CN110426755B - Ground temperature data quality control method based on improved reverse distance weighting - Google Patents
Ground temperature data quality control method based on improved reverse distance weighting Download PDFInfo
- Publication number
- CN110426755B CN110426755B CN201910686271.XA CN201910686271A CN110426755B CN 110426755 B CN110426755 B CN 110426755B CN 201910686271 A CN201910686271 A CN 201910686271A CN 110426755 B CN110426755 B CN 110426755B
- Authority
- CN
- China
- Prior art keywords
- value
- target station
- average temperature
- daily average
- rmse
- 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
- 238000003908 quality control method Methods 0.000 title claims abstract description 21
- 238000002474 experimental method Methods 0.000 claims abstract description 41
- 238000000034 method Methods 0.000 claims abstract description 38
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 16
- 239000002245 particle Substances 0.000 claims description 12
- 238000005457 optimization Methods 0.000 abstract description 6
- 238000010586 diagram Methods 0.000 description 4
- 238000011160 research Methods 0.000 description 4
- 230000000694 effects Effects 0.000 description 3
- 238000011156 evaluation Methods 0.000 description 3
- 230000002159 abnormal effect Effects 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000002093 peripheral effect Effects 0.000 description 1
- 238000001556 precipitation Methods 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 238000007637 random forest analysis Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01W—METEOROLOGY
- G01W1/00—Meteorology
- G01W1/10—Devices for predicting weather conditions
Landscapes
- Environmental & Geological Engineering (AREA)
- Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Atmospheric Sciences (AREA)
- Biodiversity & Conservation Biology (AREA)
- Ecology (AREA)
- Environmental Sciences (AREA)
- Complex Calculations (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
本发明提供了一种基于改进反距离加权的地面气温资料质量控制方法，步骤如下：S1采集目标站和其一定范围内的n个参考站的日均温实际观测数据；S2建立原反距离加权法模型进行实验，预估目标站的日均温；S3建立改进的反距离加权法模型进行实验预估：计算目标站的日均温预估值与日均温实际值的第一均方根误差RMSE；重复S1、S2，计算目标站的日均温预估值与日均温实际值的第二均方根误差RMSE；重复m次，得到m组预估值和对应的RMSE，m及n通过PSO寻优得到；S4根据RMSE的值大小赋予预估值结果的权重值，通过每次实验的预估值乘以对应权值，得出最终预估结果。本发明算法精确度、稳定性和适应性更好，偏差值较小。
The invention provides a method for quality control of surface air temperature data based on improved inverse distance weighting. The steps are as follows: S1 collects the actual observation data of daily average temperature of the target station and n reference stations within a certain range; S2 establishes the original inverse distance weighting method model to conduct experiments to estimate the daily average temperature of the target station; S3 establishes an improved inverse distance weighting method model for experimental prediction: calculate the first root mean square of the predicted value of the daily average temperature of the target station and the actual value of the daily average temperature Error RMSE; Repeat S1 and S2 to calculate the second root mean square error RMSE between the predicted value of the daily average temperature of the target station and the actual value of the daily average temperature; repeat m times to obtain m groups of predicted values and the corresponding RMSE, m and n is obtained through PSO optimization; S4 assigns the weight value of the estimated value result according to the value of RMSE, and the final estimated result is obtained by multiplying the estimated value of each experiment by the corresponding weight. The algorithm of the invention has better accuracy, stability and adaptability, and smaller deviation value.
Description
Technical Field
The invention belongs to the technical field of quality control of ground temperature data, and particularly relates to a ground temperature data quality control method based on improved reverse distance weighting.
Background
The meteorological data is the basis for meteorological research, and in order to guarantee that the meteorological research process truly reflects weather and change rules thereof, the quality control must be carried out on the meteorological data before the research, so that the authenticity and the usability of the meteorological data are ensured. In recent years, with intensive research on climate change in various countries, many quality control methods for meteorological data have appeared. The traditional quality control method mainly analyzes the rationality of meteorological data according to the principles of meteorology, meteorological principle and climatology, the time and space change rules of each element of meteorological data and the mutual relation among the elements as clues. The method mainly comprises the following steps: at present, the methods are widely applied to the quality control of ground meteorological data, and the comprehensive application effect of the methods is reflected to be good through actual application.
With the increase of the space density of the data of the automatic station, especially the use of the data of high-density precipitation, air temperature and wind, the data quality of single elements needs to be effectively controlled. If the data is single element data or newly-built measuring stations or measuring points, the quality control of the observed data can only be carried out by an interpolation estimation method due to the lack of historical data or the mutual comparison of the elements. Currently, there are many methods for interpolation estimation, such as: interpolation methods such as Inverse Distance Weighting (IDW) and Spatial Regression (SRT). The original IDW is mainly weighted by the distance between a target station and a reference station, and the closer the distance is, the larger the weight of the station is, so that the processing effect on complex terrain and abnormal weather is poor. The method is improved on the basis of the original IDW, so that the precision, stability and adaptability of the algorithm are obviously improved.
Disclosure of Invention
The invention aims to provide a ground temperature data quality control method based on improved inverse distance weighting aiming at the defects of the original IDW, and the accuracy, stability and adaptability of the algorithm are improved.
In order to achieve the purpose, the invention adopts the following technical scheme:
a ground temperature data quality control method based on improved inverse distance weighting comprises the following steps:
s1, acquiring the daily average temperature actual observation data of the target station and n reference stations within a certain range;
s2, establishing an original inverse distance weighting method model for experiment, and estimating the daily average temperature of the target station through actual observation data of the reference station to obtain a predicted value of the daily average temperature of the target station;
s3, establishing an improved inverse distance weighting method model for experimental prediction:
calculating a first Root Mean Square Error (RMSE) of a daily average temperature estimated value of the target station and a daily average temperature actual value of the target station;
repeating S1 and S2, randomly selecting n reference stations in the range of the target station, and calculating a second Root Mean Square Error (RMSE) of the estimated daily average temperature value of the target station and the actual daily average temperature value of the target station; repeating the steps m times to obtain m groups of estimated values and corresponding RMSE;
and S4, according to the value of RMSE, giving the weight value of the estimated value result obtained in each experiment in S3, and multiplying the estimated value obtained in each experiment by the corresponding weight value to obtain the final estimated result of the daily average temperature of the target station.
In order to optimize the technical scheme, the specific measures adopted further comprise:
the reference station is n stations randomly selected within 150 kilometers around the target station.
The formula of the original inverse distance weighting method model is as follows:
in the formula, Z(s)0) Is a target station s0A predicted value of (2); z (i) is the actual observation at reference station i; lambda [ alpha ]iThe weight value of the reference station i to the target station is taken as the weight value; n is the number of reference stations; di0The distance between the reference station i and the target station is taken as the distance; p defaults to 2.
The formula for calculating RSME is as follows:
in the formula, yrIs the actual observed value of day r, yr' is an estimated value for day r, r is the number of days observed, and t is the total number of days observed.
In step S4, the smaller the RMSE value, the larger the weight value of the corresponding estimated result.
The RMSE weighting method formula is as follows:
wherein X' is a final estimated result; w is aiThe weight coefficient of the ith experiment; ziThe estimated result of the ith experiment is obtained; a is the parameter value of the exponential function; r isiThe root mean square error of the i-th experiment.
The formula of the exponential function is that y is ax Wherein 0 is<a<1。
And finding out the optimal value of the experiment iteration times m and the optimal value of the random site number n by using a particle swarm algorithm.
The actual annual average temperature observation data is collected in S1.
The invention has the beneficial effects that: on the basis of an original reverse distance weighting method, the idea of random forests is introduced, a single reverse distance weighting method is used as a weak regression to carry out combined prediction, an exponential function weighting method is introduced, the estimation error RMSE of each reverse distance weighting method is used as an evaluation standard, the smaller the RMSE is, the larger the corresponding estimation result weight is, and the larger the RMSE is, the smaller the corresponding estimation result weight is. Experiments prove that the algorithm accuracy, stability and adaptability of the method are better than those of the original inverse distance weighting method and the space regression interpolation method, and the deviation value of the experimental result of the method is smaller.
Drawings
FIG. 1 is a flow diagram of the ID-IDW algorithm of the present invention.
Fig. 2 is a diagram of a distribution of four target stations and stations within 150 km of the surroundings according to an embodiment of the invention.
FIG. 3 is a diagram illustrating Particle Swarm Optimization (PSO) results according to an embodiment of the present invention.
FIG. 4 is a graph of α -RMSE relationships for an embodiment of the present invention.
FIG. 5 is a graph of the results of RMSE and MAE indicators for an embodiment of the present invention.
FIG. 6 is a diagram showing ID-IDW result fluctuation according to an embodiment of the present invention.
Detailed Description
The present invention will now be described in further detail with reference to the accompanying drawings.
The invention discloses a ground temperature data quality control method based on improved reverse distance weighting, as shown in figure 1, the basic flow of the invention is as follows:
s1, acquiring the daily average temperature actual observation data of the target station and n reference stations within a certain range.
In order to verify the generalization capability of the algorithm in space and time, four different sites of Jiangsu Nanjing, Sichuan Wenjiang, Hunan Xintian and Shanxi Taiyuan are selected as target sites, all sites within a range of 150 kilometers around the four sites are simultaneously selected as reference sites, and the four sites and the surrounding sites are distributed as shown in FIG. 2. All data with south Jing of Jiangsu as a target station are 2014-year-day average temperature data, data with Wenjiang of Sichuan as a target station are 2005-year-day average temperature data, data with New field of Hunan as a target station are 2007-year-day average temperature data, and data with Taiyuan of Shanxi as a target station are 2010-year-day average temperature data.
S2, an original Inverse Distance Weighting (IDW) model is established for experiment, the daily average temperature of the target station is estimated according to actual observation data of the reference station, and a predicted value of the daily average temperature of the target station is obtained.
The formula of the original IDW model is as follows:
in the formula, Z(s)0) Is a target station s0A predicted value of (2); z (i) is the actual observation at reference station i; lambda [ alpha ]iThe weight value of the reference station i to the target station is taken as the weight value; n is the number of reference stations; di0The distance between the reference station i and the target station is taken as the distance; p defaults to 2.
IDW is mainly performed to control the quality of a target station based on actual measurement data of peripheral stations, and the assignment of weight is performed by distance. If abnormal weather or wrong data occurs at a reference station close to the reference station, the accuracy of the IDW algorithm is reduced, which is a main disadvantage of IDW.
S3, establishing an improved inverse distance weighting method (ID-IDW) model for experimental prediction.
And S4, according to the value of RMSE, giving the weight value of the estimated value result obtained in each experiment in S3, and multiplying the estimated value obtained in each experiment by the corresponding weight value to obtain the final estimated result of the daily average temperature of the target station.
1) Randomly selecting n stations within 150 kilometers around a selected target station as a reference station of the 1 st experiment, estimating the daily average temperature of the target station by using an IDW algorithm, calculating the Root Mean Square Error (RMSE) of the estimated value and the actual value of the target station, and calculating the RSME according to the following formula:
wherein y isrIs the actual observed value of day r, yr' is the estimated value of the r day, r is the number of days observed, and the present embodiment is the estimated average temperature of the day of the whole year, so the total number of days is 365 days.
Randomly selecting n stations from the original data as reference stations to carry out the 2 nd experiment, and calculating the estimated value and RMSE of the target station; repeating the experiment m times to obtain m groups of estimated values and RMSE values, and giving a weight value to the result of each experiment according to the size of each group of RMSE values, wherein the smaller the RMSE, the larger the weight value is, the larger the corresponding RMSE is, and the smaller the weight value is. And the final target station day average temperature estimation result is obtained by multiplying each group of experimental results by the corresponding weight.
The exponential function weighting method comprises the following steps: by exponential function image y ═ ax(0<a<1) It can be seen that the smaller the value of the positive half axis of the x-axis, the larger the value of y. Where x denotes the root mean square error, i.e. the smaller the root mean square error, the greater the weight, the formula is as follows:
wherein X' is the final prediction result; w is aiThe weight coefficient of the ith experiment; ziIs the ith timeAn experiment estimation result; a is the parameter value of the exponential function; r isiThe root mean square error of the i-th experiment.
2) Determination and selection of parameters m and n
As shown in fig. 3, the optimal number of iterations m and the number of random sites n are found by Particle Swarm Optimization (PSO). The particle swarm optimization method comprises the following steps:
wherein the q-dimensional particle is represented by xi＝(xi1,xi2,xi3……xiq) Each particle corresponding to a velocity vi＝(vi1,vi2,vi3……viq) Each particle is searched for considering its own optimum piAnd particle swarm optimization pgW is the inertial weight, c1Tracking the optimal historical weight coefficient for the particle, c2For the optimal weight coefficient of the particle tracking population, delta and gamma are intervals of [0, 1 ]]And r is a constraint factor.
Taking the Guangdong Zijin station as an example, searching for the optimal solution 1000 times by using a particle swarm optimization algorithm, setting the m value to be 5 to 200, and setting the n value to be 5 to 20, wherein the root mean square error obtained by ID-IDW each time is taken as a fitness value to be calculated. FIG. 3 is a graph of the results obtained from 1000 seeks, with the final RMSE remaining substantially around 0.35, assuming a convergent regime, and resulting in final seek m and n values. According to the process, the invention respectively performs experiments on four sites of Nanjing, Wenjiang, Xintian and Taiyuan to obtain the values of corresponding m and n shown in the table 1.
TABLE 1
3) Value of a in index weight
The invention uses an exponential function weighting method, the smaller the root mean square error is, the larger the weight is, and the selection of a in the exponential function is determined by a plurality of groups of experiments. The method takes four stations of Nanjing, Wenjiang, Xintian and Taiyuan as target stations to carry out experiments respectively, wherein a is more than 0 and less than 1, takes the four selected target stations as an example, respectively takes different values of a to carry out experiments to obtain a relation curve of a relative to RMSE, and judges the optimal value of a. Fig. 4 is a graph of experimental results of four stations, where the abscissa is a value, and the ordinate is an RMSE value corresponding to a value, and it can be seen from a relationship graph obtained from a and RMSE that the larger the value of a is, the larger the RMSE value is, that is, the larger the error is, and according to the experimental results, when a is 0.01, the target station is used to perform an experiment with four stations of nanjing, wenjiang, new field, and taiyuan as target stations, and the root mean square error obtained is the minimum, which is 0.16, 0.105, 0.23, and 0.323, respectively, so that the present invention selects a to be 0.01 to perform the next experiment, and brings the a value into the next experiment, and estimates the air temperature data of the target station by using the ID-IDW algorithm of the present invention.
The results of the experiment are shown in FIGS. 5 and 6, wherein FIG. 5 shows the results of RMSE and MAE corresponding to the ID-IDW, IDW and SRT experiments. It can be seen that the ID-IDW performs best in the results of this experiment, and then the SRT, IDW is the relative poor point of the experiment. The RMSE result of the ID-IDW at the Nanjing station of the target station is approximately about 0.2, and the MAE is about 0.3; the RMSE result of the SRT at the Nanjing station of the target station is about 0.33, and the MAE is about 0.42; similarly, it can be seen that the RMSE of IDW at the target station Nanjing station is about 0.4, and the MAE is about 0.5, compared to the applicability of ID-IDW at this station, which is better than the performance of IDW and SRT. And similarly, the rest 3 sites are analyzed and compared to draw a similar conclusion, and the ID-IDW has a more prominent effect on the experimental result.
In order to verify the stability of the method, in this embodiment, 50 sets of experiments are performed on the same site respectively to see whether the fluctuation range of the RMSE and MAE is too large and the stability of the algorithm is not high due to the superiority and inferiority of the selected site, where fig. 6 shows the fluctuation results of the RMSE and MAE corresponding to the 50 sets of experiments on the four sites respectively. The results of 50 groups of experiments of the Nanjing station at the target station show that the RMSE change tends to be stable, the result is kept about 0.2, and the MAE is about 0.3; similarly, the experimental results of the target station in Wenjiangjun show that RMSE is about 0.1, MAE is about 0.25, RMSE is about 0.25 and MAE is about 0.38; RMSE is about 0.35 by taking the Tatarian as a target station, and MAE is about 0.48. Four groups of experiments basically prove that the stability of the algorithm is high, and the deviation value of the experimental result is small.
The experimental results were analyzed:
comparing and analyzing an experimental result obtained by the ID-IDW algorithm with the original IDW and SRT algorithms, wherein the evaluation of the model adopts Root Mean Square Error (RMSE) and Mean Absolute Error (MAE) for evaluation, wherein the RMSE mainly measures the error between an actual value and a predicted value, and the smaller the RMSE is, the smaller the difference between the actual value and the predicted value is, and the more accurate the predicted result is; the MAE can accurately reflect the actual prediction error, and the smaller the MAE is, the smaller the difference between the predicted value and the true value is. The experimental results are more ideal when both RMSE and MAE are relatively small.
The above is only a preferred embodiment of the present invention, and the protection scope of the present invention is not limited to the above-mentioned embodiments, and all technical solutions belonging to the idea of the present invention belong to the protection scope of the present invention. It should be noted that modifications and embellishments within the scope of the invention may be made by those skilled in the art without departing from the principle of the invention.
Claims (8)
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910686271.XA CN110426755B (en) | 2019-07-26 | 2019-07-26 | Ground temperature data quality control method based on improved reverse distance weighting |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910686271.XA CN110426755B (en) | 2019-07-26 | 2019-07-26 | Ground temperature data quality control method based on improved reverse distance weighting |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110426755A CN110426755A (en) | 2019-11-08 |
| CN110426755B true CN110426755B (en) | 2021-09-14 |
Family
ID=68411011
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910686271.XA Active CN110426755B (en) | 2019-07-26 | 2019-07-26 | Ground temperature data quality control method based on improved reverse distance weighting |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110426755B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111460695B (en) * | 2020-05-07 | 2023-08-25 | 南京信息工程大学 | A Quality Control Method of Surface Air Temperature Based on Cosine Similarity and Moving Surface Fitting |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102778702A (en) * | 2012-06-21 | 2012-11-14 | 重庆市电力公司 | Weather forecasting method and system for transformer station |
| JP2016200453A (en) * | 2015-04-08 | 2016-12-01 | 株式会社Ｎｔｔファシリティーズ | Meteorological data estimation method and electric power plant site evaluation method |
| CN109800921A (en) * | 2019-01-30 | 2019-05-24 | 北京师范大学 | A kind of Regional Fall Wheat yield estimation method based on remote sensing phenology assimilation and particle swarm optimization algorithm |
-
2019
- 2019-07-26 CN CN201910686271.XA patent/CN110426755B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102778702A (en) * | 2012-06-21 | 2012-11-14 | 重庆市电力公司 | Weather forecasting method and system for transformer station |
| JP2016200453A (en) * | 2015-04-08 | 2016-12-01 | 株式会社Ｎｔｔファシリティーズ | Meteorological data estimation method and electric power plant site evaluation method |
| CN109800921A (en) * | 2019-01-30 | 2019-05-24 | 北京师范大学 | A kind of Regional Fall Wheat yield estimation method based on remote sensing phenology assimilation and particle swarm optimization algorithm |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110426755A (en) | 2019-11-08 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106407627B (en) | A method and system for modeling wind speed probability distribution | |
| CN111327377B (en) | Method, device, equipment and storage medium for field intensity prediction | |
| CN106488548B (en) | A kind of determination method and device of indoor multipath error | |
| CN116822567B (en) | An optimization method for evaporative waveguide prediction model parameters | |
| CN106851571B (en) | Decision tree-based rapid KNN indoor WiFi positioning method | |
| CN108318856A (en) | The target positioning of fast accurate and tracking under a kind of heterogeneous network | |
| JP4998039B2 (en) | Observation data assimilation method | |
| CN102325370A (en) | High-precision three-dimensional positioner for wireless sensor network node | |
| CN110909303B (en) | An Adaptive Spatio-temporal Heterogeneity Inverse Distance Interpolation Method | |
| CN115494450B (en) | High-precision ultra-wideband indoor positioning tracking and control method and device | |
| CN114880373B (en) | Soil sampling method, system, storage medium and electronic equipment | |
| CN112733394A (en) | Atmospheric parameter inversion method and device | |
| CN110426755B (en) | Ground temperature data quality control method based on improved reverse distance weighting | |
| CN101873605A (en) | A Classification Method of Adaptive Propagation Environment in Network Planning | |
| CN113935402A (en) | Training method and device for time difference positioning model and electronic equipment | |
| CN105764088B (en) | A TDOA Cellular Location Method Based on Genetic Algorithm | |
| CN109190647B (en) | An Active and Passive Data Fusion Method | |
| CN118707528B (en) | A typhoon wind field inversion method and system | |
| CN109557567A (en) | A kind of big-dipper satellite selection method of modified gibbs sampler | |
| CN113573406A (en) | Fingerprint information positioning method based on small amount of wireless signal strength | |
| CN108848447B (en) | A Differential DV_Distance Node Localization Method Using Unknown Node Correction | |
| CN110765420A (en) | PSO-FI-based ground automatic meteorological station air temperature observation data quality control method | |
| CN113887058B (en) | Chloride ion deposition rate prediction method considering influence of distance from coastline and wind speed | |
| CN114630267B (en) | Indoor fusion positioning method based on error correction | |
| CN109886497A (en) | Surface air temperature interpolation method based on latitude-improved inverse distance weighting method |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| TR01 | Transfer of patent right | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20230531 Address after: Room 3831, Building 3, No. 89 Shuanggao Road, Gaochun Economic Development Zone, Gaochun District, Nanjing City, Jiangsu Province, 211302 Patentee after: Nanjing Kebo Environmental Technology Co.,Ltd. Address before: 210044 No. 219 Ning six road, Jiangbei new district, Nanjing, Jiangsu Patentee before: Nanjing University of Information Science and Technology |