CN111521990A - Rainfall analysis method based on multilayer radar echo data - Google Patents
Rainfall analysis method based on multilayer radar echo data Download PDFInfo
- Publication number
- CN111521990A CN111521990A CN202010389815.9A CN202010389815A CN111521990A CN 111521990 A CN111521990 A CN 111521990A CN 202010389815 A CN202010389815 A CN 202010389815A CN 111521990 A CN111521990 A CN 111521990A
- Authority
- CN
- China
- Prior art keywords
- data
- radar echo
- rainfall
- multilayer
- echo 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.)
- Granted
Links
- 238000004458 analytical method Methods 0.000 title claims abstract description 35
- 238000000034 method Methods 0.000 claims abstract description 34
- 238000012549 training Methods 0.000 claims abstract description 27
- 238000007781 pre-processing Methods 0.000 claims abstract description 9
- 238000002592 echocardiography Methods 0.000 claims abstract description 4
- 238000002310 reflectometry Methods 0.000 claims description 32
- 238000004364 calculation method Methods 0.000 claims description 12
- 238000000354 decomposition reaction Methods 0.000 claims description 9
- 238000006243 chemical reaction Methods 0.000 claims description 3
- 150000001875 compounds Chemical class 0.000 claims description 3
- 230000006870 function Effects 0.000 claims description 3
- 238000005070 sampling Methods 0.000 claims description 3
- 238000013473 artificial intelligence Methods 0.000 abstract description 2
- 239000010410 layer Substances 0.000 description 40
- 230000000875 corresponding effect Effects 0.000 description 16
- 239000011159 matrix material Substances 0.000 description 9
- 238000010586 diagram Methods 0.000 description 4
- 230000002596 correlated effect Effects 0.000 description 3
- 238000012545 processing Methods 0.000 description 3
- 229910002056 binary alloy Inorganic materials 0.000 description 2
- 238000013479 data entry Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 239000002356 single layer Substances 0.000 description 2
- 238000012271 agricultural production Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000013213 extrapolation Methods 0.000 description 1
- 238000001914 filtration Methods 0.000 description 1
- 230000002045 lasting effect Effects 0.000 description 1
- 230000002265 prevention Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 1
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
- G01S7/00—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00
- G01S7/02—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00 of systems according to group G01S13/00
- G01S7/41—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00 of systems according to group G01S13/00 using analysis of echo signal for target characterisation; Target signature; Target cross-section
-
- 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)
- Computer Networks & Wireless Communication (AREA)
- General Physics & Mathematics (AREA)
- Electromagnetism (AREA)
- Radar Systems Or Details Thereof (AREA)
Abstract
The invention belongs to the field of artificial intelligence, and particularly relates to a rainfall analysis method based on multilayer radar echo data. The method comprises the steps of obtaining historical data and historical rainfall data of original radar echoes; preprocessing original radar echo data, and analyzing radar echo reflection data at the same moment to obtain a multilayer radar echo data set; forming a training data set, and obtaining a rainfall analysis model based on multilayer radar echo data through training; and inputting the data into a rainfall analysis model to obtain an analysis result of the rainfall level. According to the invention, the original radar echo data is decomposed and preprocessed to obtain the multi-layer radar echo data values, so that the accuracy of the final result is improved, and the execution efficiency is improved while the accuracy is ensured. The invention can be used for analyzing rainfall in the meteorological field.
Description
Technical Field
The invention belongs to the field of artificial intelligence, and particularly relates to a rainfall analysis method based on multilayer radar echo data.
Background
The rainfall condition plays a vital role in daily life, agricultural production, water conservancy flood prevention and other aspects, and the analysis and prediction of the rainfall condition are one of the research hotspots in the meteorological field.
For the analysis and prediction of rainfall conditions, the existing mature methods are numerical prediction and methods based on radar echo extrapolation. These methods all achieve certain effects and can be applied to actual pre-use. However, these methods have problems such as inaccurate short-term prediction, excessive calculation amount, multi-step accumulated error, and the like.
Disclosure of Invention
The purpose of the invention is as follows:
the invention aims to solve the defects in the prior art and provides a method for preprocessing original radar echo data to obtain multilayer radar echo data, using the multilayer radar echo data for training a LightGBM algorithm model and using the multilayer radar echo data to finish rainfall analysis. Under the condition of ensuring the analysis accuracy, the operation efficiency is effectively improved.
The technical scheme is as follows:
a rainfall analysis method based on multilayer radar echo data comprises the following steps:
acquiring historical data and historical rainfall data of an original radar echo;
preprocessing the original radar echo data in the step (1), and analyzing radar echo reflection data at the same moment to obtain a multilayer radar echo data set;
step (3), correlating the rainfall value at the same time with the multilayer radar echo data reflectivity value data set established in the step (2), taking the analyzed multilayer radar echo data as an input attribute, and taking historical rainfall data as a corresponding input result to form a final training data set;
inputting the data set integrated in the step (3) into a LightGBM ensemble learning algorithm model for training to obtain a rainfall analysis model based on multilayer radar echo data;
and (5) inputting the preprocessed multilayer radar echo data of a certain spatial geographic position at a certain moment by using a rainfall analysis model established based on a LightGBM ensemble learning algorithm, and then obtaining an analysis result of the rainfall level.
Further, the original radar echo data in the step (2) are analyzed, so that a multilayer radar echo reflectivity value at the same geographic position at a certain moment is obtained.
Further, the preprocessing of the raw radar echo data refers to: performing further conversion and decomposition processing on the obtained original radar echo data, wherein the original binary coded radar echo data are transcoded into a decimal form, data on the same longitude and latitude coordinate are subjected to traversal decomposition according to a low-to-high acquisition mode of the radar echo data, and finally reflectivity values on 21 layers of radar echo reflecting layers on the corresponding coordinates are obtained; in the process of collecting the multi-layer radar echo data, reflection clutter can be generated when the height is too low or too high, so that echo reflection values of the highest and lowest 2 layers are deleted, and only the middle 17 layers are reserved as the multi-layer radar echo reflection values of an input model.
Further, in the step (3), historical rainfall data and a multi-layer radar echo data reflectivity value data set are correlated, so that a final training data set is formed.
Further, the historical rainfall data and the multilayer radar echo data reflectivity value data set are correlated, so that a final training data set is formed, specifically: and (3) correlating the decomposed multilayer radar echo reflectivity value data set with historical rainfall data recorded in the same time period, wherein the former is used as input, and the latter is used as a final data set formed by corresponding output results, and the process is as follows:
the rainfall recording is carried out by taking 10 minutes as a unit as an interval, radar echo data acquisition is carried out by taking 6 minutes as a unit, and the two types of data only have time points which are completely matched at three positions of 0 minute, 30 minutes and 60 minutes; in the process of integrating the two data sets, firstly traversing the data of the decomposed multilayer radar echo reflectivity data sets one by one, then matching the time of each multilayer radar echo reflectivity data with the data acquisition time in the rainfall data set one by one, and merging the two types of data with the same recording time, thereby forming a final training data set.
Further, in the step (5), the preprocessed multilayer radar echo data of a certain spatial geographical position at a certain time is input, and an analysis result of the corresponding rainfall level is obtained.
Further, a LightGBM algorithm model is used for calculating and analyzing the preprocessed multilayer radar echo reflection data to obtain a classification result of a corresponding rainfall level, and the method is characterized by comprising the following steps:
the LightGBM algorithm model uses a Gradient-based One-Side Sampling algorithm to sample a multi-layer radar echo data set, in the calculation process, multi-layer radar echo data samples with larger gradients greatly contribute to calculation information gain, so that data samples which do not contribute to calculation information gain are discarded and help is left, and the calculated data samples are used for learning a new weak classifier;
the formula is as follows:
wherein x isiTraining a dataset { x ] for multi-layer radar echoes1,…,xnVector in (v), giSet of gradient results for the loss function g1,…,gnThe gradient result generated by each iteration in the previous step; vJThe gain value produced at node d for feature j; a and B respectively rank the subsets of the computed radar echo data sets, Al＝{xi∈A:xij≤d}，Ar＝{xi∈A:xij>d}，Bl＝{xi∈B:xij≤d}，Br＝{xi∈B:xij>d }; n is the number of instances of the compound,
for features in the acquired radar echo data subset lj the number of instances on node d,
the number of instances of the feature j on the node d in the acquired radar echo data subset r;
the specific iteration steps of the algorithm are as follows:
s1, sorting in descending order according to the absolute value of the gradient of each multi-layer radar echo data sample;
and S2, selecting a% of samples in the sorted result set to generate a multilayer radar echo sample point subset A with a larger gradient, wherein the value range of a is 0 to 100.
And S3, randomly selecting B% of the residual (1-a)% multi-layer radar echo data sample sets to generate a sample point data set B with a smaller gradient, wherein the value range of B is 0 to 100.
S4, merging the two sample subsets A and B;
s5 multiplying the data subset B by the weight coefficient
S6, using the sampled samples to learn a new weak learner;
and S7, continuously repeating all the steps until a specified iteration number is reached or the gradient converges.
Further, a rainfall analysis model based on multi-layer radar echo data is established by using a LightGBM integrated learning model; in actual use, the multi-layer radar echo reflection value on a certain coordinate at a certain time is used as the input of the model, and then the rainfall classification level is output as the result after the model is analyzed.
The advantages and effects are as follows:
the invention has the beneficial effects that: the original radar echo data are decomposed and preprocessed to obtain multi-layer radar echo data values, and the accuracy of the final result is improved. In addition, historical rainfall data and multilayer radar echo data are correlated through time to form a training data set to train the LightGBM algorithm model, and finally rainfall can be analyzed through the algorithm model, so that the accuracy is guaranteed, and the execution efficiency is improved. The invention can be used for analyzing rainfall in the meteorological field.
Drawings
FIG. 1 is a process diagram for analyzing rainfall based on multi-layer radar echo data;
FIG. 2 is a schematic illustration of decoding raw binary radar data;
FIG. 3 is a schematic diagram of a radar echo file composition structure;
FIG. 4 is a decomposed numerical matrix of multi-layer radar returns;
FIG. 5 is a schematic diagram of a relationship between a multilayer radar echo image and an original single-layer radar echo image;
FIG. 6 is a schematic diagram illustrating operations for correlating multi-layer radar echo data with rainfall data.
Detailed Description
The invention provides a rainfall analysis method based on multilayer radar echo data, wherein rainfall is analyzed by combining a LightGBM integrated learning algorithm model with the multilayer radar echo data, so that the efficiency is improved, and the accuracy is ensured.
As shown in fig. 1, a rainfall analysis method based on multilayer radar echo data includes the following steps:
acquiring historical data and historical rainfall data of an original radar echo;
preprocessing the original radar echo data in the step (1), and analyzing radar echo reflection data at the same moment to obtain a multilayer radar echo data set;
step (3), correlating the rainfall value at the same time with the multilayer radar echo data reflectivity value data set established in the step (2), taking the analyzed multilayer radar echo data as an input attribute, and taking historical rainfall data as a corresponding input result to form a final training data set;
inputting the data set integrated in the step (3) into a LightGBM ensemble learning algorithm model for training to obtain a rainfall analysis model based on multilayer radar echo data;
and (5) inputting the preprocessed multilayer radar echo data of a certain spatial geographic position at a certain moment by using a rainfall analysis model established based on a LightGBM ensemble learning algorithm to obtain an analysis result of the rainfall level.
And (3) analyzing the original radar echo data in the step (2) so as to obtain a multilayer radar echo reflectivity value at the same geographic position at a certain moment.
As shown in fig. 2 and 3, the preprocessing performed on the raw radar echo data refers to: performing further conversion and decomposition processing on the obtained original radar echo data, wherein the original binary coded radar echo data are transcoded into a decimal form, data on the same longitude and latitude coordinate are subjected to traversal decomposition according to a low-to-high acquisition mode of the radar echo data, and finally reflectivity values on 21 layers of radar echo reflecting layers on the corresponding coordinates are obtained;
as shown in fig. 2, the original binary header file information in the file is first converted into readable header file information, and the binary image data in the file is converted into a visible radar image according to the key attribute data therein.
As shown in fig. 3, the radar echo file is mainly composed of two parts, i.e., header file data information and radar echo information.
In the process of collecting the multi-layer radar echo data, reflection clutter can be generated when the height is too low or too high, so that echo reflection values of the highest and lowest 2 layers are deleted, and only the middle 17 layers are reserved as the multi-layer radar echo reflection values of an input model.
And (3) correlating the historical rainfall data with the multilayer radar echo data reflectivity value data set so as to form a final training data set.
As shown in fig. 4, 5, and 6, the step of correlating the historical rainfall data with the multi-layer radar echo data reflectivity value data set to form a final training data set specifically includes: and (3) correlating the decomposed multilayer radar echo reflectivity value data set with historical rainfall data recorded in the same time period, wherein the former is used as input, and the latter is used as a final data set formed by corresponding output results, and the process is as follows:
the rainfall recording is carried out by taking 10 minutes as a unit as an interval, radar echo data acquisition is carried out by taking 6 minutes as a unit, and the two types of data only have time points which are completely matched at three positions of 0 minute, 30 minutes and 60 minutes; in the process of integrating the two data sets, firstly traversing the data of the decomposed multilayer radar echo reflectivity data sets one by one, then matching the time of each multilayer radar echo reflectivity data with the data acquisition time in the rainfall data set one by one, and merging the two types of data with the same recording time, thereby forming a final training data set.
And (5) inputting the multi-layer radar echo data of a certain spatial geographical position at a certain moment obtained after preprocessing to obtain a corresponding rainfall level analysis result.
And (3) performing calculation analysis on the preprocessed multilayer radar echo reflection data by using a LightGBM algorithm model to obtain a classification result of corresponding rainfall level, wherein the classification result is characterized in that:
the LightGBM algorithm model uses a Gradient-based One-Side Sampling algorithm to sample a multi-layer radar echo data set, in the calculation process, multi-layer radar echo data samples with larger gradients greatly contribute to calculation information gain, so that data samples which do not contribute to calculation information gain are discarded and help is left, and the calculated data samples are used for learning a new weak classifier;
the formula is as follows:
wherein x isiTraining a dataset { x ] for multi-layer radar echoes1,…,xnVector in (v), giSet of gradient results for the loss function g1,…,gnGenerated by each iterationThe result of the gradient of (c); vJThe gain value produced at node d for feature j; a and B respectively rank the subsets of the computed radar echo data sets, Al＝{xi∈A:xij≤d}，Ar＝{xi∈A:xij>d}，Bl＝{xi∈B:xij≤d}，Br＝{xi∈B:xij>d }; n is the number of instances of the compound,
for the number of instances of feature j on node d in the acquired subset of radar echo data i,
is the number of instances of feature j on node d in the acquired subset r of radar echo data.
The specific iteration steps of the algorithm are as follows:
s1, sorting in descending order according to the absolute value of the gradient of each multi-layer radar echo data sample;
s2, selecting a% of samples in the sorted result set to generate a multilayer radar echo sample point subset A with a larger gradient;
and S3, randomly selecting B% of the data in the residual (1-a)% multilayer radar echo data sample set to generate a sample point data set B with a smaller gradient, wherein the value range of a and B is 0 to 100.
S4, merging the two sample subsets A and B;
s5 multiplying the data subset B by the weight coefficient
S6, using the sampled samples to learn a new weak learner;
and S7, continuously repeating all the steps until a specified iteration number is reached or the gradient converges.
Establishing a rainfall analysis model based on multi-layer radar echo data by using a LightGBM ensemble learning model; in actual use, the multi-layer radar echo reflection value on a certain coordinate at a certain moment is used as the input of the model, and then the model outputs the corresponding rainfall classification level as a result after analysis.
As shown in fig. 5, each of the original single-layer radar echo images is stacked from a plurality of radar echo images acquired at different heights.
As shown in fig. 6, the original radar file is analyzed, then the file generation time in the file name is used to match the acquired historical rainfall data acquisition time, and the multi-layer radar echo reflection data and rainfall data which are consistent in time are integrated into a training data set.
Example 1
As shown in fig. 1, the rainfall analysis method based on multi-layer radar echo data includes the following steps:
step 1: the method comprises the steps of obtaining an original radar echo data file, coding the file in a binary system mode, and transcoding the binary system radar echo file to obtain two parts of data, namely head information data and metadata, contained in the original radar file in a decomposition mode.
Step 2: acquiring more key longitude and latitude attribute data and radar coordinate attribute data in the head information data, analyzing corresponding metadata according to the attribute data, and then obtaining a total of 21 multilayer radar echo reflectivity value matrixes contained in each original radar echo file, wherein the values on each coordinate point in the matrixes correspond to longitude and latitude coordinates on an actual geographic position one by one. The multi-layer radar echo reflectivity numerical matrix obtained after decomposition represents the reflectivity situation acquired by the meteorological radar at different heights above the actual geographic area.
As shown in fig. 4, each piece of data includes longitude and latitude coordinates and decomposed radar echo reflectivity values of each layer.
And step 3: and (3) filtering the multilayer radar echo reflectivity numerical matrix obtained by decomposing each radar file in the step (2), removing the multilayer radar echo reflection numerical matrix of each 2 layers obtained after decomposition, only keeping the radar echo reflection numerical matrix of the middle 17 layers, carrying out standardized processing on each filtered multilayer radar echo numerical matrix, and storing the multilayer radar echo reflection numerical matrix obtained by decomposing each radar echo file in an independent folder in a classified manner.
And 4, step 4: and separating the rainfall data from the meteorological database to obtain a rainfall acquisition time, a longitude and latitude coordinate of an acquisition site and a rainfall which are series of attribute data so as to form a rainfall data set. According to different rainfall sizes, the rainfall data set is graded to obtain 4 types of data, and each rainfall data is added with a grade label expressed by numbers.
And 5: and (3) reading the multilayer radar echo reflection value matrix folders obtained in the step (3) in sequence, firstly obtaining the name of each folder, and obtaining a timestamp corresponding to the acquisition time of the radar data. And matching the timestamp with the historical rainfall data obtained in the step 4 one by one, finding out rainfall data entries with the same acquisition time, reading and recording corresponding reflectivity values at corresponding positions of each multi-layer radar echo reflection value matrix obtained through decomposition after acquiring the longitude and latitude data, and obtaining a complete multi-layer radar echo reflectivity value-rainfall intensity data entry. And sequentially executing the operations on each folder to obtain a training data set required by algorithm training, wherein the data set comprises 18 rows of attributes, the first 17 rows are radar echo reflectivity values acquired by the weather radar at different heights above an area at a certain rainfall moment, and the 18 th row is rainfall intensity.
Step 6: and (5) inputting the training data set obtained in the step (5) into the LightGBM algorithm model for training, continuously adjusting various parameters of the model, enabling the accuracy of the final classification result of the algorithm model to be the best, and storing the model in a lasting mode.
And 7: and inputting the decomposed radar echo data set into a previously trained LightGBM algorithm model aiming at radar echo data needing to be researched and analyzed, and finally obtaining the corresponding rainfall classification level.
Claims (8)
1. A rainfall analysis method based on multilayer radar echo data is characterized in that: the method comprises the following steps:
acquiring historical data and historical rainfall data of an original radar echo;
preprocessing the original radar echo data in the step (1), and analyzing radar echo reflection data at the same moment to obtain a multilayer radar echo data set;
step (3), correlating the rainfall value at the same time with the multilayer radar echo data reflectivity value data set established in the step (2), taking the analyzed multilayer radar echo data as an input attribute, and taking historical rainfall data as a corresponding input result to form a final training data set;
inputting the data set integrated in the step (3) into a LightGBM ensemble learning algorithm model for training to obtain a rainfall analysis model based on multilayer radar echo data;
and (5) inputting the preprocessed multilayer radar echo data of a certain spatial geographic position at a certain moment by using a rainfall analysis model established based on a LightGBM ensemble learning algorithm, and then obtaining an analysis result of the rainfall level.
2. The method of claim 1, wherein the method comprises: and (3) analyzing the original radar echo data in the step (2) so as to obtain a multilayer radar echo reflectivity value at the same geographic position at a certain moment.
3. The rainfall analysis method based on multilayer radar echo data according to claim 1 or 2, wherein: the preprocessing of the raw radar echo data refers to: the resulting raw radar echo data is further processed by conversion and decomposition, wherein,
transcoding the radar echo data of the original binary coding into a decimal form, traversing and decomposing the data on the same longitude and latitude coordinate according to a low-to-high acquisition mode of the radar echo data, and finally acquiring the reflectivity values on 21 layers of radar echo reflecting layers on the corresponding coordinates;
in the process of collecting the multi-layer radar echo data, reflection clutter can be generated when the height is too low or too high, so that echo reflection values of the highest and lowest 2 layers are deleted, and only the middle 17 layers are reserved as the multi-layer radar echo reflection values of an input model.
4. The method of claim 1, wherein the method comprises: and (3) correlating the historical rainfall data with the multilayer radar echo data reflectivity value data set so as to form a final training data set.
5. The method of claim 4, wherein the method comprises: correlating historical rainfall data with a multilayer radar echo data reflectivity value data set, so as to form a final training data set, specifically: and (3) correlating the decomposed multilayer radar echo reflectivity value data set with historical rainfall data recorded in the same time period, wherein the former is used as input, and the latter is used as a final data set formed by corresponding output results, and the process is as follows:
the rainfall recording is carried out by taking 10 minutes as a unit as an interval, radar echo data acquisition is carried out by taking 6 minutes as a unit, and the two types of data only have time points which are completely matched at three positions of 0 minute, 30 minutes and 60 minutes; in the process of integrating the two data sets, firstly traversing the data of the decomposed multilayer radar echo reflectivity data sets one by one, then matching the time of each multilayer radar echo reflectivity data with the data acquisition time in the rainfall data set one by one, and merging the two types of data with the same recording time, thereby forming a final training data set.
6. The method of claim 1, wherein the method comprises: and (5) inputting the preprocessed multilayer radar echo data of a certain spatial geographic position at a certain moment to obtain a corresponding rainfall level analysis result.
7. The method of claim 6, wherein the method comprises: and (3) performing calculation analysis on the preprocessed multilayer radar echo reflection data by using a LightGBM algorithm model to obtain a classification result of corresponding rainfall level, wherein the classification result is characterized in that:
the LightGBM algorithm model uses a Gradient-based One-Side Sampling algorithm to sample a multi-layer radar echo data set, in the calculation process, multi-layer radar echo data samples with larger gradients greatly contribute to calculation information gain, so that data samples which do not contribute to calculation information gain are discarded and help is left, and the calculated data samples are used for learning a new weak classifier;
the formula is as follows:
wherein x isiTraining a dataset { x ] for multi-layer radar echoes1,…,xnVector in (v), giSet of gradient results for the loss function g1,…,gnThe gradient result generated by each iteration in the previous step; vJThe gain value produced at node d for feature j; a and B respectively rank the subsets of the computed radar echo data sets, Al＝{xi∈A:xij≤d}，Ar＝{xi∈A:xij>d}，Bl＝{xi∈B:xij≤d}，Br＝{xi∈B:xij>d }; n is the number of instances of the compound,
for the number of instances of the feature j on the node d in the acquired radar echo data subset l，
The number of instances of the feature j on the node d in the acquired radar echo data subset r;
the specific iteration steps of the algorithm are as follows:
s1, sorting in descending order according to the absolute value of the gradient of each multi-layer radar echo data sample;
s2, selecting a% of samples before the sorted result set to generate a multilayer radar echo sample point subset A with a larger gradient, wherein the value range of a is 0 to 100;
s3, randomly selecting B% of the rest (1-a)% of multi-layer radar echo data sample sets to generate a sample point data set B with a smaller gradient, wherein the value range of B is 0 to 100;
s4, merging the two sample subsets A and B;
s5 multiplying the data subset B by the weight coefficient
S6, using the sampled samples to learn a new weak learner;
and S7, continuously repeating all the steps until a specified iteration number is reached or the gradient converges.
8. The method of claim 7, wherein the method comprises: establishing a rainfall analysis model based on multi-layer radar echo data by using a LightGBM ensemble learning model; in actual use, the multi-layer radar echo reflection value on a certain coordinate at a certain time is used as the input of the model, and then the rainfall classification level is output as the result after the model is analyzed.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010389815.9A CN111521990B (en) | 2020-05-11 | 2020-05-11 | Rainfall analysis method based on multilayer radar echo data |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010389815.9A CN111521990B (en) | 2020-05-11 | 2020-05-11 | Rainfall analysis method based on multilayer radar echo data |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111521990A true CN111521990A (en) | 2020-08-11 |
| CN111521990B CN111521990B (en) | 2022-04-29 |
Family
ID=71908758
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010389815.9A Expired - Fee Related CN111521990B (en) | 2020-05-11 | 2020-05-11 | Rainfall analysis method based on multilayer radar echo data |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111521990B (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112180375A (en) * | 2020-09-14 | 2021-01-05 | 成都信息工程大学 | Meteorological radar echo extrapolation method based on improved TrajGRU network |
| CN113253275A (en) * | 2021-04-22 | 2021-08-13 | 南京航空航天大学 | Rainfall estimation method based on improved RBF neural network |
| CN113296074A (en) * | 2021-07-28 | 2021-08-24 | 成都远望探测技术有限公司 | Optical flow extrapolation method based on multi-layer CAPPI of meteorological radar |
| CN113642741A (en) * | 2021-08-12 | 2021-11-12 | 北京百度网讯科技有限公司 | Model, data generation method, data generation device, electronic device and storage medium |
| CN115792847A (en) * | 2022-11-08 | 2023-03-14 | 江西师范大学 | Quantitative precipitation estimation method based on neural network and echo vertical information |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| AU2062399A (en) * | 1998-01-23 | 1999-08-09 | Novimet | Technique for estimating rainfall from a meteorological radar with polarisation diversity |
| CN104316930A (en) * | 2014-10-29 | 2015-01-28 | 安徽四创电子股份有限公司 | Raininess estimation method based on dual polarization Doppler weather radar detection |
| CN109799550A (en) * | 2019-03-20 | 2019-05-24 | 北京百度网讯科技有限公司 | Method and apparatus for predicting rainfall intensity |
| CN110738252A (en) * | 2019-10-14 | 2020-01-31 | 广州地理研究所 | Space autocorrelation machine learning satellite precipitation data downscaling method and system |
| CN110888186A (en) * | 2019-11-29 | 2020-03-17 | 天津大学 | Method for forecasting hail and short-time heavy rainfall based on GBDT + LR model |
-
2020
- 2020-05-11 CN CN202010389815.9A patent/CN111521990B/en not_active Expired - Fee Related
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| AU2062399A (en) * | 1998-01-23 | 1999-08-09 | Novimet | Technique for estimating rainfall from a meteorological radar with polarisation diversity |
| CN104316930A (en) * | 2014-10-29 | 2015-01-28 | 安徽四创电子股份有限公司 | Raininess estimation method based on dual polarization Doppler weather radar detection |
| CN109799550A (en) * | 2019-03-20 | 2019-05-24 | 北京百度网讯科技有限公司 | Method and apparatus for predicting rainfall intensity |
| CN110738252A (en) * | 2019-10-14 | 2020-01-31 | 广州地理研究所 | Space autocorrelation machine learning satellite precipitation data downscaling method and system |
| CN110888186A (en) * | 2019-11-29 | 2020-03-17 | 天津大学 | Method for forecasting hail and short-time heavy rainfall based on GBDT + LR model |
Non-Patent Citations (4)
| Title |
|---|
| CHUANG ZHANG ET AL.: "Weather Visibility Prediction Based on Multimodal Fusion", 《IEEE ACCESS》 * |
| LI DIAO ET AL.: "Short-term Weather Forecast Based on Wavelet Denoising and Catboost", 《2019 CHINESE CONTROL CONFERENCE (CCC)》 * |
| YOUNG-MI LEE ET AL.: "The Development of a Rainfall Correction Technique based on Machine Learning for Hydrological Applications", 《JOURNAL OF ENVIRONMENTAL SCIENCE INTERNATIONAL》 * |
| 郭尚瓒 等: "基于神经网络和模型集成的短时降雨预测方法", 《气象科技进展》 * |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112180375A (en) * | 2020-09-14 | 2021-01-05 | 成都信息工程大学 | Meteorological radar echo extrapolation method based on improved TrajGRU network |
| CN112180375B (en) * | 2020-09-14 | 2022-12-20 | 成都信息工程大学 | Weather radar echo extrapolation method based on improved TrajGRU network |
| CN113253275A (en) * | 2021-04-22 | 2021-08-13 | 南京航空航天大学 | Rainfall estimation method based on improved RBF neural network |
| CN113296074A (en) * | 2021-07-28 | 2021-08-24 | 成都远望探测技术有限公司 | Optical flow extrapolation method based on multi-layer CAPPI of meteorological radar |
| CN113296074B (en) * | 2021-07-28 | 2022-02-22 | 成都远望探测技术有限公司 | Optical flow extrapolation method based on multi-layer CAPPI of meteorological radar |
| CN113642741A (en) * | 2021-08-12 | 2021-11-12 | 北京百度网讯科技有限公司 | Model, data generation method, data generation device, electronic device and storage medium |
| CN113642741B (en) * | 2021-08-12 | 2023-08-15 | 北京百度网讯科技有限公司 | Model, data generation method and device, electronic equipment and storage medium |
| CN115792847A (en) * | 2022-11-08 | 2023-03-14 | 江西师范大学 | Quantitative precipitation estimation method based on neural network and echo vertical information |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111521990B (en) | 2022-04-29 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111521990B (en) | Rainfall analysis method based on multilayer radar echo data | |
| CN107563381B (en) | Multi-feature fusion target detection method based on full convolution network | |
| CN112991354B (en) | High-resolution remote sensing image semantic segmentation method based on deep learning | |
| CN115934990B (en) | Remote sensing image recommendation method based on content understanding | |
| CN113052211A (en) | Pruning method based on characteristic rank and channel importance | |
| CN102054274A (en) | Method for full automatic extraction of water remote sensing information in coastal zone | |
| CN110222767B (en) | Three-dimensional point cloud classification method based on nested neural network and grid map | |
| CN113033714B (en) | Object-oriented full-automatic machine learning method and system for multi-mode multi-granularity remote sensing image | |
| CN113312993B (en) | Remote sensing data land cover classification method based on PSPNet | |
| CN115984850A (en) | Lightweight remote sensing image semantic segmentation method based on improved Deeplabv3+ | |
| CN112070078A (en) | Deep learning-based land utilization classification method and system | |
| CN110555841A (en) | SAR image change detection method based on self-attention image fusion and DEC | |
| CN116883853A (en) | Crop space-time information remote sensing classification method based on transfer learning | |
| CN114841412A (en) | Method for predicting pH value of sea cucumber growing water | |
| CN106611016A (en) | Image retrieval method based on decomposable word pack model | |
| CN114936694A (en) | Photovoltaic power prediction method based on double integration models | |
| CN117079130B (en) | Intelligent information management method and system based on mangrove habitat | |
| Sari et al. | Daily rainfall prediction using one dimensional convolutional neural networks | |
| CN114511787A (en) | Neural network-based remote sensing image ground feature information generation method and system | |
| CN116756572A (en) | Construction method based on mangrove ecological system distribution data set | |
| Li et al. | Image recognition and empirical application of desert plant species based on convolutional neural network | |
| Barnes | Image-driven data mining for image content segmentation, classification, and attribution | |
| CN114663791A (en) | Branch recognition method for pruning robot in unstructured environment | |
| CN111401519B (en) | Deep neural network unsupervised learning method based on similarity distance in object and between objects | |
| Ouchra et al. | Comparing Unsupervised Land Use Classification of Landsat 8 OLI Data Using K-means and LVQ Algorithms in Google Earth Engine: A Case Study of Casablanca. |
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 | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20220429 |
|
| CF01 | Termination of patent right due to non-payment of annual fee |