empty report can be greatly reduced under the condition of sacrificing the smaller missing report, and the weather forecast accuracy is improved.
Based on the basic facts of the inspection, a conceptual diagram of a grid rainfall forecast release method with multiple forecast modes fused is provided, as shown in fig. 2. Taking 3h precipitation forecast as an example, grid precipitation takes F i,j I represents the abscissa and j represents the ordinate. Assuming that the precipitation forecast of ECMWF mode is the optimal forecast passing the early-stage inspection, selecting the ECMWF mode as a background field (figure 2 a), and when NCEP (figure 2 b) and SCMOCG (figure 2 c) diverge from the background field, for strong precipitation, performing high-threshold fusion when the precipitation value at a certain grid point of the background field is less than 5.0mm/3h, namely a first precipitation threshold, and the precipitation amount of NCEP or SCMOCG is more than or equal to 5.0mm/3 h. The specific method comprises the following steps: setting a precipitation amount threshold for NCEP and SCMOCG according to the early-stage test result (assuming that NCEP and SCMOCG high thresholds are precipitation amount not less than 11.0mm/3h and not less than 9.0mm/3h respectively, namely the respective precipitation amount threshold, the precipitation amount threshold can be adjusted in different embodiments, and the precipitation amount threshold can be the same as or different from the first precipitation amount threshold), respectivelyUnder the condition of high precipitation threshold, the failure rate of NCEP or SCMOCG forecast is lower than 0.2 or lower, and the NCEP or SCMOCG is fused into a background field at 5.0mm/3 h. FIG. 2d shows NCEP precipitation of 11.0mm/3h or more and meets grid precipitation with failure rate less than 0.2, no grid points meeting the conditions under the condition of high precipitation threshold, FIG. 2e shows SCMOCG precipitation of 9.0mm/3h or more and meets grid point precipitation with failure rate less than 0.2, F under the condition of high precipitation threshold 3,2 Reach the standard F 3,3 If not, only F is fused 3,2 Lattice points. After the precipitation high-threshold fusion is completed, aiming at weak precipitation, low-threshold emptying is carried out. The specific method comprises the following steps: assuming that the criteria for blanking is set to 0.5mm/3h, i.e. the second precipitation threshold, when the background field precipitation is less than 0.5mm/3h, the other modes do not forecast precipitation, and the weather forecast accuracy (i.e. the forecast accuracy of the precipitation low threshold) of the mode at a given grid point is higher than 80%, the blanking is performed without forecasting precipitation. F in FIG. 2a 2,2 The lattice precipitation is 0.5mm/3h, NCEP is in F 2,2 Grid points do not forecast precipitation, the weather forecast accuracy of the model on the grid is higher than 80%, the air-extinction condition is met, and grid points F in a background field are removed according to an algorithm 2,2 The final threshold fusion and threshold extinction result is shown in fig. 2f, which is also the prediction result of the new prediction mode.
It should be noted that this embodiment takes 3h precipitation prediction as an example, and the present invention is applicable to precipitation prediction of any duration, such as 1h, 6h, 12h, etc. And the criteria for strong and weak precipitation may also be set, as in this embodiment, with precipitation of 5.0mm/3h as the strong precipitation fusion limit, i.e. the first precipitation threshold, and precipitation of 0.5mm/3h as the weak precipitation extinction limit, i.e. the second precipitation threshold. Similarly, the precipitation amounts of a plurality of different grades can be set, different precipitation amount thresholds can be set respectively, and fusion and emptying can be performed respectively, so that the precipitation amount setting method is not limited to the two grades of strong precipitation and weak precipitation in the embodiment. The preset fusion threshold in this embodiment is lower than 0.2 or lower, and the preset extinction threshold is higher than 80%. Different embodiments can set corresponding fusion threshold values and extinction threshold values according to actual requirements. The precipitation amount threshold value can be set according to the early stage grading inspection result, and the precipitation amount threshold value of each mode can be the same or different and can be the same or different from the first precipitation amount threshold value. As in the embodiment, the precipitation thresholds of NCEP and SCMOCG are respectively equal to or more than 11.0mm/3h and equal to or more than 9.0mm/3h. Similarly, the precipitation amount low threshold value may be the same or different for each mode, and may be the same or different from the second precipitation amount threshold value. And different precipitation high threshold values and precipitation low threshold values can be set between different grids in the same background field. The forecast failure rate and the forecast accuracy rate are obtained through calculation according to the set precipitation threshold value, the provided forecast failure rate calculation formula, the forecast accuracy rate calculation formula and the like.
It is worth to say that, on the basis of the highest scoring prediction mode as the background field, a new prediction mode obtained after correcting the background field by adopting a mode of threshold fusion and threshold extinction is adopted to predict precipitation, so that a prediction result with high accuracy can be obtained. Similarly, in order to make the new prediction mode reach the scoring standard, many times of correction and correction analysis can be performed, and then precipitation prediction is performed, that is, after one time of correction and correction, whether the new prediction mode reaches the scoring standard can be known, if the new prediction mode does not reach the scoring standard, the threshold is changed to continue correction until the obtained new prediction mode reaches the scoring standard.
4. Correction analysis
And carrying out grading inspection on the forecast mode after correction for a preset time length to obtain grading after correction.
In this embodiment, the corrected prediction mode is formed by fusing the precipitation high threshold and emptying the precipitation low threshold, the newly obtained corrected prediction mode is checked daily, specifically for 3 hours and 24 hours, and if the score after correction does not reach the expected result, the precipitation high threshold and the precipitation low threshold are readjusted according to the checking result.
(1) 3 hour forecast effect scoring test
The new forecast mode forecast performance after the whole research period from 5 months 1 month 2018 to 10 months 31 months 2019 is calculated back, and the more the added precipitation modes are, the more favorable the forecast performance is improved, but for the convenience of expression, three precipitation mode inspection thresholds of ECMWF, NCEP and SCMOCG are still selected, namely, the influence precipitation is more than or equal to 2.0mm/3h, and the weather is more than or equal to 0.1mm/3h. The first precipitation threshold value is set to be more than or equal to 2.0mm/3h, and the second precipitation threshold value is set to be less than or equal to 0.5mm/3h. In fact, the thresholds of the different modes may have some differences, and the actual service is better beneficial to improving the prediction accuracy by dynamically checking the prediction performance of the different modes to obtain the thresholds of the different modes.
The test shows that ECMWF has better performance in the south of Shaanxi, the middle of Gansu, the north of Sichuan and the middle of inner Mongolia, the maximum TS score of a single lattice point is 0.358, and the average TS score in a region is 0.169; NCEP reaches 0.44 in the TS score of north and inner Mongolia juncture of Ningxia, and the TS score on a single lattice point is highest, but the whole performance is poor in other areas, especially the precipitation frequency is higher in the southwestern Shanxi region of not less than 2.0mm/3 h. In the case of better ECMWF scores, the NCEP's TS score is near 0, with a regional average TS score of only 0.117. The spatial distribution of the SCMOCG precipitation TS scoring large-value area is the same as that of ECMWF, the TS scoring of the south part, the Gansu south part and the Sichuan northwest part in Shaanxi is slightly higher than that of ECMWF, but the area average and single lattice point maximum values are 0.161 and 0.357 respectively, and the whole is lower than that of ECMWF.
The difference value between the TS score and the ECMWF score of the corrected forecasting mode precipitation amount of more than or equal to 2.0mm/3h is examined, and the TS score of most places of the corrected forecasting mode is improved compared with the existing forecasting mode. The TS score of the forecast mode after correction is 0.173 in the whole area which is more than or equal to 2.0mm/3h, and the forecast mode is improved by 0.01, 0.012 and 0.056 relative to ECMWF, SCMOCG and NCEP respectively, and the positive effect is obvious. By examining the number of hits and the difference between the number of empty hits and the ECMWF mode, which are equal to or greater than 2.0mm/3h, the number of hits in the whole study area of the corrected prediction mode is increased, particularly in the northwest of Sichuan, but the number of empty hits is increased in the Sichuan. Analysis shows that TS scores in the middle part of Shaanxi, the middle part of Gansu, the middle part of Ningxia and the like are obviously increased, the TS scores have a good corresponding relation with the areas with relatively small increase of the number of hits and the areas with more increase of the number of empty reports, such as Sichuan and the like, and the first precipitation threshold value and the precipitation high threshold value can be adjusted for correction, so that the empty reports are reduced.
The investigation of the weather forecast accuracy can also be called as the comparison and inspection result of the forecast accuracy of the non-forecast precipitation, and ECMWF, NCEP and SCMOCG all show the characteristics of low forecast accuracy in the southwest of the research area, high accuracy in the middle and north, and the accuracy in the southwest part of the research area being less than 68%. From the area average, SCMOCG shows best, the accuracy rate of weather forecast reaches 81.2%, NCEP times reach 80.4%, ECMWF is worst 77.4%, which shows that ECMWF has a large number of empty reports. However, SCMOCG does not fully absorb the advantages of ECMWF and NCEP, both the ECMWF and NCEP are high-value areas of weather forecast accuracy at the junctions of the middle part of shanxi and Henan, the maximum is more than 88%, and the accuracy of SCMOCG in the areas is relatively low. The weather forecast accuracy of the forecast mode after correction is obviously improved, not only in the southwest, but also in the great value area of the weather forecast accuracy in the northwest, the average area is 86.8%, the single point maximum reaches 97.9%, and the amplification is obvious. From the times of empty report and missing report, the forecasting mode after correction has missing report, especially the missing report times in the north of Sichuan and the south of Shaanxi are obvious, the maximum number of missing report times is more than 90, but the method obviously reduces the empty report for the whole area, the maximum number of missing report times is more than 240, and the accuracy of weather forecast is improved as a whole. In addition, because the second precipitation threshold value is adopted for emptying, the second precipitation threshold value is lower, and in the embodiment, the treatment is carried out when the rainfall is less than or equal to 0.5mm/3h, the large-magnitude precipitation cannot be missed due to emptying, and the overall benefit is very obvious.
(2) 24-hour predictive effect scoring test
As shown in table 1, the statistical back calculation result of the prediction mode after correction shows that: and the TS scoring and weather forecasting accuracy of the large rainfall are obviously improved in the period of 3 hours. The forecast test results of different periods of 5-9 months in 2019 are shown in the table one, and the forecast accuracy of the corrected forecast mode is obviously improved compared with the existing precipitation of different modes, and the forecast method has obvious positive effect on guiding and forecasting SCMOCG. The score of each period is improved, and the positive effect is obvious.
TABLE 1
In order to more clearly understand the forecast performance of the segment precipitation correction result in the accumulation period, the spatial performance of the 24-hour accumulated precipitation check scores and the storm forecast of the forecast mode after correction in 5-9 months in 2019 are given. It should be noted that, considering the availability of the mode data in the actual service, the 12-36h segment precipitation prediction results reported by ECMWF and NCEP mode 00UTC are tested. The SCMOCG can be used for forecasting the future 24-hour rainfall, which is reported by 00UTC, according to the actual aging. It can be seen that the accuracy of the 24-hour weather forecast of the ECMWF mode is consistent with the 3-hour overall spatial distribution, the southwest part is lower, the northeast part is higher, the maximum single point on the northwest part is 0.904, the spatial average ACC is 0.738, and the accuracy of the weather forecast is obviously lower than that of the 3-hour weather forecast. The spatial average weather forecast accuracy 0.713 of the NCEP model performed the worst. Compared with the prior art, the accuracy of the weather forecast of the SCMOCG is higher than that of the ECMWF and NCEP modes, the spatial average of the whole research area is 0.785, the single-point maximum accuracy in Henan province of the east is 0.925, and the weather forecast performance is good. The spatial average of the weather forecast accuracy of the multi-mode fusion method is 0.801, which is higher than that of SCMOCG by more than 1.6 percent, and the forecast accuracy is obviously improved in the Sichuan and northwest areas of the research area and in the Ningxia and inner Mongolian junctions of northwest areas from the aspect of spatial distribution, but the weather forecast accuracy is reduced in the Yan' an and the Ulmin of the northern areas of Shaanxi. This may be related to the same threshold being used throughout the area, so different grids may use different thresholds, where threshold refers to all adjustable thresholds in the text, to improve accuracy of the forecast.
Compared with TS scores of the prediction modes after correction, TS scores of ECMWF, NCEP, SCMOCG and the prediction modes after correction are respectively 0.134, 0.08, 0.137 and 0.147, and the multi-mode fusion method (the mode after correction is called as the multi-mode fusion method in the document) is improved by 1% compared with the whole SCMOCG, but the same problems exist in spatial distribution and weather prediction accuracy, and the local TS scores are reduced while the whole is improved, and the storm scores in northern part of Shaanxi elm are lower than the predicted TS scores of ECMWF or NCEP. The method is related to the fact that the same threshold value is adopted in the whole area, so that correction can be carried out by adopting different threshold values according to different areas, and further a better forecasting effect is obtained. The threshold herein refers to all adjustable thresholds herein.
And the existing prediction mode precipitation prediction results after the one-time strong autumn rain process of Shanxi in 9 months of 2019 is selected to analyze, so that the rainstorm area for observing precipitation is mainly in the middle and west of Shanxi in 9 months of 13 days 12 to 9 months of 14 days 12 days, the station observation shows that the precipitation amount of 695 stations exceeds 50mm in 24 hours, the precipitation amount of 12 stations exceeds 100mm, and the maximum is 147.8mm in Zheng county in Han Zhongzhong. The rainfall prediction of 24 hours before ECMWF basically does not predict a storm area, the rainfall large rainfall area reported by 12 hours before ECMWF is consistent with the whole observation, but the storm scope is obviously smaller, and the area is less than 1/5 of the observation area. The NCEP forecast precipitation is obviously smaller in range than the ECMWF and weaker in magnitude, only moderate rain is forecast in the area, and the forecast indication meaning of the heavy rain is smaller. The Grapes mode has the best forecasting effect at the moment of approach, the scope and the magnitude of a storm area are basically consistent with the observation, but in long-term service inspection, the grade of the Grapes mode is lower than ECMWF as a whole, so that if the Grapes mode is selected from the angle of a forecaster, the probability of selecting the Grapes mode product as the actual service rainfall forecast is smaller. The SCMOCG forecast of the guide product is slightly better than ECMWF in the southern part of Shaanxi, but is worse than ECMWF in the middle part of Shaanxi, and the forecast mode after correction is more consistent with the observation in both the form and the area of the storm area. From the TS score of the storm forecast, the TS score of the ECMWF is 0.13 hours before, the TS score of the NCEP rainfall forecast is 0.0, the grades are 0.534, the SCMOCG is 0.426, and the forecast score of the forecast mode after correction is 0.691 minutes, so that the effect of the forecast mode after correction is very good.
In the embodiment, in order to reduce the empty report and missing report conditions of grid forecast as much as possible and effectively improve the forecast accuracy, a multi-mode fusion grid precipitation forecast correction method is provided, three-source fusion grid precipitation live data of a national weather information center are utilized, fusion correction is carried out on the selected numerical mode with optimal scores on the basis of objective performance of multiple modes of inspection and analysis, and the following conclusion is obtained:
(1) The rainfall forecasting performances of different modes are greatly different, and the situation that rainfall products with good overall forecasting performance are missed and other rainfall products hit exists; the ECMWF, NCEP and SCMOCG are inspected to find that for precipitation forecast in sunny and rainy days, the number of times of missing report of the three products is obviously lower than that of empty report, and based on the characteristics, the grid points of the precipitation forecast in different modes can be utilized for emptying. For the same mode, the larger the forecast precipitation exceeds the inspection threshold, the smaller the likelihood of an empty report.
(2) According to the inspection result, a grid precipitation prediction correction method based on multi-mode fusion is provided, namely, a mode with the best inspection score is selected as a background field, high-threshold fusion and low-threshold blank elimination are respectively adopted for large precipitation and weak precipitation with obvious influence, and precipitation prediction products of other modes are fused into grid precipitation prediction of the background field. For strong precipitation, the main method is to set a first precipitation threshold (a high threshold), check modes except the background field, so that the failure rate meeting the requirement of a user for precipitation prediction under the condition of the high threshold of the precipitation of other modes is lower than 20% or lower, and then merge grid precipitation of the mode into the background field. For weather forecast, a second precipitation threshold (a low threshold) is set, when grid points of a background field forecast weak precipitation, namely, the grid points of other modes do not report precipitation, the historical weather forecast accuracy of the grid points of other modes, namely, the forecast accuracy of the non-forecast precipitation is checked, and when the accuracy is higher than 80%, the weak precipitation in the precipitation background field is removed, and no precipitation is reported.
(3) Historical back calculations show that the TS score of the corrected prediction mode is 0.173 in the whole area which is more than or equal to 2.0mm/3h, and is respectively improved by 0.01, 0.012 and 0.056 relative to 0.169, 0.117 and 0.161 of ECMWF, SCMOCG and NCEP. The accuracy rate of weather forecast is 86.8%, which is improved by 5.6%, 6.4% and 9.4% respectively compared with 81.2% of SCMOCG, 80.4% of NCEP and 77.4% of ECMW.
It is worth to say that both back calculation and actual business show that the technology can effectively improve the accuracy of weather forecast and the score of storm forecast TS, and when the high threshold fusion is adopted for precipitation forecast products outside a background field in business, the threshold is adjusted in real time, so that a better forecast effect is achieved. The intelligent grid precipitation prediction void report missing report phenomenon is improved, the performance error in the multi-mode integration process is reduced through fusion correction, objective defects existing in the prior art are effectively overcome, and the precipitation prediction accuracy is further improved.
Different from the traditional integration method, the multi-mode fusion grid precipitation correction method directly selects the possibly correct forecast from different precipitation products according to the inspection result, does not involve complex weight calculation, and pertinently carries out threshold fusion correction on strong and weak precipitation, compared with the prior art, the method effectively improves precipitation forecast accuracy, belongs to a new technology for intelligent grid forecast correction, has better service application effect, and has a leading effect in industry.
As shown in fig. 3, the present invention provides a precipitation prediction correction device based on multimode fusion, which includes: the system comprises a data acquisition module, a scoring calculation module and a precipitation correction module, wherein,
the data acquisition module is used for reading historical data of a plurality of forecasting modes in a historical time period, wherein the historical data comprises historical forecasting data and historical observation data;
the scoring calculation module is used for scoring and checking each forecasting mode according to the historical data acquired by the data acquisition module to acquire the score of each forecasting mode;
and the rainfall correction module is used for correcting the background field by adopting a mode of threshold fusion and threshold extinction on the basis of taking the prediction mode with the highest score as the background field, so as to obtain the corrected prediction mode.
In one embodiment of the present invention, the score calculating module is specifically configured to calculate the TS score of each prediction mode according to the historical data under a preset check limit.
In one embodiment of the present invention, the precipitation correction module includes a background selection unit, a failure calculation unit, and a fusion judgment unit, wherein,
The background selection unit is used for taking the prediction mode with the highest score as a background field;
the failure calculation unit is used for calculating the failure rate of prediction under the respective precipitation amount threshold value of each prediction mode in the prediction modes inconsistent with the highest scoring mode when the historical prediction precipitation amount of the prediction mode with the highest scoring is smaller than the first precipitation amount threshold value and the historical prediction precipitation amounts of other prediction modes except the prediction mode with the highest scoring are larger than or equal to the first precipitation amount threshold value in one grid of the background field;
and the fusion judging unit is used for judging whether all the forecasting failure rates are lower than a preset fusion threshold value, if yes, fusing the grids, and otherwise, not processing the grids.
In one embodiment of the invention, the precipitation correction module comprises a background selection unit, a success calculation unit and a blanking judgment unit,
the background selection unit is used for taking the forecast mode with the highest score as a background field;
the success calculation unit is configured to, in one grid of the background field, read prediction accuracy of prediction modes corresponding to all the non-predicted precipitation, where the historical prediction precipitation of the prediction mode with the highest score is greater than zero and smaller than a second precipitation threshold, and non-predicted precipitation exists in other prediction modes except for the prediction mode with the highest score;
The blanking judging unit is used for judging whether the forecasting accuracy rate higher than a preset blanking threshold exists in all the forecasting accuracy rates, if yes, blanking is carried out on the grids, and otherwise, the grids are not processed.
In one embodiment of the present invention, the score calculating module is further configured to perform score inspection on the corrected forecast model obtained by the precipitation correction module, to obtain a corrected score.
In one embodiment of the present invention, the score calculating module is further configured to perform a score check on the corrected prediction mode obtained by the precipitation correction module for a preset time length, so as to obtain a corrected score.
In an embodiment of the present invention, the device further includes a correction improvement module, configured to determine whether the corrected score reaches a preset standard score, if yes, perform precipitation prediction in a corrected prediction mode, and if not, correct the background field after adjusting the first precipitation threshold and/or the second precipitation threshold and/or the precipitation high threshold and/or the precipitation low threshold and/or the preset fusion threshold and/or the preset blank threshold.
The content of information interaction and execution process between each module and unit in the device is based on the same conception as the embodiment of the method of the present invention, and specific content can be referred to the description in the embodiment of the method of the present invention, which is not repeated here.
It is noted that relational terms such as first and second, and the like, are used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Moreover, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the statement "comprises one" does not exclude that an additional identical element is present in a process, method, article or apparatus that comprises the element.
Those of ordinary skill in the art will appreciate that: all or part of the steps for implementing the above method embodiments may be implemented by hardware related to program instructions, and the foregoing program may be stored in a computer readable storage medium, where the program, when executed, performs steps including the above method embodiments; and the aforementioned storage medium includes: various media in which program code may be stored, such as ROM, RAM, magnetic or optical disks.
Finally, it should be noted that: the foregoing description is only illustrative of the preferred embodiments of the present invention, and is not intended to limit the scope of the present invention. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention are included in the protection scope of the present invention.
Claims (8)
1. A precipitation prediction correction method based on multi-mode fusion, the method comprising:
reading historical data of a plurality of forecasting modes in a historical time period, wherein the historical data comprises historical forecasting data and historical observing data;
performing scoring inspection on each forecasting mode according to the historical data to obtain the score of each forecasting mode;
on the basis of taking the prediction mode with the highest score as a background field, correcting the background field by adopting a mode of threshold fusion and threshold nulling;
on the basis of taking the prediction mode with the highest score as a background field, correcting the background field by adopting a mode of threshold fusion and threshold blanking, wherein the method comprises the following steps:
taking the prediction mode with the highest score as a background field;
for strong precipitation, when the predicted precipitation of the prediction mode with the highest score is smaller than a first precipitation threshold value and the predicted precipitation of other prediction modes except the prediction mode with the highest score is larger than or equal to the first precipitation threshold value, calculating the prediction failure rate of each prediction mode in the prediction modes inconsistent with the prediction mode with the highest score under the respective precipitation quantity threshold value;
Judging whether all the forecasting failure rates are lower than a preset fusion threshold value, if yes, fusing the grids, otherwise, not processing the grids; the method comprises the steps of,
taking the prediction mode with the highest score as a background field;
aiming at weak precipitation, in one grid of the background field, the predicted precipitation of the prediction mode with the highest score is larger than zero and smaller than a second precipitation threshold, and the non-predicted precipitation exists in other prediction modes except for the prediction mode with the highest score, and the prediction accuracy of all the prediction modes corresponding to the non-predicted precipitation under the respective precipitation low threshold is calculated;
judging whether all the prediction accuracy rates are higher than a preset extinction threshold, if yes, carrying out extinction on the grids, otherwise, not processing the grids.
2. The method for correcting a rainfall forecast based on multimode fusion according to claim 1, wherein the step of performing a scoring test on each forecast mode according to the historical data to obtain a score of each forecast mode comprises the following steps:
under the preset test condition, calculating TS scores of all forecast modes according to the historical data.
3. The method for correcting a rainfall forecast based on multi-mode fusion according to claim 1, wherein on the basis of taking a forecast mode with the highest score as a background field, correcting the background field by adopting a mode of threshold fusion and threshold blanking, and further comprising:
on the basis of taking the prediction mode with the highest score as a background field, correcting the background field in a mode of threshold fusion and threshold nulling to obtain a corrected prediction mode;
and carrying out grading test on the forecast mode after correction to obtain grading after correction.
4. A method of correcting a forecast of precipitation based on multimode fusion according to claim 3, wherein scoring the corrected forecast pattern to obtain a corrected score comprises:
and carrying out grading inspection on the forecast mode after correction for a preset time length to obtain grading after correction.
5. The method for correcting a precipitation forecast based on multimode fusion according to any of claims 3 or 4, characterized in that said method further comprises:
judging whether the corrected score reaches a preset standard score, if so, adopting a corrected prediction mode to predict the precipitation, otherwise, adjusting the first precipitation threshold value and/or the second precipitation threshold value and/or the precipitation high threshold value and/or the precipitation low threshold value and/or the preset fusion threshold value and/or the preset emptying threshold value, and correcting the background field.
6. A precipitation prediction correction device based on multimode fusion, the device comprising: the system comprises a data acquisition module, a scoring calculation module and a precipitation correction module, wherein,
the data acquisition module is used for reading historical data of a plurality of forecasting modes in a historical time period, wherein the historical data comprises historical forecasting data and historical observation data;
the scoring calculation module is used for scoring and checking each forecasting mode according to the historical data acquired by the data acquisition module to acquire the score of each forecasting mode;
the rainfall correction module is used for correcting the background field by adopting a mode of threshold fusion and threshold blanking on the basis of taking the prediction mode with the highest score as the background field;
on the basis of taking the prediction mode with the highest score as a background field, correcting the background field by adopting a mode of threshold fusion and threshold blanking, wherein the method comprises the following steps:
taking the prediction mode with the highest score as a background field;
for strong precipitation, when the predicted precipitation of the prediction mode with the highest score is smaller than a first precipitation threshold value and the predicted precipitation of other prediction modes except the prediction mode with the highest score is larger than or equal to the first precipitation threshold value, calculating the prediction failure rate of each prediction mode in the prediction modes inconsistent with the prediction mode with the highest score under the respective precipitation quantity threshold value;
Judging whether all the forecasting failure rates are lower than a preset fusion threshold value, if yes, fusing the grids, otherwise, not processing the grids; the method comprises the steps of,
taking the prediction mode with the highest score as a background field;
aiming at weak precipitation, in one grid of the background field, the predicted precipitation of the prediction mode with the highest score is larger than zero and smaller than a second precipitation threshold, and the non-predicted precipitation exists in other prediction modes except for the prediction mode with the highest score, and the prediction accuracy of all the prediction modes corresponding to the non-predicted precipitation under the respective precipitation low threshold is calculated;
judging whether all the prediction accuracy rates are higher than a preset extinction threshold, if yes, carrying out extinction on the grids, otherwise, not processing the grids.
7. A readable storage medium having executable instructions thereon, which when executed, cause a computer to perform the method comprised in any of claims 1-5.
8. A computing device, comprising: one or more processors, memory, and a program, wherein the one or more programs are stored in the memory and configured to perform the method of any of claims 1-5 by the one or more processors.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011362926.7A CN112488385B (en) | 2020-11-27 | 2020-11-27 | Precipitation prediction correction method and device based on multi-mode fusion |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011362926.7A CN112488385B (en) | 2020-11-27 | 2020-11-27 | Precipitation prediction correction method and device based on multi-mode fusion |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112488385A CN112488385A (en) | 2021-03-12 |
| CN112488385B true CN112488385B (en) | 2024-04-16 |
Family
ID=74936582
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011362926.7A Active CN112488385B (en) | 2020-11-27 | 2020-11-27 | Precipitation prediction correction method and device based on multi-mode fusion |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112488385B (en) |
Families Citing this family (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113592132B (en) * | 2021-04-29 | 2022-04-05 | 山东省气象科学研究所(山东省海洋气象科学研究所、山东省气象局培训中心) | Rainfall objective forecasting method based on numerical weather forecast and artificial intelligence |
| CN114325879B (en) * | 2021-12-20 | 2022-09-30 | 广东省气象台(南海海洋气象预报中心) | Quantitative precipitation correction method based on classification probability |
| CN116108971B (en) * | 2022-12-20 | 2023-07-25 | 中国民用航空局空中交通管理局航空气象中心 | Convection weather forecasting method and system based on numerical mode convection coverage rate |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111159640A (en) * | 2019-11-20 | 2020-05-15 | 北京玖天气象科技有限公司 | Small rain emptying method, system, electronic equipment and storage medium suitable for grid forecast |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105808948B (en) * | 2016-03-08 | 2017-02-15 | 中国水利水电科学研究院 | Automatic correctional multi-mode value rainfall ensemble forecast method |
| US20180292573A1 (en) * | 2017-04-05 | 2018-10-11 | Weather Analytics Llc | Corrective Ensemble Forecasting System for Tropical Cyclones |
-
2020
- 2020-11-27 CN CN202011362926.7A patent/CN112488385B/en active Active
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111159640A (en) * | 2019-11-20 | 2020-05-15 | 北京玖天气象科技有限公司 | Small rain emptying method, system, electronic equipment and storage medium suitable for grid forecast |
Non-Patent Citations (2)
| Title |
|---|
| 基于小波分析的西北区智能网格气温客观预报方法的检验评估;刘新伟等;大气科学学报;第43卷(第4期);673-686 * |
| 陕西省精细化网格预报业务系统技术方法;王建鹏等;气象科技;第46卷(第5期);910-918 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112488385A (en) | 2021-03-12 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112488385B (en) | Precipitation prediction correction method and device based on multi-mode fusion | |
| Howarth et al. | Changes in extreme precipitation in the northeast United States: 1979–2014 | |
| Wang et al. | Trends and low frequency variability of extra-tropical cyclone activity in the ensemble of twentieth century reanalysis | |
| Wolff et al. | Beyond the basics: Evaluating model-based precipitation forecasts using traditional, spatial, and object-based methods | |
| Wang | On the calculation and correction of equitable threat score for model quantitative precipitation forecasts for small verification areas: The example of Taiwan | |
| CN111242404B (en) | Extreme evaluation method and system for heavy rainfall induced flood incident | |
| CN110334404B (en) | Accurate identification method for watershed scale sudden dry and droughts | |
| Palaniswami et al. | Change point detection and trend analysis of rainfall and temperature series over the Vellar River basin. | |
| CN116910041B (en) | Daily correction method for remote sensing precipitation product based on scale analysis | |
| Zhang et al. | Spatio-temporal variations in extreme drought in China during 1961–2015 | |
| Chen et al. | Spatial spread‐skill relationship in terms of agreement scales for precipitation forecasts in a convection‐allowing ensemble | |
| Stein et al. | Neighborhood-based contingency tables including errors compensation | |
| CN116362419A (en) | Urban flood control early warning system and method | |
| Tozer et al. | Large‐scale ocean‐atmospheric processes and seasonal rainfall variability in South Australia: potential for improving seasonal hydroclimatic forecasts | |
| Sciuto et al. | Quality control of daily rainfall data with neural networks | |
| Zhu et al. | Predictability and skill of convection-permitting ensemble forecast systems in predicting the record-breaking “21· 7” extreme rainfall event in Henan Province, China | |
| Smith et al. | The hourly wind-bias adjusted precipitation data set from the Environment and Climate Change Canada automated surface observation network (2001–2019) | |
| Alerskans et al. | Optimizing spatial quality control for a dense network of meteorological stations | |
| Shin et al. | Predictability of PM2. 5 in Seoul based on atmospheric blocking forecasts using the NCEP global forecast system | |
| Jie et al. | Using a deterministic time-lagged ensemble forecast with a probabilistic threshold for improving 6–15 day summer precipitation prediction in China | |
| Dittus et al. | An investigation of some unexpected frost day increases in southern Australia | |
| Mo et al. | Impact of annual maximum wind speed in mixed wind climates on wind hazard for mainland China | |
| Wasko et al. | A systematic review of climate change science relevant to Australian design flood estimation | |
| CN112526641A (en) | Method, system and equipment for identifying quality of rainfall observed value in real time | |
| Gupta et al. | Extreme value analysis by graphical and Lieblein techniques for rainfall on western coast of India |
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 |