CN111766642A - Daily precipitation forecast system for landfall tropical cyclone - Google Patents

Abstract

The invention relates to a daily precipitation forecasting system for a landing tropical cyclone according to the present invention, which includes: a generalized initial value building module, receiving the path of a historical TC, obtaining the forecast path of a target TC at the starting time of a certain path, and converting the target Prediction path of TC The observation path before the starting time of the path is merged into the target TC path; the TC speed information of the target TC and the historical TC in a specific daily-scale time period is obtained by processing; the initial value similarity discrimination module constructs the daily-scale similar area , identify the closest point and the shortest distance of the historical TC, calculate the similarity between the historical TC path and the target TC path, mark the historical TC whose moving speed difference with the target TC reaches the threshold, and mark the historical TC whose shortest distance is greater than a certain threshold, select m The best similar historical TCs are sent to the ensemble forecasting module; the ensemble forecasting module obtains the specific daily precipitation fields of the best historical TCs and assembles them. The present invention has good prediction performance for the daily precipitation of landfalling TC in China.

Description

Daily precipitation forecast system for landfall tropical cyclone

technical field

The invention relates to a weather forecast technology, in particular to a daily precipitation forecast system for landing tropical cyclones based on a dynamic-statistic-similar ensemble forecast model.

Background technique

For weather forecasting, some scholars have proposed the Dynamic Statistical Similarity Ensemble Prediction (DSAEF) theory (model) based on accurate models. Subsequently, the DSAEF forecasting technology (DSAEF_LTP_A 1.0 system) for precipitation in the process of landfalling tropical cyclones came into being, and the example proved that the forecasting performance of this system is good. However, there is currently a lack of forecasting techniques for DSAEF models for the daily precipitation of landfalling tropical cyclones. The daily precipitation forecasting system of the landing tropical cyclone of the present invention is the forecasting technology of the DSAEF model for the daily precipitation of the landing tropical cyclone (DSAEF_LTP_D 1.0 system).

SUMMARY OF THE INVENTION

The purpose of the present invention is to provide a daily precipitation forecasting system for landing tropical cyclones, which is used to solve the above-mentioned problems of the prior art.

The present invention is a daily precipitation forecasting system for landing tropical cyclones, which includes: a generalized initial value building module, which receives the path of the historical TC, obtains the forecast path of the target TC at the starting time of a certain path, and uses the forecast path of the target TC according to the forecast path of the target TC. The observation paths before the path start time are merged into the target TC path; the TC speed information of the target TC and the historical TC in a specific daily-scale time period is obtained by processing; the initial value similarity judgment module constructs the daily-scale similar area, and identifies the nearest historical TC. Points and the shortest distance, calculate the similarity between the historical TC path and the target TC path, mark the historical TC whose moving speed difference with the target TC reaches the threshold, and mark the historical TC whose shortest distance is greater than a certain threshold, and select m best similar histories The TC is sent to the ensemble forecast module; the ensemble forecast module obtains the specific daily precipitation field of the best historical TC and assembles it.

According to an embodiment of the landfall tropical cyclone daily precipitation forecast system of the present invention, the generalized initial value building module includes a TC path acquisition module and a TC daily movement speed acquisition module; the TC path acquisition module is used to acquire targets from a numerical weather forecast model The forecast path of the TC at the start time of a certain path, and the forecast path of the target TC and the observation path before the start time of the path are merged into the target TC path. The TC path and the historical TC path are sent to the initial value similarity determination module; the TC daily movement speed acquisition module is used to obtain the TC movement speed information of the target TC and the historical TC in a certain 24-hour time period.

According to an embodiment of the daily precipitation forecasting system for landing tropical cyclones of the present invention, the initial value similarity determination module includes a daily-scale similar area building module, which takes the target TC daily precipitation forecast starting point as any point on the target TC path, and is similar to One end point A of the diagonal line of the area slides at the point A1 before the end point A and several TC positions before it, and the other end point B of the diagonal line of the similar area slides at the TC position after the point A1, and the farthest is the maximum forecast point of time.

According to an embodiment of the landfall tropical cyclone daily precipitation forecasting system of the present invention, the initial value similarity determination module further includes: sending the constructed similar area information to the path similarity index calculation module and the shortest distance identification module; the path similarity index calculation module The module is used to calculate the path similarity area index for the target TC and all historical TCs one by one in the similar area, and arrange the historical TCs from low to high according to the size of the TSAI, and send the sorting result to the best similar historical TC for determination Module; the shortest distance identification module is used to identify the observation point of each historical TC that is closest to the starting point of the daily precipitation forecast of the target TC in a similar area, as well as the corresponding shortest distance, and eliminate the history whose shortest distance is greater than a certain set value. TC; TC movement speed similarity discrimination module, receives the movement speed information of target TC and historical TC generated by the TC daily movement speed acquisition module; then calculates the difference between all historical TC and target TC movement speed one by one, and marks the absolute value of the difference For historical TCs greater than a certain threshold, the marked historical TC numbers are sent to the best similar historical TC determination module; the best similar historical TC determination module first receives the TSAI value sorting results of all historical TCs generated in the TC path similarity calculation module , receive the historical TC number, and remove all marked historical TCs; the remaining historical TCs are determined as the best similar historical TCs, and the TC numbers are sent to the TC ensemble forecasting module.

According to an embodiment of the landfall tropical cyclone daily precipitation forecasting system of the present invention, the shortest distance identification module includes a closest point identification module and a shortest distance identification module; the closest point identification module in a selected similar area, calculates all historical records in turn The distance between all the observation points of the TC and the starting point of the target TC daily precipitation forecast, send the calculated closest point information of all historical TCs to the TC daily precipitation acquisition module, and send the shortest distance information to the shortest distance discrimination module; the shortest distance discrimination module then After receiving the shortest distance information, mark the historical TC, and send the marked historical TC number to the best similar historical TC determination module.

According to an embodiment of the landfall tropical cyclone daily precipitation forecasting system of the present invention, the TC precipitation ensemble forecasting module includes a similar TC daily precipitation acquisition module and a precipitation ensemble module; the similar TC daily precipitation acquisition module is used to receive the best similar historical TC to determine The number of the best similar historical TC generated in the module, calculate the accumulated precipitation in 24 hours after its closest point, and then use the objective weather map analysis method to identify the TC precipitation, so as to obtain the corresponding similar TC daily precipitation field, and use It is sent to the precipitation aggregation module; the precipitation aggregation module is used to aggregate the TC daily precipitation fields generated in the similar TC daily precipitation acquisition modules into one precipitation field. The precipitation forecast results within 24 hours after the start of the daily precipitation forecast.

According to an embodiment of the daily precipitation forecasting system for landing tropical cyclones of the present invention, the TC daily moving speed acquisition module receives the closest point information of all historical TCs generated by the closest point identification module, and calculates that the target TC is after the starting point of the target TC daily precipitation forecast. The average moving speed within 24 hours is used as the moving speed of the target TC, the average moving speed of all historical TCs within 24 hours after the closest point is calculated as their respective moving speeds, and the moving speed information of the target TC and all historical TCs is sent to the TC Movement speed similarity discrimination module.

According to an embodiment of the daily precipitation forecasting system for landing tropical cyclones of the present invention, the TC moving speed similarity determination module receives the moving speed information of the target TC and all historical TCs generated in the TC daily moving speed acquisition module, and calculates all historical TCs The difference between the moving speed of the target TC and the target TC is marked, and the historical TC whose difference meets a certain condition is marked, and the marked historical TC number is sent to the best similar historical TC determination module.

According to an embodiment of the daily precipitation forecasting system for landfall tropical cyclones of the present invention, the best similar historical TC determination module receives the TSAI value ranking results of all historical TCs generated in the TC path similarity calculation module, and receives the shortest distance discrimination module and All the marked historical TC numbers in the TC moving speed similarity discrimination module, remove all marked historical TCs, determine the remaining top m historical TCs as the best similar historical TCs, and send the TC numbers to the TC ensemble forecasting module .

According to an embodiment of the landfall tropical cyclone daily precipitation forecasting system of the present invention, the similar TC daily precipitation acquisition module receives the numbers of m best similar historical TCs generated in the best similar historical TC determination module, and calculates the number of the m best similar historical TCs generated in the best similar historical TC determination module, The accumulated precipitation in the 24 hours after the point, and then use the objective synoptic map analysis method to identify the TC precipitation, so as to obtain the corresponding m similar TC daily precipitation fields, and send them to the TC precipitation ensemble forecasting module.

The landfalling tropical cyclone daily precipitation forecasting system based on the dynamic-statistics-similar ensemble forecasting model of the present invention has good forecasting performance for the landfalling Chinese TC daily precipitation (especially the TC precipitation one day before the landfall).

Description of drawings

Fig. 1 shows the module diagram of the daily precipitation forecasting system of landfall tropical cyclone based on the dynamic-statistics-similar ensemble forecasting model of the present invention;

Figure 2a shows a schematic diagram of the construction of similar areas on a daily scale;

Figure 2b shows a schematic diagram of the shortest distance identification.

Detailed ways

In order to make the purpose, content, and advantages of the present invention clearer, the specific embodiments of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments.

Fig. 1 shows the module diagram of the landfall tropical cyclone (landfall TC) daily precipitation forecast system based on the dynamic-statistics-similar ensemble forecast model of the present invention. As shown in Fig. 1, the landfall tropical cyclone daily precipitation forecast system of the present invention includes: : Generalized initial value building module 1, initial value similarity discrimination module 2 and TC precipitation ensemble forecasting module 3.

As shown in Figure 1, the generalized initial value building module 1, which receives the path of the historical TC, and is used for processing to obtain the TC moving speed information of the target TC path, the target TC and the historical TC on a specific daily scale time period, the system of the present invention ( The generalized initial value of DSAEF_LTP_D 1.0) contains two factors: TC path and TC speed. Further, the generalized initial value building module 1 includes a TC path acquisition module 11 and a TC daily movement speed acquisition module 12; a TC path acquisition module 11, which is used to obtain the forecast of the target TC at the start time of a certain path from the numerical weather forecast model. It combines the forecast path of the target TC and the observation path before the start time of the path into a complete path, which is the target TC path. In addition, the TC path acquisition module 11 is also used to obtain the paths of all historical TCs (such as the 1960 TC path). Up to now), and send the target TC path and the historical TC path to the initial value similarity judgment module 2; the TC daily movement speed acquisition module 12 is used to obtain the TC movement speed information of the target TC and the historical TC in a certain 24-hour time period , which is subject to the identification result of the shortest distance identification module 23 in the initial value similarity identification module 2 .

As shown in Figure 1, the initial value similarity judgment module 2 is used to judge the similarity of the TC path and the TC moving speed contained in the generalized initial value, and finally select 5 (the setting of 5 is only an example) the best similarity The historical TC is sent to the TC precipitation ensemble forecasting module 3. Further, the initial value similarity judging module 2 includes a daily-scale similar area building module 21 , a path similarity index calculation module 22 , a shortest distance identification module 23 , a TC moving speed similarity judging module 24 and a best similar historical TC determination module 25 .

As shown in Fig. 1, the daily-scale similar area building module 21, which constructs a similar area (rectangular box) on a daily scale (referring to the scale of the moving distance of the TC within a day) through the target TC path, and this area is affected by the target TC daily precipitation forecast starting point and the farthest forecast point of the path, but its selection should also include various possibilities. Specifically, as shown in Fig. 2a, a schematic diagram of the construction of similar areas on a daily scale is shown. In principle, the starting point (A1) of the target TC daily precipitation forecast can be any point on the target TC path (the line in the figure) (the TC observation or forecast position at intervals of several hours, such as six-hour intervals), depending on forecast requirements. One end (point A) of the diagonal line of the similar area can be slid at several TC positions (solid dots) before and after point A1, and the other end (point B) can be slid at the TC position (hollow dots) after point A1, The farthest is the location of the maximum forecast time (point B1, that is, the farthest forecast point on the path), for example, the A-side can be the location 0, 12 or 24 hours before point A1, and the A-side can be the location 18, 24, 30, 36 hours after point A1 or The location 0, 6, 12 hours before point B1, which depends on the actual forecast demand (the similar area shown in Fig. 2a is just an example). Finally, the constructed similar area information is sent to the path similarity index calculation module 22 and the shortest distance identification module 23 .

The path similarity index calculation module 22 shown in FIG. 1 is used to calculate the path similarity area index (TSAI) one by one for the target TC and all the historical TCs in the similar area, and according to the size of the TSAI, the historical TC is calculated from low to low High ranking, and send the ranking result to the best similar historical TC determination module 25.

The shortest distance identification module 23 shown in FIG. 1 is used to identify the observation point (C1) closest to the target TC daily precipitation forecast starting point (A1) of each historical TC in a similar area, and the corresponding shortest distance ( d0), and remove the historical TC whose shortest distance is greater than a certain set value. Specifically, the shortest distance identification module 23 includes the closest point identification module 231 and the shortest distance identification module 232; as shown in the schematic diagram of the shortest distance identification in Figure 2b, the closest point identification module 231 refers to a selected similar area, and calculates all the The distance between all observation points (blue dots) of the historical TC (the line is an example historical TC path) and the starting point (A1) of the daily precipitation forecast of the target TC, the closest observation point is marked as point C1, and the straight line between point C1 and point A1 The distance d0 is the shortest distance between the historical TC and the target TC. Finally, the calculated closest point (C1) information of all historical TCs is sent to the TC daily precipitation acquisition module 12, and the shortest distance (d0) information is sent to the shortest distance discrimination module. 232; the shortest distance discrimination module 232, after receiving the shortest distance (d0) information, marks the historical TC of d0 greater than 190km (the setting greater than 190km is only an example, and can also be set to range from 0km to 500km, or set to infinity). , and send the marked historical TC number to the best similar historical TC determination module 25 .

As shown in FIG. 1, the TC daily moving speed acquisition module 12 first receives the closest point information of all historical TCs generated by the closest point identification module 231; ) within 24 hours as the moving speed of the target TC, calculate the average moving speed of all historical TCs within 24 hours after its closest point (point C1 in Figure 2b) as their respective moving speeds; finally, the target The moving speed information of the TC and all historical TCs is sent to the TC moving speed similarity judging module 24 .

As shown in FIG. 1, the TC speed similarity determination module 24 first receives the target TC and the speed information of all historical TCs generated by the TC daily speed acquisition module 12; then calculates the speed of all historical TCs and target TCs one by one. The difference value, mark the absolute value of the difference value is greater than 8km/h (the setting of the absolute value of the difference value exceeding 8km/h is only an example, it can also be set from the absolute value of the difference value greater than 0km/h to the absolute value of the difference value greater than 50km/h. etc., or set as a historical TC with a difference greater than zero, a difference less than zero, or a difference infinite); finally, the marked historical TC number is sent to the best similar historical TC determination module 25.

The best similar historical TC determination module 25 shown in FIG. 1 firstly receives the TSAI value ranking results of all historical TCs generated in the TC path similarity calculation module 22, and receives the shortest distance determination module 232 and the similarity determination of the TC moving speed. All marked historical TC numbers in module 24; then, all marked historical TCs are eliminated; finally, the remaining top 5 (the settings of the first 5 are only examples) historical TCs are determined as the best similar historical TCs, and Send their TC numbers to the similar TC daily precipitation acquisition module 31 in the TC ensemble forecast module 3 .

As shown in Figure 1, the TC precipitation ensemble forecasting module 3 is used to obtain the daily precipitation field of the best historical TC and use a suitable ensemble scheme to assemble it. Further, the ensemble forecast module 3 includes a similar TC daily precipitation acquisition module 31 and a precipitation aggregation module 32; the similar TC daily precipitation acquisition module 31 is used to receive the numbers of the five best similar historical TCs generated in the best similar historical TC determination module 25. , and then calculate their respective accumulated precipitation (original precipitation field) 24 hours after their closest point (point C1 in Figure 2b), and then use the objective synoptic map analysis method (OSAT) to identify them for TC precipitation, so as to obtain the corresponding The 5 similar TC daily precipitation fields are sent to the precipitation aggregation module 32; the precipitation aggregation module 32 is used to aggregate the 5 TC daily precipitation fields generated in the similar TC daily precipitation acquisition module 31 into one precipitation field, and the aggregation scheme is Take the maximum value for each station (the setting of taking the maximum value for each station is only an example, and it can also be set as a reasonable solution such as taking the average of each station), and the precipitation field obtained by the aggregation is the target TC 24 hours after the point A1 (the starting point of daily precipitation forecast). The hourly precipitation forecast results.

The name of the invention is the Landfalling Tropical Cyclone Daily Precipitation Forecast System (DSAEF_LTP_D, Dynamical-Statistical-Analog Ensemble Forecast for LandfallingTropical cyclones Daily Precipitation) based on the dynamic-statistical-similar ensemble forecasting model. The generation version of the application technology (DSAEF_LTP_D 1.0). The DSAEF model includes a generalized initial value building module, an initial value similarity discrimination module and an ensemble forecasting module. In DSAEF_LTP_D 1.0, the generalized initial value building module is used to obtain the path information of the target tropical cyclone (TC) and historical TC, obtain the target TC and The TC speed information of historical TCs in a specific daily-scale time period, the generalized initial value constructed by the TC contains two physical factors (referred to as factors or variables) of the TC path and the TC speed; the initial value similarity discrimination module is used to construct the daily-scale Similar areas, identify the closest point (C1) and the shortest distance (d0) of the historical TC, calculate the similarity between the historical TC path and the target TC path, mark the historical TC whose moving speed difference between the target TC and the target TC meets certain conditions, and mark the shortest distance (d0) ) is greater than a certain value of historical TCs, and finally select m best similar historical TCs and send them to the ensemble forecasting module; .

The landfalling tropical cyclone daily precipitation forecasting system based on the dynamic-statistics-similar ensemble forecasting model of the present invention has good forecasting performance for the landfalling Chinese TC daily precipitation (especially the TC precipitation one day before the landfall). The present invention can more accurately predict the precipitation on the landing TC day.

The above are only the preferred embodiments of the present invention. It should be pointed out that for those skilled in the art, without departing from the technical principle of the present invention, several improvements and modifications can also be made. These improvements and modifications It should also be regarded as the protection scope of the present invention.

Claims (10)

1. a landing tropical cyclone daily precipitation forecast system, is characterized in that, comprises: The generalized initial value building module receives the path of the historical TC, obtains the forecast path of the target TC at the start time of a certain path, and merges the forecast path of the target TC and the observation path before the start time of the path into the target TC path; TC speed information of target TC and historical TC in a specific daily time period; The initial value similarity discrimination module constructs a similar area on a daily scale, identifies the closest point and the shortest distance of the historical TC, calculates the similarity between the historical TC path and the target TC path, marks the historical TC whose moving speed difference with the target TC reaches the threshold, and marks the shortest For historical TCs whose distance is greater than a certain threshold, select m best similar historical TCs and send them to the ensemble forecasting module; the ensemble forecasting module obtains the specific daily precipitation fields of the best historical TCs and assembles them. 2. landing tropical cyclone daily precipitation forecasting system as claimed in claim 1, is characterized in that, generalized initial value building module comprises TC path acquisition module and TC daily moving speed acquisition module; TC path acquisition module is used for from numerical weather forecast model. Obtain the forecast path of the target TC at the start time of a certain path, and merge the forecast path of the target TC with the observation path before the start time of the path to form the target TC path. The TC path acquisition module also obtains the paths of all historical TCs. And the target TC path and the historical TC path are sent to the initial value similarity determination module; the TC daily movement speed acquisition module is used to obtain the TC movement speed information of the target TC and the historical TC in a certain 24-hour time period. 3. The daily precipitation forecasting system for landing tropical cyclones as claimed in claim 1, wherein the initial value similarity judging module comprises a daily scale similar area building module, which takes the target TC daily precipitation forecast starting point as any arbitrary value on the target TC path. One point, one end point A of the diagonal line of the similar area slides at the point A1 before the end point A and several TC positions before it, and the other end point B of the diagonal line of the similar area slides at the TC position after the point A1, the farthest is the point where the maximum forecast time is located. 4. landing tropical cyclone daily precipitation forecasting system as claimed in claim 3 is characterized in that, initial value similarity discrimination module also comprises: Send the constructed similar area information to the path similarity index calculation module and the shortest distance identification module; the path similarity index calculation module is used to calculate the path similarity area index for the target TC and all historical TCs one by one in the similar area, and according to the TSAI The size of the historical TC is arranged from low to high, and the ranking result is sent to the best similar historical TC determination module; the shortest distance identification module is used to identify the daily precipitation forecast starting point of each historical TC from the target TC in a similar area The nearest observation point and the corresponding shortest distance, and the historical TC whose shortest distance is greater than a certain set value is eliminated; The TC movement speed similarity judgment module receives the movement speed information of the target TC and the historical TC generated by the TC daily movement speed acquisition module; then calculates the difference between the movement speed of all historical TCs and the target TC one by one, and marks the absolute value of the difference is greater than a certain value Threshold historical TC, send the marked historical TC number to the best similar historical TC determination module; The best similar historical TC determination module first receives the TSAI value sorting results of all historical TCs generated in the TC path similarity calculation module, receives the historical TC numbers, and removes all marked historical TCs; the remaining historical TCs are determined as the best similar history TC, and send the TC number to the TC ensemble forecast module. 5. the landing tropical cyclone daily precipitation forecasting system as claimed in claim 4, is characterized in that, the shortest distance identification module comprises the nearest point identification module and the shortest distance discrimination module; The nearest point identification module is in a certain selected similar area, successively. Calculate the distance between all observation points of all historical TCs and the starting point of the target TC daily precipitation forecast, send the calculated closest point information of all historical TCs to the TC daily precipitation acquisition module, and send the shortest distance information to the shortest distance discrimination module; The discrimination module marks the historical TC after receiving the shortest distance information, and sends the marked historical TC number to the best similar historical TC determination module. 6. The daily precipitation forecasting system for landing tropical cyclones as claimed in claim 4, wherein the TC precipitation ensemble forecasting module comprises a similar TC daily precipitation acquisition module and a precipitation ensemble module; the similar TC daily precipitation acquisition module is used to receive the best similar The number of the best similar historical TC generated in the historical TC determination module, calculate the accumulated precipitation 24 hours after the nearest point, and then use the objective synoptic map analysis method to identify the TC precipitation, so as to obtain the corresponding similar TC daily precipitation field , and send it to the precipitation aggregation module; the precipitation aggregation module is used to aggregate the TC daily precipitation fields generated in the similar TC daily precipitation acquisition modules into one precipitation field. It is the precipitation forecast result of the target TC within 24 hours after the starting point of the daily precipitation forecast. 7. landing tropical cyclone daily precipitation forecasting system as claimed in claim 2, is characterized in that, TC daily moving speed acquisition module receives the nearest point information of all historical TCs that nearest point identification module produces, calculates target TC on target TC day precipitation. The average moving speed within 24 hours after the forecast starting point is taken as the moving speed of the target TC, the average moving speed of all historical TCs within 24 hours after its closest point is calculated as their respective moving speeds, and the moving speed information of the target TC and all historical TCs is calculated. It is sent to the TC moving speed similarity discrimination module. 8. landing tropical cyclone daily precipitation forecasting system as claimed in claim 2, is characterized in that, TC moving speed similarity discrimination module receives the moving speed information of target TC and all historical TCs produced in TC daily moving speed acquisition module, calculates. The difference between the moving speed of all historical TCs and the target TC is marked, and the historical TCs whose differences meet certain conditions are marked, and the marked historical TC number is sent to the best similar historical TC determination module. 9. The daily precipitation forecasting system for landing tropical cyclones as claimed in claim 2, wherein the best similar historical TC determination module receives the TSAI value sorting results of all historical TCs generated in the TC path similarity calculation module, and receives the shortest distance All marked historical TC numbers in the discrimination module and TC speed similarity discrimination module, remove all marked historical TCs, determine the remaining top m historical TCs as the best similar historical TCs, and send the TC numbers to the TC Ensemble forecast module. 10. The daily precipitation forecasting system for landing tropical cyclones as claimed in claim 9, wherein the similar TC daily precipitation acquisition module receives the number of m best similar historical TCs generated in the best similar historical TC determination module, and calculates the respective After accumulating precipitation in the 24 hours after its closest point, the objective weather map analysis method is used to identify TC precipitation to obtain the corresponding m similar TC daily precipitation fields and send them to the TC precipitation ensemble forecasting module.

Images