CN112051627A - Method, device and medium for correcting numerical mode multi-level rainfall forecast value - Google Patents
Method, device and medium for correcting numerical mode multi-level rainfall forecast value Download PDFInfo
- Publication number
- CN112051627A CN112051627A CN202010847121.5A CN202010847121A CN112051627A CN 112051627 A CN112051627 A CN 112051627A CN 202010847121 A CN202010847121 A CN 202010847121A CN 112051627 A CN112051627 A CN 112051627A
- Authority
- CN
- China
- Prior art keywords
- precipitation
- value
- forecast
- magnitude
- correction
- 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
Links
Images
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01W—METEOROLOGY
- G01W1/00—Meteorology
- G01W1/10—Devices for predicting weather conditions
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
- Y02A10/00—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE at coastal zones; at river basins
- Y02A10/40—Controlling or monitoring, e.g. of flood or hurricane; Forecasting, e.g. risk assessment or mapping
Landscapes
- Environmental & Geological Engineering (AREA)
- Engineering & Computer Science (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 discloses a correcting method of a numerical mode multi-level rainfall forecast value, which calculates a corresponding rainfall correction value and carries out TS scoring through the rainfall forecast value and an actual rainfall value in a training period corresponding to a forecast day and a preset rainfall magnitude standard so as to determine and obtain an optimal rainfall correction threshold value sequence for correcting the rainfall forecast value of the forecast day. The invention also provides a corresponding rainfall forecast value correcting device. By adopting the embodiment of the invention, the precipitation correction forecast score under each precipitation magnitude in the training period corresponding to the forecast day is calculated to determine the optimal precipitation correction threshold sequence for correcting the precipitation prediction value on the forecast day, so that the day-by-day correction of the precipitation prediction value is realized, the precipitation prediction error of each precipitation magnitude is effectively reduced, and the accuracy of precipitation prediction is improved.
Description
Technical Field
The invention relates to the technical field of meteorological prediction, in particular to a method, a device and a medium for correcting a numerical mode multi-level rainfall forecast value.
Background
With the continuous development of numerical weather forecast, numerical forecast products become more and more important in the business forecast of forecasters. Numerical pattern prediction and correction are one of the mainstream methods in weather prediction. Numerical model forecasting uses knowledge of atmospheric dynamics and meteorology primarily to generate predictions of weather indicators. However, the numerical prediction product also inevitably has errors due to the uncertainty of the initial field and the numerical pattern itself. Different modes are different in systematic errors of precipitation forecast, for example, the global mode of organizations such as the European middle weather forecast center (ECMWF) and the Japanese weather parlor (JMA) has the condition that strong precipitation forecast is missed, and the regional mode such as the mesoscale mode (AREM) of the Wuhan rainstorm research of the China weather bureau is stronger than the actual condition, so that the inconsistency of the modes in the errors of the precipitation forecast with different precipitation intensity is reflected.
Therefore, the deviation correction is usually performed on the numerical prediction product by using some statistical post-processing techniques to obtain a product with higher prediction skills. Or depending on the subjective experience of the forecaster, and depending on the expert experience according to the forecasting result, manually generating a local forecasting correction result as a final weather forecasting result. However, in the process of implementing the invention, the inventor finds that the prior art has at least the following problems: the method for correcting the rainfall forecast value in the prior art cannot completely eliminate the forecast error and cannot obtain a relatively accurate rainfall correction result.
Disclosure of Invention
The invention aims to provide a method, a device and a medium for correcting a multi-level rainfall forecast value in a numerical mode, which are used for determining an optimal rainfall correction threshold sequence by calculating rainfall correction forecast scores under all rainfall levels in a training period, so that the rainfall prediction values are corrected day by day, the rainfall prediction errors of all the rainfall levels are effectively reduced, and the rainfall prediction accuracy is improved.
In order to achieve the above object, an embodiment of the present invention provides a method for correcting a numerical mode multi-level precipitation forecast value, including:
determining a training period corresponding to a forecast day, and acquiring a rainfall forecast value of each forecast station to each training day in the training period and an actual rainfall value of each training day;
calculating precipitation correcting values after different precipitation correcting threshold values are processed under each precipitation magnitude according to all precipitation predicted values in the training period; n precipitation magnitude values and a precipitation threshold value corresponding to each precipitation magnitude value are preset, wherein N is greater than 1;
aiming at the same precipitation magnitude, performing forecast TS scoring according to precipitation correction values after different precipitation correction threshold values are processed, actual precipitation values and precipitation threshold values corresponding to the precipitation magnitude; taking the precipitation correction threshold corresponding to the TS scoring result with the highest score as the optimal precipitation correction threshold of the precipitation magnitude, so as to obtain the optimal precipitation correction threshold corresponding to each precipitation magnitude as the optimal precipitation correction threshold sequence;
and correcting the rainfall forecast value of the forecast day according to the optimal rainfall correction threshold sequence to obtain a rainfall correction value of the forecast day as a final rainfall forecast result of the forecast day.
As an improvement of the above scheme, the calculating, according to all the rainfall forecast values in the training period, the rainfall settlement value after being processed by different rainfall settlement threshold values at each rainfall magnitude specifically includes:
presetting precipitation correction threshold F corresponding to each precipitation magnitude kkThe n candidate values are used as a precipitation correction threshold value candidate value sequence { F) corresponding to the precipitation magnitude kk1，Fk2，…，Fkn}; wherein k represents the number of each precipitation magnitude, and k is more than or equal to 1 and less than or equal to N;
and aiming at the same precipitation magnitude k, calculating to obtain precipitation correction values corresponding to all the precipitation correction values after different precipitation correction threshold values are processed according to each precipitation correction value and each candidate value in the precipitation correction threshold value candidate value sequence corresponding to the precipitation magnitude.
As an improvement of the above scheme, for the same precipitation magnitude k, calculating, according to each precipitation forecast value and each candidate value in the precipitation correction threshold candidate value sequence corresponding to the precipitation magnitude, precipitation correction values corresponding to all the precipitation forecast values after different precipitation correction threshold treatments, specifically including:
when the precipitation magnitude k is equal to 1, calculating precipitation correction values corresponding to all the precipitation prediction values according to each precipitation prediction value and each candidate value in the precipitation correction threshold candidate value sequence corresponding to the precipitation magnitude according to the following first calculation formula:
when the precipitation magnitude is 1< k < N, calculating precipitation correction values corresponding to all the precipitation prediction values according to each precipitation prediction value and each candidate value in the precipitation correction threshold candidate value sequence corresponding to the precipitation magnitude according to the following second calculation formula:
when the precipitation magnitude k is equal to N, calculating precipitation correction values corresponding to all the precipitation prediction values according to each precipitation prediction value and each candidate value in the precipitation correction threshold candidate value sequence corresponding to the precipitation magnitude according to the following third calculation formula:
wherein y is a precipitation setting value, x is a precipitation forecast value, and FkA precipitation correction threshold value corresponding to the precipitation magnitude k is determined, and the corresponding precipitation correction threshold value candidate value sequence is { F }k1，Fk2，…，Fkn}；OkFor each precipitation level k corresponding to a dropAnd (4) water threshold value.
As an improvement of the above scheme, for the same precipitation magnitude, performing forecast TS scoring according to a precipitation correction value, an actual precipitation value, and a precipitation threshold corresponding to each precipitation magnitude after treatment by using different precipitation correction thresholds specifically includes:
aiming at the same precipitation magnitude k, according to precipitation correction values after different precipitation correction threshold values are processed, actual precipitation values and precipitation threshold values corresponding to each precipitation magnitude, forecasting TS scoring is carried out through the following TS scoring formula:
wherein TS represents the forecast TS scoring result; NA represents that any precipitation corrected value of any training day in the training period and the actual precipitation value of the training day both reach the precipitation threshold value O corresponding to the precipitation magnitude kkThe number of times of (c); NB represents that any precipitation order value of any training day in the training period reaches the precipitation threshold value O corresponding to the precipitation order value kkAnd the actual precipitation value of the training day does not reach the precipitation threshold value O corresponding to the precipitation magnitude kkThe number of times of (c); NC represents that any precipitation correction value on any training day in the training period does not reach the precipitation threshold value O corresponding to the precipitation magnitude kkAnd the actual precipitation value of the training day reaches the precipitation threshold value O corresponding to the precipitation magnitude kkThe number of times.
As an improvement of the above scheme, the correcting the rainfall forecast value on the forecast day according to the optimal rainfall correction threshold sequence to obtain a rainfall correction value on the forecast day as a final rainfall forecast result on the forecast day specifically includes:
correcting a threshold sequence { F) according to the optimal precipitation1，F2，…，FNAnd the predicted rainfall value x 'of the forecast day is calculated by the following rainfall correction formula to obtain the predicted rainfall value y' of the forecast day:
wherein, OkA precipitation threshold value for each precipitation magnitude k.
As an improvement of the above scheme, the determining the training period corresponding to the forecast day specifically includes:
acquiring 20 days before the forecast day and 20 days after the forecast day on the same day of the previous year as training periods of the forecast day.
As an improvement of the above scheme, the presetting of k precipitation magnitudes and the precipitation threshold corresponding to each precipitation magnitude specifically include:
presetting 10 precipitation magnitude levels and precipitation threshold O corresponding to each precipitation magnitude levelk(ii) a Wherein, the precipitation threshold value that each precipitation magnitude corresponds is respectively: 0.1mm, 1mm, 5mm, 10mm, 25mm, 35mm, 50mm, 75mm, 100mm, 150 mm.
The embodiment of the invention also provides a correcting device for the numerical mode multi-level precipitation forecast value, which comprises the following steps:
the rainfall data acquisition module is used for determining a training period corresponding to the forecast day, and acquiring a rainfall forecast value of each forecast station to each training day in the training period and an actual rainfall value of each training day;
the rainfall correction value calculating module is used for calculating rainfall correction values processed by different rainfall correction threshold values under each rainfall magnitude according to all rainfall forecast values in the training period; n precipitation magnitude values and a precipitation threshold value corresponding to each precipitation magnitude value are preset, wherein N is greater than 1;
the optimal precipitation correction threshold calculation module is used for carrying out forecast TS scoring according to precipitation correction values and actual precipitation values which are processed by different precipitation correction thresholds and precipitation thresholds corresponding to each precipitation magnitude aiming at the same precipitation magnitude; taking the precipitation correction threshold corresponding to the highest TS score as the optimal precipitation correction threshold of the precipitation magnitude, so as to obtain the optimal precipitation correction threshold corresponding to each precipitation magnitude as the optimal precipitation correction threshold sequence;
and the precipitation forecast result obtaining module is used for correcting the precipitation forecast value of the forecast day according to the optimal precipitation correction threshold sequence to obtain a precipitation correction value of the forecast day as a final precipitation forecast result of the forecast day.
An embodiment of the present invention further provides an apparatus for correcting a numerical mode multi-level precipitation forecast value, which includes a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, and when the processor executes the computer program, the processor implements a method for correcting a numerical mode multi-level precipitation forecast value according to any one of claims 1 to 7.
An embodiment of the present invention further provides a computer-readable storage medium, where the computer-readable storage medium includes a stored computer program, where the computer program, when running, controls an apparatus where the computer-readable storage medium is located to execute the method for correcting the numerical mode multiple precipitation forecast value according to any one of claims 1 to 7.
Compared with the prior art, the method, the device and the medium for correcting the numerical-mode multi-level rainfall forecast value disclosed by the invention calculate the corresponding rainfall correction value and carry out TS scoring through the rainfall forecast value, the actual rainfall value and the preset rainfall magnitude standard in the training period corresponding to the forecast day so as to determine and obtain the optimal rainfall correction threshold value sequence for correcting the rainfall forecast value of the forecast day. The embodiment of the invention adopts the quasi-symmetrical sliding training period of 1 year to carry out the statistical analysis of the optimal rainfall correction threshold value, can better self-adaptively predict the seasonal background similar to the day before and after, and ensures the timeliness of the training data. In addition, the historical rainfall forecast value and the actual rainfall value are used as training data, can be directly obtained through a numerical weather forecast mode, are suitable for obtaining grid point data or station data, and are simple and easy to obtain. The precipitation correction forecast score under each precipitation magnitude in the training period corresponding to the forecast day is calculated to determine the optimal precipitation correction threshold sequence for correcting the precipitation forecast value on the forecast day, so that the precipitation forecast value is corrected day by day, the precipitation forecast errors on the forecast days on different precipitation magnitudes are effectively reduced, and the accuracy of the precipitation forecast is improved.
Drawings
FIG. 1 is a flowchart illustrating steps of a method for correcting a predicted precipitation value in a numerical mode according to an embodiment of the present invention;
fig. 2 is a schematic structural diagram of a correcting device for a numerical mode multi-level precipitation forecast value according to a second embodiment of the present invention;
fig. 3 is a schematic structural diagram of another device for correcting the value of the numerical mode multiple precipitation forecast according to the third embodiment of the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
Fig. 1 is a schematic flow chart illustrating steps of a method for correcting a numerical mode multi-level precipitation forecast value according to an embodiment of the present invention. The method for correcting the numerical-mode multi-level precipitation forecast value provided by the embodiment of the invention is implemented through steps S1 to S4:
s1, determining the training period corresponding to the forecast day, and obtaining the forecast value of each forecast station to the precipitation of each training day in the training period, and the actual precipitation value of each training day.
In the embodiment of the invention, in order to correct the rainfall forecast value of the forecast day, the historical date is required to be acquired as the training period of the forecast day, and the learning training is carried out according to the rainfall forecast value and the actual rainfall value of each training day in the training period, so that the correction of the rainfall forecast value of the forecast day is realized.
As a preferred embodiment, in order to better adapt to similar seasonal backgrounds before and after the forecast day and ensure timeliness of participating in training data, a one-year quasi-symmetric sliding training period mode is adopted for correcting the forecast value of rainfall on the forecast day. Specifically, 20 days before the forecast day and 20 days after the same day of the previous year on the forecast day are acquired as the training period of the forecast day.
It should be noted that, there are several stations each day to predict the rainfall amount of the same day, so there is only one actual rainfall value z on the same training day, and there are multiple rainfall prediction values x on the same training day. For example, 40 training days exist in a training period corresponding to the forecast day, and 100 stations perform rainfall prediction each day, so that the forecast value of the rainfall in the training period is 4000 (the numerical value can be repeated), and the actual rainfall value is 40.
It is to be understood that the above-described scenarios and values are exemplary only and are not to be construed as specifically limiting the present invention. In practical application, the corresponding forecast value of the rainfall and the actual rainfall value can be obtained according to the actual training period duration and the number of forecast stations, and the beneficial effects obtained by the method are not influenced.
S2, calculating precipitation correcting values after different precipitation correcting threshold values are processed under each precipitation magnitude according to all precipitation forecast values in the training period; n precipitation magnitude values and precipitation threshold values corresponding to each precipitation magnitude value are preset, and N is larger than 1.
In the embodiment of the invention, N precipitation magnitude levels k and a precipitation threshold O corresponding to each precipitation magnitude level k are presetkWherein N is>1. The precipitation levels k represent the graduations of the precipitation amounts in one day (24 hours), each of which is provided with a corresponding precipitation threshold Ok. By way of example, the precipitation level may be divided according to existing division criteria: light rain: the rainfall is less than 10 mm; rain: the rainfall is 10-25 mm; heavy rain: the rainfall is 25-50 mm; rainstorm: the rainfall is 50-100 mm; heavy rainstorm: the rainfall is 100-250 mm.
In a preferred embodiment, the forecast value is entered for the precipitation on the forecast dayThe method has the advantages that more accurate correction is carried out, and the precipitation magnitude and the corresponding precipitation threshold are more finely divided. Specifically, 10 precipitation levels k are preset, that is, N is 10, and a precipitation threshold O corresponding to each precipitation level k is setkRespectively setting as follows: 0.1mm, 1mm, 5mm, 10mm, 25mm, 35mm, 50mm, 75mm, 100mm, 150 mm.
It can be understood that, in the embodiment of the present invention, the number N of precipitation levels to be corrected may be flexibly set according to the occupation condition of the computing resources, and meanwhile, the precipitation thresholds of the respective levels may also be adaptively corrected, so as to improve the performance of the mode precipitation forecast without affecting the beneficial effects obtained by the present invention.
Further, the calculating of the precipitation correction value after being processed by different precipitation correction threshold values at each precipitation magnitude according to all the precipitation forecast values in the training period specifically includes steps S21 and S22:
s21, presetting precipitation correction threshold F corresponding to each precipitation magnitude kkThe n candidate values are used as a precipitation correction threshold value candidate value sequence { F) corresponding to the precipitation magnitude kk1，Fk2，…，Fkn}; wherein k represents the number of each precipitation magnitude, and k is more than or equal to 1 and less than or equal to N.
In the embodiment of the invention, a corresponding precipitation correction threshold F is preset for each precipitation magnitude kkThe n candidate values are used for determining the optimal precipitation correction threshold value from the candidate values of each precipitation correction threshold value in the subsequent process, and the optimal precipitation correction threshold value is used as the optimal precipitation correction threshold value of the corresponding precipitation magnitude.
Preferably, 10 candidate values are set for each precipitation correction threshold k, and then a precipitation correction threshold candidate value sequence { F } corresponding to each precipitation magnitude k is obtainedk1，Fk2，…，Fk10}. As an example, for a precipitation level k equal to 1, the corresponding precipitation correction threshold F is set1Is { F }1，1，F1，2，…，F1，10And for the precipitation magnitude k equal to 2, correcting the corresponding precipitation threshold value F2Is { F }2，1，F2，2，…，F2，10And so on.
It should be noted that the higher the level of the precipitation level k is, the corresponding precipitation correction threshold F iskThe larger the value of (a). Therefore, in setting the sequence of precipitation threshold candidate values, F should be guaranteedkEach candidate value in the corresponding precipitation threshold candidate value sequence is larger than Fk-1Each candidate in the corresponding sequence of precipitation threshold candidates.
In one embodiment, the precipitation threshold candidate sequence is set with a sequence length of 2mm precipitation and a numerical interval between each candidate of 0.1 mm. By way of example, let F1Setting the corresponding precipitation threshold candidate value sequence as {0.1mm, 0.2mm, 0.3mm, …, 2mm }, and setting F2The corresponding precipitation threshold candidate sequence is set to {2.1mm, 2.2mm, 2.3mm, …, 4mm }, and so on.
It should be understood that the above numerical values are only examples and do not constitute specific limitations of the present invention.
And S22, aiming at the same precipitation magnitude k, calculating to obtain precipitation correction values corresponding to all the precipitation correction values after different precipitation correction threshold values are processed according to each precipitation prediction value and each candidate value in the precipitation correction threshold value candidate value sequence corresponding to the precipitation magnitude.
In the embodiment of the invention, each precipitation forecast value x in the training period is corrected through a preset precipitation threshold candidate value sequence to obtain a precipitation correction value y.
Preferably, step S22 specifically includes steps S221 to S223:
s221, when the precipitation magnitude k is equal to 1, calculating precipitation correction values corresponding to all the precipitation prediction values according to each precipitation prediction value and each candidate value in the precipitation correction threshold candidate value sequence corresponding to the precipitation magnitude, and according to the following first calculation formula:
s222, when the precipitation magnitude is 1< k < N, calculating precipitation correction values corresponding to all the precipitation prediction values according to each precipitation prediction value and each candidate value in the precipitation correction threshold candidate value sequence corresponding to the precipitation magnitude, and according to the following second calculation formula:
s223, when the precipitation magnitude k is equal to N, calculating precipitation correction values corresponding to all the precipitation prediction values according to each precipitation prediction value and each candidate value in the precipitation correction threshold candidate value sequence corresponding to the precipitation magnitude, and according to the following third calculation formula:
wherein y is a precipitation setting value, x is a precipitation forecast value, and FkA precipitation correction threshold value corresponding to the precipitation magnitude k is determined, and the corresponding precipitation correction threshold value candidate value sequence is { F }k1，Fk2，…，Fkn}；OkA precipitation threshold value for each precipitation magnitude k.
S3, aiming at the same precipitation magnitude, performing forecast TS scoring according to precipitation correction values, actual precipitation values and precipitation thresholds corresponding to the precipitation magnitudes after different precipitation correction threshold processing; and taking the precipitation correction threshold corresponding to the highest TS scoring result as the optimal precipitation correction threshold of the precipitation magnitude, so as to obtain the optimal precipitation correction threshold corresponding to each precipitation magnitude, and taking the optimal precipitation correction threshold as an optimal precipitation correction threshold sequence.
According to the precipitation setting value y, the actual precipitation value z and the precipitation threshold value OkPerforming TS scoring to determine the optimal precipitation correction threshold F in the 10 candidate values of the precipitation threshold candidate value sequencekSo that during the subsequent operation, the threshold value F is corrected according to the optimal precipitationkThe correction of the rainfall forecast value of the forecast day is realized.
Specifically, for the same precipitation magnitude k, according to precipitation correction values and actual precipitation values after treatment of different precipitation correction threshold values and precipitation threshold values corresponding to each precipitation magnitude, forecasting TS scoring is performed through the following TS scoring formula:
wherein TS represents the forecast TS scoring result; NA represents the correct number of precipitation forecasts in the training period; NB represents the number of precipitation empty reports in the training period; and NC represents the number of times of missed reports of rainfall in the training period. That is, NA represents that any precipitation corrected value of any training day in the training period and the actual precipitation value of the training day both reach the precipitation threshold O corresponding to the precipitation magnitude kkThe number of times of (c); NB represents that any precipitation order value of any training day in the training period reaches the precipitation threshold value O corresponding to the precipitation order value kkAnd the actual precipitation value of the training day does not reach the precipitation threshold value O corresponding to the precipitation magnitude kkThe number of times of (c); NC represents that any precipitation correction value on any training day in the training period does not reach the precipitation threshold value O corresponding to the precipitation magnitude kkAnd the actual precipitation value of the training day reaches the precipitation threshold value O corresponding to the precipitation magnitude kkThe number of times.
In the embodiment of the invention, the precipitation correction threshold value F corresponding to each precipitation magnitude k is determined by adopting an iterative correction processing modek. In one embodiment, set N to 10 precipitation levels is used as an example.
First, for a precipitation level k equal to 1, a precipitation correction threshold F is determined by the first calculation formula1The value of (c). Taking 40 training days in a training period corresponding to a forecast day, and 100 stations for rainfall prediction each day, the forecast values of the rainfall in the training period are 4000, and the actual rainfall values are 40 as an example. According to a first calculation formula, correcting 4000 rainfall forecast values x and the rainfall F within a training period1The corresponding candidate value sequence is { F1，1，F1，2，…，F1，10The first candidate value in (f) }F1，1Comparing, and when the forecast value x of the precipitation is less than F1，1If so, making the precipitation amount preset value y corresponding to the precipitation amount forecast value x equal to 0; when the predicted precipitation value x is larger than or equal to F1，1And then, keeping the precipitation correction value y corresponding to the precipitation prediction value x unchanged, namely, keeping y equal to x, and obtaining the first group of data of the precipitation correction value after the emptying threshold processing. And (4) performing forecast TS scoring on the first group of precipitation correction value data. When any precipitation setting value y and the actual precipitation value z of the training day reach the corresponding precipitation threshold value O1I.e. the presence of y ≧ O1And z is more than or equal to O1When, the numerical value of NA is added with 1; when any precipitation booking value y reaches the corresponding precipitation threshold value O1And the actual rainfall value z of the training day does not reach the corresponding rainfall threshold value O1I.e. the presence of y ≧ O1And z is less than O1When, the value of NB is increased by 1; when any precipitation setting value y does not reach the corresponding precipitation threshold value O1And the actual precipitation value z of the training day reaches the corresponding precipitation threshold value O1I.e. the presence of y < O1And z is more than or equal to O1When so, the numerical value of NC is incremented by 1. After numerical values of NA, NB and NC are determined, calculating to obtain a precipitation correction threshold value candidate value F according to the TS scoring formula1，1The TS scoring result of (1).
Further, 4000 rainfall forecast values x in the training period and a rainfall correction threshold F are compared1The corresponding candidate value sequence is { F1，1，F1，2，…，F1，10The second candidate value F in1，2Comparing, and when the forecast value x of the precipitation is less than F1，2Then, the precipitation amount preset value y corresponding to the precipitation amount predicted value x is made equal to 0. When the predicted precipitation value x is larger than or equal to F1，2And keeping the precipitation amount correction value y corresponding to the precipitation amount forecast value x unchanged, namely, keeping y equal to x, and obtaining a second group of data of the precipitation amount forecast value after the emptying threshold value processing. And (4) performing forecast TS scoring on the second group of rainfall correction value data, and calculating to obtain a rainfall correction threshold value candidate value F1，2The TS scoring result of (1).
And so on, thereby obtaining 10 groups of drops after the correction threshold value candidate values of different precipitation amounts are processed under the condition that the same precipitation magnitude k is 1Data of water forecast values, and corresponding 10 TS scoring results. The higher the TS score result, namely the closer the TS value is to 1, the better the precipitation forecast effect is. Taking the precipitation correction threshold value candidate value corresponding to the TS scoring result with the highest value as the optimal precipitation correction threshold value F of the precipitation magnitude1。
Then, for the precipitation magnitude k being 2, the second calculation formula is used, and the calculated optimal precipitation correction threshold value F is obtained1To determine the precipitation correction threshold F2The value of (c). Correcting 4000 rainfall forecast values x and the rainfall in a training period by using a threshold value F2The corresponding candidate value sequence is { F2，1，F2，2，…，F2，10The first candidate value F in2，1Comparing, and when the forecast value of the precipitation quantity meets F1≤x＜F2，1Then, the precipitation amount predicted value x is made to correspond to the precipitation amount set value
When the precipitation prediction value x is F2，1Then, the precipitation amount correction value y corresponding to the precipitation amount prediction value x is kept unchanged, that is, y is equal to x, and data of the first set of precipitation amount correction values are obtained. The forecast TS scoring is carried out on the first group of rainfall correction value data, and the candidate value F of the rainfall correction threshold value is obtained through calculation2，1The TS scoring result of (1). And so on, finally determining the corresponding optimal precipitation correction threshold F when the precipitation magnitude k is 22。
Optimal precipitation correction threshold F3～F9Method for determining (D) and (F)2Similarly, the description is omitted here.
Then, for the precipitation level k equal to 10, the precipitation correction threshold F is determined by the third calculation formula10The value of (c). Correcting 4000 rainfall forecast values x and the rainfall in a training period by using a threshold value F10The corresponding candidate value sequence is { F10，1，F10，2，…，F10，10The first candidate value F in10，1Comparing, when the forecast value of the precipitation satisfies x is more than or equal to F10，1Then, the precipitation amount predicted value x is made to correspond to the precipitation amount set value
When the predicted precipitation value x is less than F10，1Then, the precipitation amount correction value corresponding to the precipitation amount prediction value x is kept unchanged, that is, y is equal to x, and data of the first set of precipitation amount correction values are obtained. The forecast TS scoring is carried out on the first group of rainfall correction value data, and the candidate value F of the rainfall correction threshold value is obtained through calculation10，1The TS scoring result of (1). And the like, so that when the precipitation magnitude k is finally determined to be 10, the corresponding optimal precipitation correction threshold F10。
By analogy, the optimal precipitation correction threshold F corresponding to each precipitation magnitude k can be obtainedkAs an optimal precipitation correction threshold sequence { F }1，F2，…，F10}。
And S4, correcting the rainfall forecast value of the forecast day according to the optimal rainfall correction threshold sequence to obtain a rainfall correction value of the forecast day as a final rainfall forecast result of the forecast day.
Preferably, the sequence of thresholds { F) is corrected according to said optimal precipitation1，F2，…，FNAnd the predicted rainfall value x 'of the forecast day is calculated by the following rainfall correction formula to obtain the predicted rainfall value y' of the forecast day:
in the embodiment of the invention, the precipitation forecast value x' according to the forecast day and the optimal precipitation correction threshold value sequence { F }1，F2，…，FNAnd determining a corresponding calculation formula of the precipitation correction value y ', and further calculating to obtain the precipitation correction value y' of the forecast day as a final precipitation forecast result of the forecast day.
The embodiment of the invention provides a correcting method of a numerical mode multi-level rainfall forecast value, which is characterized in that a corresponding rainfall forecast value is calculated and TS scoring is carried out according to the rainfall forecast value and an actual rainfall value in a training period corresponding to a forecast day and a preset rainfall magnitude standard, so as to determine an optimal rainfall correction threshold sequence for correcting the rainfall forecast value of the forecast day. The embodiment of the invention adopts the quasi-symmetrical sliding training period of 1 year to carry out the statistical analysis of the optimal rainfall correction threshold value, can better self-adaptively predict the seasonal background similar to the day before and after, and ensures the timeliness of the training data. In addition, the historical rainfall forecast value and the actual rainfall value are used as training data, can be directly obtained through a numerical weather forecast mode, are suitable for obtaining grid point data or station data, and are simple and easy to obtain. The precipitation correction forecast score under each precipitation magnitude in the training period corresponding to the forecast day is calculated to determine the optimal precipitation correction threshold sequence for correcting the precipitation forecast value on the forecast day, so that the precipitation forecast value is corrected day by day, the precipitation forecast errors on the forecast days on different precipitation magnitudes are effectively reduced, and the accuracy of the precipitation forecast is improved.
Fig. 2 is a schematic structural diagram of a correcting device for a numerical mode multi-level precipitation forecast value according to a second embodiment of the present invention. The second embodiment of the present invention provides a correcting device 20 for a numerical mode multi-level precipitation forecast value, which includes: the rainfall data acquisition module 21, the rainfall correction value calculation module 22, the optimal rainfall correction threshold calculation module 23 and the rainfall forecast result acquisition module 24; wherein the content of the first and second substances,
the rainfall data obtaining module 21 is configured to determine a training period corresponding to a forecast day, and obtain a rainfall forecast value of each forecast station for each training day in the training period, and an actual rainfall value of each training day.
The precipitation correcting value calculating module 22 is configured to calculate precipitation correcting values after being processed by different precipitation correcting thresholds at each precipitation magnitude according to all precipitation forecast values in the training period; n precipitation magnitude values and a precipitation threshold value corresponding to each precipitation magnitude value are preset, and N is larger than 1.
The optimal precipitation correction threshold calculation module 23 is configured to perform forecast TS scoring according to precipitation correction values, actual precipitation values, and precipitation thresholds corresponding to each precipitation magnitude, which are processed by different precipitation correction thresholds, for the same precipitation magnitude; and taking the precipitation correction threshold corresponding to the highest TS score as the optimal precipitation correction threshold of the precipitation magnitude, so as to obtain the optimal precipitation correction threshold corresponding to each precipitation magnitude as the optimal precipitation correction threshold sequence.
And the precipitation forecast result obtaining module 24 is configured to correct the precipitation forecast value on the forecast day according to the optimal precipitation correction threshold sequence, so as to obtain a precipitation correction value on the forecast day, which is used as a final precipitation forecast result on the forecast day.
In a preferred embodiment, the precipitation correction value calculating module 22 is specifically configured to:
presetting precipitation correction threshold F corresponding to each precipitation magnitude kkThe n candidate values are used as a precipitation correction threshold value candidate value sequence { F) corresponding to the precipitation magnitude kk1，Fk2，…，Fkn}; wherein k represents the number of each precipitation magnitude, and k is more than or equal to 1 and less than or equal to N.
And aiming at the same precipitation magnitude k, calculating to obtain precipitation correction values corresponding to all the precipitation correction values after different precipitation correction threshold values are processed according to each precipitation correction value and each candidate value in the precipitation correction threshold value candidate value sequence corresponding to the precipitation magnitude.
As a preferred embodiment, for the same precipitation magnitude k, calculating, according to each precipitation forecast value and each candidate value in the precipitation correction threshold candidate value sequence corresponding to the precipitation magnitude, precipitation correction values corresponding to all the precipitation forecast values after different precipitation correction threshold treatments, specifically including:
when the precipitation magnitude k is equal to 1, calculating precipitation correction values corresponding to all the precipitation prediction values according to each precipitation prediction value and each candidate value in the precipitation correction threshold candidate value sequence corresponding to the precipitation magnitude according to the following first calculation formula:
when the precipitation magnitude is more than 1 and less than k and less than N, calculating precipitation correction values corresponding to all the precipitation prediction values according to each precipitation prediction value and each candidate value in the precipitation correction threshold value candidate value sequence corresponding to the precipitation magnitude and according to the following second calculation formula:
when the precipitation magnitude k is equal to N, calculating precipitation correction values corresponding to all the precipitation prediction values according to each precipitation prediction value and each candidate value in the precipitation correction threshold candidate value sequence corresponding to the precipitation magnitude according to the following third calculation formula:
wherein y is a precipitation setting value, x is a precipitation forecast value, and FkA precipitation correction threshold value corresponding to the precipitation magnitude k is determined, and the corresponding precipitation correction threshold value candidate value sequence is { F }k1，Fk2，…，Fkn}；OkA precipitation threshold value for each precipitation magnitude k.
Preferably, the performing, for the same precipitation magnitude, forecast TS scoring according to the precipitation correction value after the treatment of different precipitation correction threshold values, the actual precipitation value, and the precipitation threshold value corresponding to each precipitation magnitude specifically includes:
aiming at the same precipitation magnitude k, according to precipitation correction values after different precipitation correction threshold values are processed, actual precipitation values and precipitation threshold values corresponding to each precipitation magnitude, forecasting TS scoring is carried out through the following TS scoring formula:
where TS represents forecast TS scoreThe result is; NA represents that any precipitation corrected value of any training day in the training period and the actual precipitation value of the training day both reach the precipitation threshold value O corresponding to the precipitation magnitude kkThe number of times of (c); NB represents that any precipitation order value of any training day in the training period reaches the precipitation threshold value O corresponding to the precipitation order value kkAnd the actual precipitation value of the training day does not reach the precipitation threshold value O corresponding to the precipitation magnitude kkThe number of times of (c); NC represents that any precipitation correction value on any training day in the training period does not reach the precipitation threshold value O corresponding to the precipitation magnitude kkAnd the actual precipitation value of the training day reaches the precipitation threshold value O corresponding to the precipitation magnitude kkThe number of times.
In a preferred embodiment, the precipitation forecast result obtaining module 24 is specifically configured to:
correcting a threshold sequence { F) according to the optimal precipitation1，F2，…，FNAnd the predicted rainfall value x 'of the forecast day is calculated by the following rainfall correction formula to obtain the predicted rainfall value y' of the forecast day:
wherein, OkA precipitation threshold value for each precipitation magnitude k.
It should be noted that, the XXX apparatus provided in the embodiment of the present invention is used for executing all the process steps of the XXX method in the above embodiment, and the working principles and beneficial effects of the two are in one-to-one correspondence, so that details are not described again.
The embodiment two of the invention provides a correcting device for the numerical-mode multi-level rainfall forecast value, which is used for calculating a corresponding rainfall forecast value and performing TS scoring through the rainfall forecast value in a training period corresponding to a forecast day, an actual rainfall value and a preset rainfall magnitude standard so as to determine and obtain an optimal rainfall correction threshold sequence for correcting the rainfall forecast value of the forecast day. The embodiment of the invention adopts the quasi-symmetrical sliding training period of 1 year to carry out the statistical analysis of the optimal rainfall correction threshold value, can better self-adaptively predict the seasonal background similar to the day before and after, and ensures the timeliness of the training data. In addition, the historical rainfall forecast value and the actual rainfall value are used as training data, can be directly obtained through a numerical weather forecast mode, are suitable for obtaining grid point data or station data, and are simple and easy to obtain. The precipitation correction forecast score under each precipitation magnitude in the training period corresponding to the forecast day is calculated to determine the optimal precipitation correction threshold sequence for correcting the precipitation forecast value on the forecast day, so that the precipitation forecast value is corrected day by day, the precipitation forecast errors on the forecast days on different precipitation magnitudes are effectively reduced, and the accuracy of the precipitation forecast is improved.
Fig. 3 is a schematic structural diagram of another device for correcting a numerical-mode multi-level rainfall forecast value according to the third embodiment of the present invention. The device 30 for correcting numerical mode multi-level precipitation forecast values provided by the embodiment of the invention comprises a processor 31, a memory 32 and a computer program stored in the memory and configured to be executed by the processor, wherein the processor implements the method for correcting numerical mode multi-level precipitation forecast values as described in the first embodiment of the invention when executing the computer program.
An embodiment of the present invention further provides a computer-readable storage medium, where the computer-readable storage medium includes a stored computer program, where when the computer program runs, the apparatus where the computer-readable storage medium is located is controlled to execute the method for correcting the numerical mode multiple precipitation forecast value according to the first embodiment.
It will be understood by those skilled in the art that all or part of the processes of the methods of the embodiments described above can be implemented by a computer program, which can be stored in a computer-readable storage medium, and when executed, can include the processes of the embodiments of the methods described above. The storage medium may be a magnetic disk, an optical disk, a Read-only memory (ROM), a Random Access Memory (RAM), or the like.
While the foregoing is directed to the preferred embodiment of the present invention, it will be understood by those skilled in the art that various changes and modifications may be made without departing from the spirit and scope of the invention.
Claims (10)
1. A method for correcting numerical mode multi-level rainfall forecast value is characterized by comprising the following steps:
determining a training period corresponding to a forecast day, and acquiring a rainfall forecast value of each forecast station to each training day in the training period and an actual rainfall value of each training day;
calculating precipitation correcting values after different precipitation correcting threshold values are processed under each precipitation magnitude according to all precipitation predicted values in the training period; n precipitation magnitude values and a precipitation threshold value corresponding to each precipitation magnitude value are preset, wherein N is greater than 1;
aiming at the same precipitation magnitude, performing forecast TS scoring according to precipitation correction values after different precipitation correction threshold values are processed, actual precipitation values and precipitation threshold values corresponding to the precipitation magnitude; taking the precipitation correction threshold corresponding to the TS scoring result with the highest score as the optimal precipitation correction threshold of the precipitation magnitude, so as to obtain the optimal precipitation correction threshold corresponding to each precipitation magnitude as the optimal precipitation correction threshold sequence;
and correcting the rainfall forecast value of the forecast day according to the optimal rainfall correction threshold sequence to obtain a rainfall correction value of the forecast day as a final rainfall forecast result of the forecast day.
2. The method for correcting the numerical mode multi-level precipitation forecast value according to claim 1, wherein the step of calculating the precipitation correction value after being processed by different precipitation correction threshold values at each precipitation level according to all the precipitation forecast values in the training period specifically comprises the steps of:
presetting precipitation correction threshold F corresponding to each precipitation magnitude kkThe n candidate values are used as precipitation orders corresponding to the precipitation magnitude kPositive threshold candidate sequence { Fk1，Fk2，…，Fkn}; wherein k represents the number of each precipitation magnitude, and k is more than or equal to 1 and less than or equal to N;
and aiming at the same precipitation magnitude k, calculating to obtain precipitation correction values corresponding to all the precipitation correction values after different precipitation correction threshold values are processed according to each precipitation correction value and each candidate value in the precipitation correction threshold value candidate value sequence corresponding to the precipitation magnitude.
3. The method according to claim 2, wherein the step of calculating, for the same precipitation magnitude k, precipitation correction values corresponding to all the precipitation prediction values after different precipitation correction threshold treatments according to each precipitation prediction value and each candidate value in the precipitation correction threshold candidate value sequence corresponding to the precipitation magnitude comprises:
when the precipitation magnitude k is equal to 1, calculating precipitation correction values corresponding to all the precipitation prediction values according to each precipitation prediction value and each candidate value in the precipitation correction threshold candidate value sequence corresponding to the precipitation magnitude according to the following first calculation formula:
when the precipitation magnitude is 1< k < N, calculating precipitation correction values corresponding to all the precipitation prediction values according to each precipitation prediction value and each candidate value in the precipitation correction threshold candidate value sequence corresponding to the precipitation magnitude according to the following second calculation formula:
when the precipitation magnitude k is equal to N, calculating precipitation correction values corresponding to all the precipitation prediction values according to each precipitation prediction value and each candidate value in the precipitation correction threshold candidate value sequence corresponding to the precipitation magnitude according to the following third calculation formula:
wherein y is a precipitation setting value, x is a precipitation forecast value, and FkA precipitation correction threshold value corresponding to the precipitation magnitude k is determined, and the corresponding precipitation correction threshold value candidate value sequence is { F }k1，Fk2，…，Fkn}；OkA precipitation threshold value for each precipitation magnitude k.
4. The method for correcting the numerical-mode multi-level rainfall forecast value of claim 1, wherein the step of performing forecast TS scoring for the same rainfall level according to the rainfall correction value after the precipitation correction threshold processing, the actual rainfall value and the rainfall threshold corresponding to each rainfall level comprises:
aiming at the same precipitation magnitude k, according to precipitation correction values after different precipitation correction threshold values are processed, actual precipitation values and precipitation threshold values corresponding to each precipitation magnitude, forecasting TS scoring is carried out through the following TS scoring formula:
wherein TS represents the forecast TS scoring result; NA represents that any precipitation corrected value of any training day in the training period and the actual precipitation value of the training day both reach the precipitation threshold value O corresponding to the precipitation magnitude kkThe number of times of (c); NB represents that any precipitation order value of any training day in the training period reaches the precipitation threshold value O corresponding to the precipitation order value kkAnd the actual precipitation value of the training day does not reach the precipitation threshold value O corresponding to the precipitation magnitude kkThe number of times of (c); NC represents that any precipitation correction value on any training day in the training period does not reach the precipitation threshold value O corresponding to the precipitation magnitude kkAnd should trainThe actual precipitation value of the practice day reaches the precipitation threshold value O corresponding to the precipitation magnitude kkThe number of times.
5. The method for correcting the numerical mode multi-level precipitation forecast value according to claim 1, wherein the method for correcting the precipitation forecast value on the forecast day according to the optimal precipitation correction threshold sequence to obtain the precipitation correction value on the forecast day as the final precipitation forecast result on the forecast day specifically comprises:
correcting a threshold sequence { F) according to the optimal precipitation1，F2，…，FNAnd the predicted rainfall value x 'of the forecast day is calculated by the following rainfall correction formula to obtain the predicted rainfall value y' of the forecast day:
wherein, OkA precipitation threshold value for each precipitation magnitude k.
6. The method for correcting the numerical-mode multi-level rainfall forecast value of claim 1, wherein the determining of the training period corresponding to the forecast day is specifically:
acquiring 20 days before the forecast day and 20 days after the forecast day on the same day of the previous year as training periods of the forecast day.
7. The method for correcting the numerical mode multi-level precipitation forecast value of claim 1, wherein the preset k precipitation levels and the precipitation threshold corresponding to each precipitation level are as follows:
presetting 10 precipitation magnitude levels and precipitation threshold O corresponding to each precipitation magnitude levelk(ii) a Wherein, the precipitation threshold value that each precipitation magnitude corresponds is respectively: 0.1mm, 1mm, 5mm, 10mm, 25mm, 35mm, 50mm, 75mm, 100mm, 150 mm.
8. A device for correcting numerical mode multi-level rainfall forecast value is characterized by comprising:
the rainfall data acquisition module is used for determining a training period corresponding to the forecast day, and acquiring a rainfall forecast value of each forecast station to each training day in the training period and an actual rainfall value of each training day;
the rainfall correction value calculating module is used for calculating rainfall correction values processed by different rainfall correction threshold values under each rainfall magnitude according to all rainfall forecast values in the training period; n precipitation magnitude values and a precipitation threshold value corresponding to each precipitation magnitude value are preset, wherein N is greater than 1;
the optimal precipitation correction threshold calculation module is used for carrying out forecast TS scoring according to precipitation correction values and actual precipitation values which are processed by different precipitation correction thresholds and precipitation thresholds corresponding to each precipitation magnitude aiming at the same precipitation magnitude; taking the precipitation correction threshold corresponding to the highest TS score as the optimal precipitation correction threshold of the precipitation magnitude, so as to obtain the optimal precipitation correction threshold corresponding to each precipitation magnitude as the optimal precipitation correction threshold sequence;
and the precipitation forecast result obtaining module is used for correcting the precipitation forecast value of the forecast day according to the optimal precipitation correction threshold sequence to obtain a precipitation correction value of the forecast day as a final precipitation forecast result of the forecast day.
9. An apparatus for correcting a numerical mode multiple precipitation forecast value, comprising a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, the processor implementing the method of correcting a numerical mode multiple precipitation forecast value according to any of claims 1 to 7 when executing the computer program.
10. A computer-readable storage medium, comprising a stored computer program, wherein the computer program, when executed, controls an apparatus on which the computer-readable storage medium is located to perform the method of correcting a numerical mode multiple precipitation forecast value according to any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010847121.5A CN112051627B (en) | 2020-08-21 | 2020-08-21 | Method, device and medium for correcting numerical mode multi-level rainfall forecast value |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010847121.5A CN112051627B (en) | 2020-08-21 | 2020-08-21 | Method, device and medium for correcting numerical mode multi-level rainfall forecast value |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112051627A true CN112051627A (en) | 2020-12-08 |
| CN112051627B CN112051627B (en) | 2022-04-12 |
Family
ID=73599667
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010847121.5A Active CN112051627B (en) | 2020-08-21 | 2020-08-21 | Method, device and medium for correcting numerical mode multi-level rainfall forecast value |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112051627B (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112819237A (en) * | 2021-02-08 | 2021-05-18 | 广东省气象台(南海海洋气象预报中心) | Precipitation multi-scale fusion forecasting method and device |
| CN113128778A (en) * | 2021-04-27 | 2021-07-16 | 最美天气(上海)科技有限公司 | Model training method based on graded TS meteorological scoring |
| CN113466968A (en) * | 2021-07-15 | 2021-10-01 | 海南省气象台 | Multi-mode rainfall forecast correction method based on frequency matching and dynamic fusion |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPH10334071A (en) * | 1997-06-02 | 1998-12-18 | Nippon Telegr & Teleph Corp <Ntt> | Method and device for parallel generalized learning for neural circuit network model |
| CN109376913A (en) * | 2018-09-30 | 2019-02-22 | 北京市天元网络技术股份有限公司 | The prediction technique and device of precipitation |
| CN110163426A (en) * | 2019-05-09 | 2019-08-23 | 中国科学院深圳先进技术研究院 | A kind of multi-mode integrates precipitation forecast method and device |
| CN111257970A (en) * | 2018-11-30 | 2020-06-09 | 中国电力科学研究院有限公司 | Rainfall forecast correction method and system based on ensemble forecast |
-
2020
- 2020-08-21 CN CN202010847121.5A patent/CN112051627B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPH10334071A (en) * | 1997-06-02 | 1998-12-18 | Nippon Telegr & Teleph Corp <Ntt> | Method and device for parallel generalized learning for neural circuit network model |
| CN109376913A (en) * | 2018-09-30 | 2019-02-22 | 北京市天元网络技术股份有限公司 | The prediction technique and device of precipitation |
| CN111257970A (en) * | 2018-11-30 | 2020-06-09 | 中国电力科学研究院有限公司 | Rainfall forecast correction method and system based on ensemble forecast |
| CN110163426A (en) * | 2019-05-09 | 2019-08-23 | 中国科学院深圳先进技术研究院 | A kind of multi-mode integrates precipitation forecast method and device |
Non-Patent Citations (1)
| Title |
|---|
| 翟振芳 等: "安徽省ECMWF数值模式降水预报性能的检验", 《气象与环境学报》 * |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112819237A (en) * | 2021-02-08 | 2021-05-18 | 广东省气象台(南海海洋气象预报中心) | Precipitation multi-scale fusion forecasting method and device |
| CN112819237B (en) * | 2021-02-08 | 2021-09-14 | 广东省气象台(南海海洋气象预报中心) | Precipitation multi-scale fusion forecasting method and device |
| CN113128778A (en) * | 2021-04-27 | 2021-07-16 | 最美天气(上海)科技有限公司 | Model training method based on graded TS meteorological scoring |
| CN113466968A (en) * | 2021-07-15 | 2021-10-01 | 海南省气象台 | Multi-mode rainfall forecast correction method based on frequency matching and dynamic fusion |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112051627B (en) | 2022-04-12 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112051627B (en) | Method, device and medium for correcting numerical mode multi-level rainfall forecast value | |
| Bremnes | Ensemble postprocessing using quantile function regression based on neural networks and Bernstein polynomials | |
| CN108428017B (en) | Wind power interval prediction method based on nuclear extreme learning machine quantile regression | |
| CN108764515A (en) | A kind of reservoir operation Application of risk decision method of Coupled Numerical meteorological model DATA PROCESSING IN ENSEMBLE PREDICTION SYSTEM | |
| CN110727665B (en) | Internet of things equipment reported data quality analysis method and system | |
| CN110019124B (en) | Method and system for revising aggregated weather forecast | |
| Bogner et al. | Combining quantile forecasts and predictive distributions of streamflows | |
| CN112508299A (en) | Power load prediction method and device, terminal equipment and storage medium | |
| Zaw et al. | Empirical statistical modeling of rainfall prediction over Myanmar | |
| CN111784023A (en) | Short-time adjacent fog visibility prediction method | |
| CN113312578A (en) | Data index fluctuation attribution method, device, equipment and medium | |
| JP2022510667A (en) | Methods and equipment for predicting energy consumption | |
| CN113807545A (en) | River and lake ecological flow forecasting and early warning method based on deep learning and physical model | |
| CN116503206A (en) | Warehouse-in runoff reconstruction method, computer equipment and medium | |
| Bennett et al. | An error model for long-range ensemble forecasts of ephemeral rivers | |
| US11625635B2 (en) | Computer-readable storage medium storing control program, control method, and control device | |
| CN110046756B (en) | Short-term weather forecasting method based on wavelet denoising and Catboost | |
| CN112700050B (en) | Method and system for predicting ultra-short-term 1 st point power of photovoltaic power station | |
| CN117435873B (en) | Data management method based on intelligent spraying dust fall | |
| CN111563236B (en) | Short-term wind speed prediction method and device | |
| CN116227785A (en) | Runoff change attribution method, device and system considering climate-vegetation-water taking influence | |
| CN115907192A (en) | Method and device for generating wind power fluctuation interval prediction model and electronic equipment | |
| CN112034535B (en) | Correcting method, system and storage medium for numerical model rainfall forecast | |
| CN112529183A (en) | Knowledge distillation-based model self-adaptive updating method | |
| CN112446550A (en) | Short-term building load probability density prediction method |
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 |