CN110031847B - Dynamic short-term quantitative rainfall estimation method combining wavelet transformation and support vector machine with radar reflectivity - Google Patents
Dynamic short-term quantitative rainfall estimation method combining wavelet transformation and support vector machine with radar reflectivity Download PDFInfo
- Publication number
- CN110031847B CN110031847B CN201811200753.1A CN201811200753A CN110031847B CN 110031847 B CN110031847 B CN 110031847B CN 201811200753 A CN201811200753 A CN 201811200753A CN 110031847 B CN110031847 B CN 110031847B
- Authority
- CN
- China
- Prior art keywords
- wavelet
- rainfall
- support vector
- radar
- data
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
- G01S13/00—Systems using the reflection or reradiation of radio waves, e.g. radar systems; Analogous systems using reflection or reradiation of waves whose nature or wavelength is irrelevant or unspecified
- G01S13/88—Radar or analogous systems specially adapted for specific applications
- G01S13/95—Radar or analogous systems specially adapted for specific applications for meteorological use
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
- Y02A90/00—Technologies having an indirect contribution to adaptation to climate change
- Y02A90/10—Information and communication technologies [ICT] supporting adaptation to climate change, e.g. for weather forecasting or climate simulation
Landscapes
- Engineering & Computer Science (AREA)
- Radar, Positioning & Navigation (AREA)
- Remote Sensing (AREA)
- Physics & Mathematics (AREA)
- Electromagnetism (AREA)
- Computer Networks & Wireless Communication (AREA)
- General Physics & Mathematics (AREA)
- Radar Systems Or Details Thereof (AREA)
Abstract
The invention discloses a dynamic quantitative precipitation estimation method combining wavelet transformation and a support vector machine with radar reflectivity, and belongs to the field of atmospheric science. Firstly, a matching library of radar echo and rainfall of an automatic station is established, a historical one-hour sample is obtained according to estimation time, wavelet transformation is carried out, a support vector machine model is established and trained in a wavelet domain, a dynamic rainfall estimation model in the wavelet domain is obtained, prediction is carried out in the wavelet domain after radar measures reflectivity factors of the rainfall echo, wavelet coefficients of the rainfall to be estimated can be obtained, and the estimated rainfall is 6 minutes after wavelet inverse transformation is carried out on the wavelet coefficients. The method combines the data description of wavelet transformation and the advantages of a support vector machine in small sample learning to carry out short-time dynamic quantitative precipitation estimation, and ensures the accuracy and the effectiveness of an estimation result. Experiments show that the estimation result of the method is superior to the effect of the rainfall estimation by using the Z-I relation no matter the root mean square error is calculated or the TS score is carried out.
Description
Technical Field
The invention belongs to the field of atmospheric science, and particularly relates to a dynamic radar reflectivity quantitative precipitation estimation method based on wavelet transformation and a support vector machine.
Background
The radar quantitative precipitation estimation is an important reference for short-time approaching early warning, and currently, radar precipitation estimation methods are mainly divided into two types, wherein one type is to determine precipitation intensity by using a Z-I relation and measure precipitation distribution, and the other type is to estimate precipitation intensity by combining the measurement of a rain gauge on points and a radar.
The basic principle of determining the precipitation intensity by using the Z-I relation is to convert the echo power into a radar reflectivity Z value based on the positive correlation relation between the reflectivity factor and the precipitation rate, establish an empirical relation between the reflectivity factor and the precipitation amount, estimate the precipitation amount according to a corresponding empirical formula after the radar measures the reflectivity factor of the precipitation echo, and accumulate the precipitation amount in a period of time. The radar rainfall measuring method has the advantages that when radar measurement errors and rainfall data have errors, the radar measurement errors and the rainfall data cannot be ordered in time, and the radar measurement errors and the rainfall data are influenced by various factors such as radar, rainfall types, radar detection heights, super refractive indexes, differences of ground rainfall, wind and the like, so that differences exist between radar estimated values and ground rainfall meter measured values, and uncertainty and randomness between radar reflectivity factors and rainfall are enhanced.
The rainfall meter and radar combined rainfall estimation utilizes rainfall meter station data and combines radar reflectivity factors to measure rainfall so as to improve the precision of radar rainfall measurement. The relatives have proposed many technical research results, such as variational method, kalman filtering method, probabilistic pairing method, terrain-based weighted random forest method, etc. However, the effect of these methods is not obvious in the application of actual weather service, and most commonly, the Z-I relationship is classified, the estimation result of the Z-I relationship is easily affected by other factors, and the estimation error is usually large.
Disclosure of Invention
In order to solve the technical problems, the invention provides a dynamic short-term quantitative rainfall estimation method combining wavelet transformation and a correlation vector machine with radar reflectivity, and aims to solve the problems of large rainfall prediction error, limited application range and the like in the conventional radar quantitative rainfall estimation problem.
The invention aims to provide a dynamic short-term quantitative precipitation estimation method combined with radar reflectivity, which comprises the following steps:
and S1, acquiring 6-minute radar echo data of a certain area and a 5-minute automatic station rainfall database.
And S2, averaging the rainfall data of the automatic station for 5 minutes to obtain precipitation data per minute, and accumulating the precipitation data in the corresponding time period according to the time of radar echo data to obtain precipitation data of 6 minutes.
And S3, matching the radar echo data and the rainfall data of the automatic station according to time and longitude and latitude to obtain a sample library in which the radar echo and the reflectivity correspond to each other one by one.
S4, selecting data corresponding to historical radar echo and rainfall of the automatic station for one hour, respectively performing stationary wavelet transformation to respectively obtain wavelet coefficients of radar echo and rainfall under high frequency and low frequency and in multiple scales, wherein the scale is related to the sample size within one hour, and the sample size obtained within one hour is set as n, 2 n1 ≤n＜2 n1+1 The scale of the wavelet transform is n 1.
And S5, taking the wavelet coefficient obtained in S3 as a training sample, respectively training a corresponding model based on the support vector machine in each scale of wavelet transformation under high frequency and low frequency, and finally obtaining n1 support vector machine models under high frequency and n1 support vector machine models under low frequency.
S6: and obtaining radar precipitation echo data of 6 minutes at the current moment, and performing stationary wavelet transformation to obtain multi-scale radar echo wavelet coefficients under high frequency and low frequency.
S7: and taking the radar echo wavelet coefficient to be estimated obtained in the S6 as input, and inputting the coefficients of the same high-frequency/low-frequency channel and the same wavelet transformation scale into corresponding support vector machine models to obtain the wavelet coefficients of the rainfall to be estimated for 6 minutes at different scales under high frequency and low frequency.
S8: and performing wavelet inverse transformation on the wavelet coefficient of the rainfall to be estimated obtained by training in the S7 to obtain the value of the rainfall estimated in the method for 6 minutes.
Compared with the prior art, the invention has the following advantages:
(1) according to the dynamic short-term quantitative rainfall estimation method combining wavelet transformation and a support vector machine with radar reflectivity, the short-term dynamic quantitative rainfall estimation is carried out by combining the data description and the denoising capability of the wavelet transformation and the advantages of the support vector machine in small sample learning through the steps, and the accuracy and the effectiveness of an estimation result are guaranteed. Experiments show that the estimation result of the method has a much better effect compared with the quantitative precipitation estimation by a Z-I relational expression no matter the root mean square error is calculated or the TS scoring is carried out.
(2) In addition, the training sample of the estimation method only adopts wavelet transformation data of the past hour, the sample size is small, the transplantation is convenient, the algorithm complexity of the support vector machine for modeling is low, the training time is short, the time of once estimation is not more than 20 seconds, and the estimation method is more suitable for the application of actual weather service.
Drawings
FIG. 1 is a flow chart of the dynamic short-term quantitative precipitation estimation method of the present invention using wavelet transform and support vector machine in combination with radar reflectivity.
Detailed Description
The present invention will be further described in the following with reference to the drawings and example 1.
Example 1
The verification set of the invention is that 0.5-degree elevation radar echo data in Doppler weather radar 9 layer body scanning in Shanghai city Yangpu region and all precipitation automatic station data in Shanghai city during 2017 years and 6 months to 8 months, and 00-hour data at 6 months, 10 days and 12 days in 2017 years are selected as specific verification data for quantitative precipitation estimation of the dynamic radar based on wavelet transformation and a support vector machine, wherein the verification process comprises the following steps:
s1, acquiring 6-minute radar echo with 0.5-degree elevation angle in Doppler weather radar 9 layer body scanning in Shanghai city Yangpu district and automatic station rainfall data with 5-minute interval in Shanghai city as a database.
And S2, averaging the rainfall data of the automatic station for 5 minutes to obtain precipitation data per minute, and accumulating the precipitation data in the corresponding time period according to the time of the radar echo data to obtain precipitation data for 6 minutes.
And S3, matching the radar echo data and the rainfall data of the automatic station according to the time and the longitude and latitude to obtain a sample library in which the radar echo and the reflectivity correspond to each other one by one.
S4, selecting data corresponding to radar echo and automatic station rainfall between 11 hours and 00 hours in 6 months and 10 days in 2017 and between 12 hours and 00 hours in 10 months and 12 days in 6 months and 10 days in 2017 (the data do not include the data 00 hours in 12 hours in 6 months and 10 days in 2017), wherein the data are 2 in magnitude 8 And performing the stationary wavelet transform with the scale of 8 to respectively obtain the wavelet coefficients of radar echo and rainfall under high frequency and low frequency and 8 scales.
And S5, taking the wavelet coefficient obtained in S3 as a training sample, respectively training corresponding models based on the support vector machine in 8 scales of wavelet transformation under high frequency and low frequency, and finally obtaining 8 support vector machine models under high frequency and 8 support vector machine models under low frequency, wherein the total number of the support vector machine models is 16.
And S6, obtaining radar precipitation echo data of 6 minutes of 12 hours in 6 months, 10 days and 12 days in 2017, and similarly performing 8-layer stationary wavelet transform to obtain radar echo wavelet coefficients under high frequency and low frequency and 8 scales.
S7: and taking the radar echo wavelet coefficient to be estimated obtained in the S6 as input, and inputting the coefficients of the same high-frequency/low-frequency channel and the same wavelet transformation scale into corresponding support vector machine models to obtain the wavelet coefficients of the rainfall to be estimated for 6 minutes under 8 scales under high frequency and low frequency.
S8: and performing wavelet inverse transformation on the wavelet coefficient of the rainfall to be estimated, which is obtained by training in the S7, to obtain a value of the rainfall of 6 minutes between 00 minutes at 12 days 6 and 10 in 2017 and 06 minutes at 12 days 6 and 10 in 2017.
Claims (5)
1. The dynamic short-term quantitative rainfall estimation method combining wavelet transformation and a support vector machine with radar reflectivity is characterized in that: the method comprises the following steps:
s1, acquiring 6 minutes of radar echo data of a certain area and a 5 minutes of automatic station rainfall database;
s2, averaging the rainfall data of the automatic station for 5 minutes to obtain precipitation data per minute, and accumulating the precipitation data in a corresponding time period according to the time of radar echo data to obtain precipitation data of 6 minutes;
s3, matching the radar echo data and the rainfall data of the automatic station according to time and longitude and latitude to obtain a sample library in which the radar echo and the reflectivity correspond to each other one by one;
s4, selecting data corresponding to historical radar echo and rainfall of the automatic station for one hour, respectively performing stationary wavelet transformation to respectively obtain wavelet coefficients of radar echo and rainfall under high frequency and low frequency and in multiple scales, wherein the scale is related to the sample size within one hour, and the sample size obtained within one hour is set as n, 2 n1 ≤n＜2 n1+1 Then the scale of the wavelet transform is n 1;
s5, taking the wavelet coefficient obtained in S4 as a training sample, respectively training a corresponding model based on a support vector machine in each scale of wavelet transformation under high frequency and low frequency, and finally obtaining n1 support vector machine models under high frequency and n1 support vector machine models under low frequency;
s6: obtaining radar precipitation echo data of 6 minutes at the current moment, and performing stationary wavelet transformation to obtain multi-scale radar echo wavelet coefficients under high frequency and low frequency;
s7: taking the radar echo wavelet coefficient to be estimated obtained in the S6 as input, and inputting the coefficients of the same high-frequency/low-frequency channel and the same wavelet transformation scale into corresponding support vector machine models to obtain the wavelet coefficients of the rainfall to be estimated for 6 minutes at different scales under high frequency and low frequency;
s8: and performing wavelet inverse transformation on the wavelet coefficient of the rainfall to be estimated obtained by training in the S7 to obtain the value of the rainfall estimated in the method for 6 minutes.
2. The method for dynamic short-lived quantitative precipitation estimation of radar reflectivity using wavelet transform and support vector machines as claimed in claim 1, wherein: in step S4, one hour of history data is selected, and the data is subjected to stationary wavelet transform.
3. The method for dynamic short-lived quantitative precipitation estimation of radar reflectivity using wavelet transform and support vector machines as claimed in claim 2, wherein: in step S4, the scale of the stationary wavelet transform is related to the amount of samples in one hour.
4. The method for dynamic short-lived quantitative precipitation estimation of radar reflectivity using wavelet transform and support vector machines as claimed in claim 1, wherein: in step S5, a support vector machine model is constructed and trained in different frequency channels and different wavelet scales in the wavelet domain, and is estimated.
5. The method for dynamic short-lived quantitative precipitation estimation of radar reflectivity using wavelet transform and support vector machines as claimed in claim 1, wherein: in step S6, the time for each precipitation estimation is 6 minutes.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811200753.1A CN110031847B (en) | 2018-09-29 | 2018-09-29 | Dynamic short-term quantitative rainfall estimation method combining wavelet transformation and support vector machine with radar reflectivity |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811200753.1A CN110031847B (en) | 2018-09-29 | 2018-09-29 | Dynamic short-term quantitative rainfall estimation method combining wavelet transformation and support vector machine with radar reflectivity |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110031847A CN110031847A (en) | 2019-07-19 |
| CN110031847B true CN110031847B (en) | 2022-08-23 |
Family
ID=67234975
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811200753.1A Active CN110031847B (en) | 2018-09-29 | 2018-09-29 | Dynamic short-term quantitative rainfall estimation method combining wavelet transformation and support vector machine with radar reflectivity |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110031847B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111650572B (en) * | 2020-07-23 | 2023-02-24 | 浪潮云信息技术股份公司 | Method and system for reducing short-time precipitation estimation deviation |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101630405A (en) * | 2009-08-14 | 2010-01-20 | 重庆市勘测院 | Multi-focusing image fusion method utilizing core Fisher classification and redundant wavelet transformation |
| CN102063715A (en) * | 2010-12-25 | 2011-05-18 | 浙江师范大学 | Method for fusing typhoon cloud pictures based on NSCT (Nonsubsampled Controurlet Transformation) and particle swarm optimization algorithm |
| CN105528619A (en) * | 2015-12-10 | 2016-04-27 | 河海大学 | SAR remote sensing image change detection method based on wavelet transform and SVM |
| CN107220628A (en) * | 2017-06-06 | 2017-09-29 | 北京环境特性研究所 | The method of infrared jamming source detection |
| CN107703493A (en) * | 2017-11-13 | 2018-02-16 | 浙江大学 | Sea clutter optimal soft survey instrument and method based on adaptive drosophila optimized algorithm Optimized Least Square Support Vector |
Family Cites Families (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7085436B2 (en) * | 2001-08-28 | 2006-08-01 | Visioprime | Image enhancement and data loss recovery using wavelet transforms |
| US20110077484A1 (en) * | 2009-09-30 | 2011-03-31 | Nellcor Puritan Bennett Ireland | Systems And Methods For Identifying Non-Corrupted Signal Segments For Use In Determining Physiological Parameters |
| US10270642B2 (en) * | 2012-12-05 | 2019-04-23 | Origin Wireless, Inc. | Method, apparatus, and system for object tracking and navigation |
-
2018
- 2018-09-29 CN CN201811200753.1A patent/CN110031847B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101630405A (en) * | 2009-08-14 | 2010-01-20 | 重庆市勘测院 | Multi-focusing image fusion method utilizing core Fisher classification and redundant wavelet transformation |
| CN102063715A (en) * | 2010-12-25 | 2011-05-18 | 浙江师范大学 | Method for fusing typhoon cloud pictures based on NSCT (Nonsubsampled Controurlet Transformation) and particle swarm optimization algorithm |
| CN105528619A (en) * | 2015-12-10 | 2016-04-27 | 河海大学 | SAR remote sensing image change detection method based on wavelet transform and SVM |
| CN107220628A (en) * | 2017-06-06 | 2017-09-29 | 北京环境特性研究所 | The method of infrared jamming source detection |
| CN107703493A (en) * | 2017-11-13 | 2018-02-16 | 浙江大学 | Sea clutter optimal soft survey instrument and method based on adaptive drosophila optimized algorithm Optimized Least Square Support Vector |
Non-Patent Citations (10)
| Title |
|---|
| Estimating soil erosion using MODIS and TM images based on support vector;S.H. Chen;《International Journal of Applied Earth Observation and Geoinformation 13 (2011)》;20111231;626-635 * |
| The quantitative precipitation estimation system for Dallas–Fort Worth (DFW) urban remote sensing network;Haonan Chen;《Journal of Hydrology》;20151231;1-13 * |
| 丹江口水库径流预测方法与趋势分析;杨正祥等;《水电能源科学》;20110225(第02期);7-9 * |
| 台风云图伪彩色增强的Berkeley小波变换法;金明星等;《计算机应用》;20100601(第06期);178-181 * |
| 基于WA-SVM的流域降雨序列预测研究及应用;张国栋等;《中国农村水利水电》;20070515(第05期);5-7、12 * |
| 基于小波变换和粗糙集的一种RCS处理方法;张善文等;《电光与控制》;20070228(第01期);7-10 * |
| 基于小波变换的自适应脉冲耦合神经网络图像融合;薛寺中等;《计算机应用》;20101231(第12期);3225-3228 * |
| 基于小波变换集成模型预测径流水位的研究;丁红等;《武汉理工大学学报》;20121230(第12期);149-155 * |
| 基于机器学习的PM2.5短期浓度动态预报模型;戴李杰等;《计算机应用》;20171110(第11期);25-31 * |
| 改进时间序列模型在降雨量预测中的应用研究;白玉洁;《计算机仿真》;20111015(第10期);149-153 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110031847A (en) | 2019-07-19 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Goudenhoofdt et al. | Evaluation of radar-gauge merging methods for quantitative precipitation estimates | |
| Wang et al. | Trends and low frequency variability of extra-tropical cyclone activity in the ensemble of twentieth century reanalysis | |
| CN111401602B (en) | Assimilation method for satellite and ground rainfall measurement values based on neural network | |
| Wang et al. | Adjusted daily rainfall and snowfall data for Canada | |
| Brocca et al. | Application of a model-based rainfall-runoff database as efficient tool for flood risk management | |
| CN105891833A (en) | Method of identifying warm cloud precipitation rate based on Doppler radar information | |
| Benestad et al. | The use of a calculus-based cyclone identification method for generating storm statistics | |
| CN109685334A (en) | A kind of new hydrological model simulation evaluation method based on Multiscale Theory | |
| Magome et al. | River discharge simulation of a distributed hydrological model on global scale for the hazard quantification | |
| CN110031847B (en) | Dynamic short-term quantitative rainfall estimation method combining wavelet transformation and support vector machine with radar reflectivity | |
| CN113592186B (en) | Hydrological prediction state variable real-time correction method based on real-time measured diameter flow data | |
| CN110186533A (en) | A kind of short-term tide prediction method in high-precision river mouth | |
| Young | Detecting and removing inhomogeneities from long-term monthly sea level pressure time series | |
| CN103033274B (en) | Measuring method of solar temperature probability density | |
| CN117787081A (en) | Hydrological model parameter uncertainty analysis method based on Morris and Sobol methods | |
| Fischer et al. | Seasonal cycle in German daily precipitation extremes | |
| CN116796649A (en) | SPEI coarse resolution data space downscaling method and device based on machine learning | |
| Sun et al. | A subjective and objective evaluation of model forecasts of Sumatra squall events | |
| Keller et al. | CEH-GEAR: 1 km resolution daily and monthly areal rainfall estimates for the UK for hydrological use | |
| Cao et al. | Altimeter significant wave height data assimilation in the South China Sea using Ensemble Optimal Interpolation | |
| CN115203643A (en) | Hydrologic and ecological factor fused water source conservation function quantitative diagnosis method and system | |
| Verma et al. | A framework for the evaluation of MRP complex precipitation in a CORDEX-SA regional climate applied to REMO | |
| Kisel’nikova | Object-based evaluation of precipitation forecast quality | |
| Liu et al. | The June 2012 North American Derecho: A testbed for evaluating regional and global climate modeling systems at cloud-resolving scales | |
| CN114004426B (en) | Dynamic adjustment method of short-time rainstorm forecast release model |
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 |