CN106447746A - Air temperature space-time prediction distribution diagram drawing method - Google Patents
Air temperature space-time prediction distribution diagram drawing method Download PDFInfo
- Publication number
- CN106447746A CN106447746A CN201610850065.4A CN201610850065A CN106447746A CN 106447746 A CN106447746 A CN 106447746A CN 201610850065 A CN201610850065 A CN 201610850065A CN 106447746 A CN106447746 A CN 106447746A
- Authority
- CN
- China
- Prior art keywords
- model
- data
- temperature
- seasonal
- value
- 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
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T11/00—2D [Two Dimensional] image generation
- G06T11/60—Editing figures and text; Combining figures or text
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
Abstract
The invention discloses an air temperature space-time prediction distribution diagram drawing method, and the method comprises the steps: collecting data; selecting a model; determining the s value of a product season ARIMA(p, d, q)(P, D, Q) s model; determining parameter values of the model; carrying out the diagnosis and verification of the ARIMA(p, d, q)(P, D, Q) s model; carrying out the preprocessing of the Kriging interpolation data; building a Kriging model variation function; carrying out the Kriging interpolation and diagnosis; drawing an air temperature space-time prediction distribution diagram. The method solves problems that the predicted space-time distribution diagram is poor in precision and feasibility because the meteorological time and space rules are usually separated in a conventional meteorological data research method in the prior art, and the conventional meteorological data research method cannot fully explore the space-time data values through single time sequence analysis and spatial interpolation analysis, will lose the spatial distribution rule of data because the conventional meteorological data research method just carries out time analysis and will lose the time rule of station data because the conventional meteorological data research method just carries out the spatial analysis.
Description
Technical field：
The invention belongs to meteorological element spatio-temporal prediction distribution drawing drawing method technology, more particularly, to a kind of temperature spatio-temporal prediction
Distribution drawing drawing method.
Background technology：
Meteorological condition and social production life are closely bound up, and the spatio-temporal distribution rule of meteorological element is always to be studied
Emphasis, traditional be generally based on the instruments such as matlab, R for meteorological time-space distribution graph and studied, at technology path
In theory study stage, the poor feasibility of technology, and traditional meteorological data research method is often meteorological time and sky
Between rule separately carry out, simple utilization time series analysis and space interpolation analysis can not fully excavate space-time data be worth,
Only carry out the space distribution rule that time series analysis can lose data, only carry out spatial analysis and can lose station data time rule again
Rule, leads to the problems such as time-space distribution graph poor accuracy predicting, poor feasibility；In actual life, time and space are not point not
Open, people with greater need for be the rule information that time and space combine, be spatially distributed discrete meteorological site and uninterruptedly see
Survey and obtain substantial amounts of meteorology space-time data, how to excavate its Time Change using these meteorological datas and deduce acquisition and do not have
The weather law of observation station space of points position is the problem of social concerns.
Content of the invention：
The technical problem to be solved in the present invention：A kind of temperature spatio-temporal prediction distribution drawing drawing method is provided, existing to solve
It is often the meteorological time using traditional meteorological data research method in technology and spatial homing is separately carried out, simple utilization
Time series analysis and space interpolation analysis can not fully be excavated space-time data and be worth, and only carry out time series analysis and can lose data
Space distribution rule, only carries out spatial analysis and can lose station data temporal regularity again, leads to the time-space distribution graph essence predicting
Really property is poor, the problems such as poor feasibility.
Inventive technique scheme：
A kind of temperature spatio-temporal prediction is distributed drawing drawing method, and it includes：
Step 1, gathered data：The daily temperature observation data that in pickup area, all meteorological observation points are recorded, is calculated
Corresponding monthly mean temperature value, forms monthly mean temperature time series；
Step 2, model select：From product seasonal ARIMA (p, d, q) (P, D, Q) s model as time prediction model,
P, d, q and P, D, Q are continuity and seasonal autoregression (AR), difference (I), the rank value of rolling average (MA) respectively, and s is season
The when points that sexual cycle is comprised, from Kriging model as spatial prediction model；
Step 3, determine the s value of product seasonal ARIMA (p, d, q) (P, D, Q) s model；
Step 4, determine ARIMA (p, d, q) (P, D, Q) s model parameter value；
Step 5, the inspection of ARIMA (p, d, q) (P, D, Q) s Model Diagnosis；
Step 6, the data to Kriging regression pre-process；
Step 7, structure Kriging model variation function；
Step 8, Kriging regression and diagnosis；
Step 9, temperature spatio-temporal prediction distribution map are drawn.
The expression formula of ARIMA (p, d, q) (P, D, Q) the s model described in step 2 is：
In formula, B is backward shift operator, BsFor season item
Backward shift operator, Δ=1-B be difference operator, Δs=1-BsReferred to as seasonal difference operator,For autoregressive coefficient, ΦΡFor
Seasonal autoregressive coefficient, νqFor moving average coefficient, θQFor seasonal moving average coefficient, s is seasonal rhythm, and D is season
Property difference order, { xt, t=0, ± 1 ... } it is a Random time sequence, { zt, t=0, ± 1 ... it is a seasonal random time
Sequence.
Step 3.1, monthly mean temperature time series is decomposed, obtain observed, trend, seasonal,
Tetra- modules of random, represent former time series, trend term, Seasonal Regularity item and the random entry of temperature value respectively；Step 3.2,
Determine s value from the periodic quantity of the season item of monthly mean temperature Time Series.
Step 4 determines that the method for ARIMA (p, d, q) (P, D, Q) s model parameter value includes：
Step 4.1, the former time series data to temperature value do seasonal difference, and it is poor successively d and D assignment 0,1,2 to be observed
Whether steadily data sequence after point, if unstable, changes the value of d and D, when data item is steady after difference, current d's and D
It is worth the parameter value for d and D；
Step 4.2, p and P, the value of q and Q are passed through AIC or SBC criterion and are judged to draw.
Step 5, ARIMA (p, d, q) (P, D, Q) the s Model Diagnosis method of inspection are：By obtaining the residual error of model, it is residual
Difference shows as white noise, then model is rational.
Step 6 carries out preprocess method to the data of Kriging regression：Data is carried out to monthly mean temperature time series
Log change and spatially carry out a rank multinomial go trend process.
Described in step 7, the expression formula of structure Kriging model variation function is：
0*nugget+0.01002*stable (1.8182,1.7873), in formula：Nugget is block gold number, and stable is flat
Steady model.
Kriging regression described in step 8 and diagnostic method are：
Step 8.1, row interpolation is entered using ArcGIS software " statistical module "；
Step 8.2, row interpolation is entered using Kriging model ordinary kriging interpolation method；
Step 8.3, compare the interpolation accuracy of two kinds of interpolation methods；
Step 8.4, by the interpolation accuracy of the two determine using prediction average temperature of the whole year data carry out Ke Lijin
The conjunction of interpolation
Rationality.
Described in step 9, the method for temperature spatio-temporal prediction distribution map drafting is：Temperature space-time is carried out by ArcGIS software pre-
Survey distribution map to draw.
Beneficial effect of the present invention：
The present invention is that time-space attribute data carries out the time and space combines a kind of method of prediction to having, there is provided a set of
The practicable techniques route that temperature spatio-temporal prediction distribution map is drawn, compensate for the time during the space-time data regularity of distribution is explored and sky
Between the insufficient problem of information excavating, be effectively combined meteorological data time and spatial data rule；The method passes through the time
The spatially data that upper and spatially regularity is predicted future time and do not had data, can by obtaining meteorological time-space distribution graph
With the analyzed area meteorology regularity of distribution, this meteorological time space distribution had not only embodied locus rule but also had comprised time change
Rule, can predict the weather law of future time non-observation space point.Time and spatial information are combined by the present invention, fully profit
With time and the spatial information of meteorological data, obtain a kind of temperature spatio-temporal prediction and be distributed drawing drawing method, obtain space to meet
The region meteorology regularity of distribution and the demand of prediction future time meteorological condition, thus effective Instructing manufacture life；Solve existing
Having in technology using traditional meteorological data research method is often the meteorological time and spatial homing is separately carried out, simple fortune
Space-time data can not fully be excavated with time series analysis and space interpolation analysis to be worth, only carry out time series analysis and can lose data
Space distribution rule, only carry out spatial analysis and can lose station data temporal regularity again, lead to the time-space distribution graph predicting
Poor accuracy, the problems such as poor feasibility.
Brief description：
Meteorological observation website distribution map in Guizhou Province's in Fig. 1 embodiment of the present invention；
Guizhou average temperature of the whole year ARIMA model predication value schematic diagram in Fig. 2 embodiment of the present invention；
Fig. 3 applies average temperature of the whole year distribution schematic diagram in Guizhou in example for the present invention.
Specific embodiment：
A kind of temperature spatio-temporal prediction is distributed drawing drawing method, and it includes：
Step 1, gathered data：The daily temperature observation data that in pickup area, all meteorological observation points are recorded, is calculated
Corresponding monthly mean temperature value, forms monthly mean temperature time series；
Step 2, model select：From product seasonal ARIMA (p, d, q) (P, D, Q) s model as time prediction model,
P, d, q and P, D, Q are continuity and seasonal autoregression (AR), difference (I), the rank value of rolling average (MA) respectively, and s is season
The when points that sexual cycle is comprised, from Kriging model as spatial prediction model；
Step 3, determine the s value of product seasonal ARIMA (p, d, q) (P, D, Q) s model；
Step 4, determine ARIMA (p, d, q) (P, D, Q) s model parameter value；
Step 5, the inspection of ARIMA (p, d, q) (P, D, Q) s Model Diagnosis；
Step 6, the data to Kriging regression pre-process；
Step 7, structure Kriging model variation function；
Step 8, Kriging regression and diagnosis；
Step 9, temperature spatio-temporal prediction distribution map are drawn.
The expression formula of ARIMA (p, d, q) (P, D, Q) the s model described in step 2 is：
In formula, B is backward shift operator, BsFor season item
Backward shift operator, Δ=1-B be difference operator, Δs=1-BsReferred to as seasonal difference operator,For autoregressive coefficient, ΦΡFor
Seasonal autoregressive coefficient, νqFor moving average coefficient, θQFor seasonal moving average coefficient, s is seasonal rhythm, and D is season
Property difference order, { xt, t=0, ± 1 ... } it is a Random time sequence, { zt, t=0, ± 1 ... it is a seasonal random time
Sequence.
Step 3.1, monthly mean temperature time series is decomposed, obtain observed, trend, seasonal,
Tetra- modules of random, represent former time series, trend term, Seasonal Regularity item and the random entry of temperature value respectively；Step 3.2,
Determine s value from the periodic quantity of the season item of monthly mean temperature Time Series.
Step 4 determines that the method for ARIMA (p, d, q) (P, D, Q) s model parameter value includes：
Step 4.1, the former time series data to temperature value do seasonal difference, and it is poor successively d and D assignment 0,1,2 to be observed
Whether steadily data sequence after point, if unstable, changes the value of d and D, when data item is steady after difference, current d's and D
It is worth the parameter value for d and D；
Step 4.2, p and P, the value of q and Q are passed through AIC or SBC criterion and are judged to draw.
Step 5, ARIMA (p, d, q) (P, D, Q) the s Model Diagnosis method of inspection are：By obtaining the residual error of model, it is residual
Difference shows as white noise, then model is rational.
Step 6 carries out preprocess method to the data of Kriging regression：Data is carried out to monthly mean temperature time series
Log change and spatially carry out a rank multinomial go trend process.
Described in step 7, the expression formula of structure Kriging model variation function is：
0*nugget+0.01002*stable (1.8182,1.7873), in formula：Nugget is block gold number, and stable is flat
Steady model.
Kriging regression described in step 8 and diagnostic method are：
Step 8.1, row interpolation is entered using ArcGIS software " statistical module "；
Step 8.2, row interpolation is entered using Kriging model ordinary kriging interpolation method；
Step 8.3, compare the interpolation accuracy of two kinds of interpolation methods；
Step 8.4, by the interpolation accuracy of the two determine using prediction average temperature of the whole year data carry out Ke Lijin
The reasonability of interpolation.
Described in step 9, the method for temperature spatio-temporal prediction distribution map drafting is：Temperature space-time is carried out by ArcGIS software pre-
Survey distribution map to draw.
With reference to example, the present invention is made to refine further with explanation：A kind of temperature spatio-temporal prediction is distributed drawing drawing method,
It includes：
Step 1, model select：ARIMA model, is a kind of widely used Time Series Forecasting Methods model, its
Basic thought is：The data that prediction object over time is formed is considered as a random sequence, with certain data model
Carry out this sequence of approximate description, this model just can be predicted from seasonal effect in time series past value and present value after identified
Future value.
Product seasonal ARIMA (p, d, q) (P, D, Q) s model is comprised, it is continuity in ARIMA model concrete form
ARIMA (p, d, q) model and the combination of seasonal ARIMA (P, D, Q) s model, wherein p, d, q and P, D, Q are continuity respectively
With the rank value of seasonal autoregression (AR), difference (I), rolling average (MA), s is the time point that a seasonal rhythm is comprised
Number.When a time series does not contain only seasonal composition, is also mixed with Out of season composition, if single adopt seasonal ARIMA
Or Out of season ARIMA model is analyzed, its prediction effect can be undesirable, at this moment can with ARIMA seasonal ARIMA (p,
D, q) (P, D, Q) s model, the present invention uses seasonal ARIMA (p, d, q) (P, D, Q) s model, and its model formation is as follows.
In formula, B is backward shift operator, BsFor the backward shift operator of season item, Δ=1-B is referred to as difference operator, Δs=1-BsReferred to as
Seasonal difference operator,For autoregressive coefficient, ΦΡFor seasonal autoregressive coefficient, νqFor moving average coefficient, θQFor season
Property moving average coefficient, s is seasonal rhythm, and D is seasonal difference order, { xt, t=0, ± 1 ... it is a random time sequence
Row, { zt, t=0, ± 1 ... it is a seasonal Random time sequence.
Kriging regression is the core of geography information Geostatistical, is by the points of measurement known to space it is estimated that point to be estimated
A kind of method of data, common Kriging is one of Kriging interpolation method, and it is a kind of linear unbias best interpolation,
Take into full account spatial structural form.
ARIMA model is the auto-correlation relation of the different time node of analysis time sequence data, comes with this relation
Unknown timing node data is extrapolated, Kriging model is with spatial autocorrelation relation (being portrayed by variation function)
Spatial spreading interpolation of data is obtained space continuous data.Two methods one kind is temporal prediction, and a kind of is spatially
" prediction ", is all to be used as theoretical foundation by variable auto-correlation relation, the present invention is entered from ARIMA model and Kriging model
The spatio-temporal prediction of row temperature record.
Step 2, gathered data：The Guizhou being provided by Guizhou Meteorological Bureau of Shanxi Province is provided and is distributed the meteorological sight in discrete 19 inside the province
The observation data of survey station point, each meteorological observation website have recorded the daily temperature value from nineteen fifty-one to 2007, is obtained by calculating
Take corresponding monthly mean temperature value, each website forms a monthly mean temperature time sequence by nineteen fifty-one to 2007
Row, 19 monthly mean temperature time serieses altogether.Data is managed in the form of a table by PostgreSQL database.
Step 3, determine the s value of product seasonal ARIMA (p, d, q) (P, D, Q) s model：Build ARIMA model firstly the need of
Obtain monthly mean temperature seasonal effect in time series feature, by decomposing with addition model to monthly mean temperature time series, point
Solution obtains tetra- modules of observed, trend, seasonal, random, represents former time series, the trend of temperature value respectively
Item, Seasonal Regularity item and random entry.Random entry is the Data Representation after former time series removes trend and Seasonal Regularity, random number
Represent that former time series feature is embodied by season item and trend term substantially according to showing as white-noise process.From the monthly mean temperature time
The season item regularity that sequence is decomposed shows as 12 periodicity, therefore, it is determined that s in ARIMA (p, d, q) (P, D, Q) s model
It is worth for 12.
Value for clear and definite s further is 12, and the data autocorrelation analysis method using different delays step-length is verified, leads to
Cross being fitted of scatterplot and straight line postponing that step-length is from 1 to 12, matching better explanation correlation is bigger, knows when delay step-length
For matching when 12 preferably, represent that i-th month of certain year is the strongest with (i+12) moon (i.e. i-th month of next year) data dependence, thus
Demonstrate the monthly mean temperature time series cycle be 12, ARIMA (p, d, q) (P, D, Q) s model in s can be determined that equal to 12.
Step 4, determine ARIMA (p, d, q) (P, D, Q) s model parameter value：Seasonal ARIMA (p, d, q) (P, D, Q) s model
Parameter Estimation is to determine the process of model parameter p, d, q, P, D, Q, it is first determined the value of d and D, and d, D are difference number of times, typically take
It is worth for one of 0,1,2, first seasonal difference is done to former data, successively differentiated data sequence is observed to d and D assignment 0,1,2
Whether steadily, if unstable, the value of modification d and D, when after difference, data item is steady, (whether difference is steadily by difference side
Method ACF and PACF result are shown), project difference activity d=1, during D=0, difference result is optimum, and after difference, data sequence tends to
Steadily.Thus judging d=1, D=0.
Model other specification determines and judges referring generally to AIC or SBC criterion.It is uncorrelated that AIC criterion embodies residual error simultaneously
Principle and succinct principle, and eliminate the subjective impact of modeler, but require data sequence Normal Distribution and carry out maximum
Possibility predication；SBC is Schwarz bayesian criterion, and it considers the impact of residual error number more than AIC, works as N>During ∞, SBC
The best model exponent number that criterion determines is often consistent with the exponent number of true model, and AIC criterion is then than the exponent number of true model
Height, when AIC value is all less with SBC value, model is preferable.Invention determines d, D value (d=1, D=0) in model first, then passes through
Auto.arima () method in R language, selects SBC criterion to carry out model selection.Experiment proves ARIMA (1,1,1) (0,0,1)
When [12], SBC is optimum, and under this model, AIC value and BIC are little compared with other model values, its corresponding variance, maximum likelihood value, AIC value
It is shown in Table 1 with BIC value.
Model is to further determine that selected model is optimum after determining, prevents model parameter from selecting mistake, to possible mould
Type is modeled respectively, and by mean error (ME), root-mean-square error (RMSE), mean absolute error (MAE), average hundred
Point error (MPE), average absolute percent error (MAPE), average absolute scale error (MASE) carry out preference pattern, these parameters
More little required model is better, and reference table 1 may further determine that ARIMA (1,1,1) (0,0,1) [12] is optimum, further really
The reasonability of fixed required model.
Table 1 ARIMA forecast model comparative analysis
Model construction can obtain the monthly mean temperature of 2007 using the prediction that Guizhou meteorological data realized in R language after finishing
Value.
Step 5, the inspection of ARIMA (p, d, q) (P, D, Q) s Model Diagnosis：If the residual error of model prediction result shows as white noise
Sound, that is, itself ACF and PCAF should not have significant difference with zero, if having several models all to meet parameter have significant difference simultaneously,
Residual sequence is the requirement of white-noise process, then can consider the goodness statistic (AIC) of model and bayesian criterion (SBC),
Square root (Std.EE) evaluation of residual sequence variance (VE) and variance evaluation, it is pre- that AIC, SBC, VE, Std.EE get over mini Mod
Survey effect better.By said method, ARIMA (1,1,1) (0,0,1) [12] model selecting is obtained to predict the outcome and examine
Survey it was demonstrated that the residual error that predicts the outcome fluctuates near null value and shows as white noise, therefore diagnostic model is rational.
Step 6, the data to Kriging regression pre-process：Average temperature of the whole year data is entered and first has to during row interpolation completely
Sufficient data fit second-order stationary and the condition of intrinsic hypothesis, therefore need data is processed before Kriging interpolation, this
Bright data is carried out just too property detection find data be not just to be distributed very much, therefore interpolation advance row data log change and
Spatially carrying out a rank multinomial goes trend to process, and data preprocessing method employs statistical module function in ArcGIS software
Carry out, realize this step data prediction by selecting data smooth way button.
Step 7, structure Kriging model variation function：ArcGIS software geo-statistic mould is adopted to the data that step 6 is processed
The Kriging regression function of block is carried out, and selects variation function model before carrying out Kriging regression first, and the variation function of matching is：
0*nugget+0.01002*stable (1.8182,1.7873), nugget are block gold number, and stable is stationary model, wherein becomes
Journey is that 1.8182 expressions think that in 1.8182km meteorological site average temperature of the whole year is related, beyond this bound variable not phase
Close.Variation function features the space structure of average temperature of the whole year, reflects data distribution correlation spatially, is by sky
The theoretical foundation of interpolation.
Step 8, Kriging regression and diagnosis：Average temperature of the whole year distribution maps in 2007 are obtained by Kriging regression and examines
Disconnected, row interpolation is entered using ArcGIS software " statistical module ", have chosen ordinary kriging interpolation method and enter row interpolation, interpolation essence
Whether rationally the evaluation one side of degree embodies interpolation method, on the other hand can describe the interpolation order of accuarcy of zones of different,
(one kind enters row interpolation for True Data to two kinds of interpolation precisions of table 2 record, and a kind of is that the data that the inventive method is predicted is carried out
Interpolation), the validity of the inventive method can be evaluated by comparative analysis.Mean error, root-mean-square error and standard are averagely missed
Difference is the smaller the better, and closer to 1, root mean square standard error value represents that interpolation result is better, two kinds of interpolation results second of comparative analysis
Preferably, the inventive method is slightly poor for kind of (gathered data calculated value interpolation) interpolation result, but the precision of interpolation poor all at 0.01
On, represent two kinds of interpolation result precision quite, the average temperature of the whole year of present invention prediction is entered row interpolation and calculated by observed data
Average temperature of the whole year interpolation precision similar.In addition root mean square standard error is more than 1, represents that interpolative prediction slightly underestimates the change in space
The opposite sex.
The interpolation precision contrast of 2 two kinds of interpolation of table
Step 9, temperature spatio-temporal prediction distribution map are drawn：Detecting step 8 interpolation method rationally, then enters promoting the circulation of qi by ArcGIS
Warm spatio-temporal prediction distribution map is drawn, thus obtaining the average temperature of the whole year Distribution value situation of 2007 of Guizhou Province region, this distribution
Figure is to observe data drafting by Guizhou Province's 19 survey station points before in 2006 to form.
Claims (9)
1. a kind of temperature spatio-temporal prediction is distributed drawing drawing method, and it includes：
Step 1, gathered data：The daily temperature observation data that in pickup area, all meteorological observation points are recorded, is calculated corresponding
Monthly mean temperature value, formed monthly mean temperature time series；
Step 2, model select：From product seasonal ARIMA (p, d, q) (P, D, Q) s model as time prediction model, p, d, q
It is the rank value of continuity and seasonal autoregression (AR), difference (I), rolling average (MA) respectively with P, D, Q, s is seasonal all
The when points that phase is comprised, from Kriging model as spatial prediction model；
Step 3, determine the s value of product seasonal ARIMA (p, d, q) (P, D, Q) s model；
Step 4, determine ARIMA (p, d, q) (P, D, Q) s model parameter value；
Step 5, the inspection of ARIMA (p, d, q) (P, D, Q) s Model Diagnosis；
Step 6, the data to Kriging regression pre-process；
Step 7, structure Kriging model variation function；
Step 8, Kriging regression and diagnosis；
Step 9, temperature spatio-temporal prediction distribution map are drawn.
2. a kind of temperature spatio-temporal prediction distribution drawing drawing method according to claim 1 it is characterised in that：
The expression formula of ARIMA (p, d, q) (P, D, Q) the s model described in step 2 is：
In formula, B is backward shift operator, BsAfter season item
Move operator, Δ=1-B is difference operator, Δs=1-BsReferred to as seasonal difference operator,For autoregressive coefficient, ΦΡFor season
Property autoregressive coefficient, νqFor moving average coefficient, θQFor seasonal moving average coefficient, s is seasonal rhythm, and D is seasonal poor
Sublevel number, { xt, t=0, ± 1 ... } it is a Random time sequence, { zt, t=0, ± 1 ... it is a seasonal random time sequence
Row.
3. a kind of temperature spatio-temporal prediction distribution drawing drawing method according to claim 1 it is characterised in that：Described in step 3
The determination method of s value include：Step 3.1, monthly mean temperature time series is decomposed, obtain observed, trend,
Tetra- modules of seasonal, random, represent former time series, trend term, Seasonal Regularity item and the random entry of temperature value respectively；
Step 3.2, determine s value from the periodic quantity of the season item of monthly mean temperature Time Series.
4. a kind of temperature spatio-temporal prediction distribution drawing drawing method according to claim 1 it is characterised in that：Step 4 determines
The method of ARIMA (p, d, q) (P, D, Q) s model parameter value includes：
Step 4.1, the former time series data to temperature value do seasonal difference, successively d and D assignment 0,1,2 are observed after difference
Data sequence whether steadily, if unstable, the value of modification d and D, when after difference, data item is steady, the value of current d and D is d
Parameter value with D；
Step 4.2, p and P, the value of q and Q are passed through AIC or SBC criterion and are judged to draw.
5. a kind of temperature spatio-temporal prediction distribution drawing drawing method according to claim 1 it is characterised in that：Step 5,
ARIMA (p, d, q) (P, D, Q) the s Model Diagnosis method of inspection is：By obtaining the residual error of model, its residual error shows as white noise,
Then model is rational.
6. a kind of temperature spatio-temporal prediction distribution drawing drawing method according to claim 1 it is characterised in that：Step 6 to gram
In the data of golden interpolation carry out preprocess method and be：Monthly mean temperature time series is carried out with the log change of data and in space
On carry out single order multinomial go trend process.
7. a kind of temperature spatio-temporal prediction distribution drawing drawing method according to claim 1 it is characterised in that：Described in step 7
Build Kriging model variation function expression formula be：
0*nugget+0.01002*stable (1.8182,1.7873), in formula：Nugget is block gold number, and stable is steady mould
Type.
8. a kind of temperature spatio-temporal prediction distribution drawing drawing method according to claim 1 it is characterised in that：Described in step 8
Kriging regression and diagnostic method are：
Step 8.1, row interpolation is entered using ArcGIS software " statistical module "；
Step 8.2, row interpolation is entered using Kriging model ordinary kriging interpolation method；
Step 8.3, compare the interpolation accuracy of two kinds of interpolation methods；
Step 8.4, by the interpolation accuracy of the two determine using prediction average temperature of the whole year data carry out Kriging regression
Reasonability.
9. a kind of temperature spatio-temporal prediction distribution drawing drawing method according to claim 1 it is characterised in that：Described in step 9
Temperature spatio-temporal prediction distribution map draw method be：Carry out temperature spatio-temporal prediction distribution map by ArcGIS software to draw.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610850065.4A CN106447746B (en) | 2016-09-26 | 2016-09-26 | A kind of temperature spatio-temporal prediction is distributed drawing drawing method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610850065.4A CN106447746B (en) | 2016-09-26 | 2016-09-26 | A kind of temperature spatio-temporal prediction is distributed drawing drawing method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN106447746A true CN106447746A (en) | 2017-02-22 |
| CN106447746B CN106447746B (en) | 2018-06-22 |
Family
ID=58169514
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610850065.4A Active CN106447746B (en) | 2016-09-26 | 2016-09-26 | A kind of temperature spatio-temporal prediction is distributed drawing drawing method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106447746B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109933839A (en) * | 2019-01-11 | 2019-06-25 | 青岛理工大学 | A kind of integrated form pole symmetric mode decomposition method about the analysis of Space Time data |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103473408A (en) * | 2013-08-28 | 2013-12-25 | 河南大学 | Method for restoring missing air temperature records on basis of spatial-temporal information fusion |
| CN104504465A (en) * | 2014-12-16 | 2015-04-08 | 国电南京自动化股份有限公司 | Power generation fuel supply prediction method |
-
2016
- 2016-09-26 CN CN201610850065.4A patent/CN106447746B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103473408A (en) * | 2013-08-28 | 2013-12-25 | 河南大学 | Method for restoring missing air temperature records on basis of spatial-temporal information fusion |
| CN104504465A (en) * | 2014-12-16 | 2015-04-08 | 国电南京自动化股份有限公司 | Power generation fuel supply prediction method |
Non-Patent Citations (4)
| Title |
|---|
| 刘付程等: "基于季节ARIMA模型的近海表层水温时间序列模拟与预测", 《淮海工学院学报(自然科学版)》 * |
| 李莎等: "基于时空变异函数的Kriging插值及实现", 《计算机工程与应用》 * |
| 石昱馨 等: "乘积季节模型在气温及降水时间序列分析中的应用", 《数学·力学·物理学·高新技术交叉研究进展——2010（13）卷 》 * |
| 米晓将: "区域尺度下月均气温的时空演化格局研究", 《中国优秀硕士学位论文全文数据库》 * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109933839A (en) * | 2019-01-11 | 2019-06-25 | 青岛理工大学 | A kind of integrated form pole symmetric mode decomposition method about the analysis of Space Time data |
| CN109933839B (en) * | 2019-01-11 | 2023-04-14 | 青岛理工大学 | Integrated pole symmetric modal decomposition method for time-space data analysis |
Also Published As
| Publication number | Publication date |
|---|---|
| CN106447746B (en) | 2018-06-22 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111859800B (en) | Space-time estimation and prediction method for PM2.5 concentration distribution | |
| Hall | Machine learning approaches to macroeconomic forecasting | |
| CN107992968B (en) | Electric energy meter metering error prediction method based on integrated time series analysis technology | |
| CN113077097B (en) | Air quality prediction method based on depth space-time similarity | |
| CN111967688B (en) | Power load prediction method based on Kalman filter and convolutional neural network | |
| Honey et al. | From rags to fishes: data-poor methods for fishery managers | |
| CN106600037B (en) | Multi-parameter auxiliary load prediction method based on principal component analysis | |
| Girardi et al. | The role of indicator selection in nowcasting euro-area GDP in pseudo-real time | |
| CN109598935A (en) | A kind of traffic data prediction technique based on ultra-long time sequence | |
| CN115495991A (en) | Rainfall interval prediction method based on time convolution network | |
| CN110210694A (en) | Space management, device, storage medium and computer equipment | |
| CN109376937B (en) | Self-adaptive scheduling end-of-term water level prediction method based on ensemble empirical mode decomposition | |
| CN105488598A (en) | Medium-and-long time electric power load prediction method based on fuzzy clustering | |
| CN106447746A (en) | Air temperature space-time prediction distribution diagram drawing method | |
| CN105354644A (en) | Financial time series prediction method based on integrated empirical mode decomposition and 1-norm support vector machine quantile regression | |
| CN106683001A (en) | Thermal power plant set identification data selection method based on historical operation data | |
| CN113868597B (en) | Regression fairness measurement method for age estimation | |
| CN113642600B (en) | Driving behavior feature extraction method based on mRMR algorithm and principal component analysis | |
| CN105069214A (en) | Process reliability evaluation method based on nonlinear correlation analysis | |
| CN108665090A (en) | Urban distribution network saturation load forecasting method based on principal component analysis Yu Verhulst models | |
| CN115496264A (en) | Method for predicting generated power of wind turbine generator | |
| CN114154667A (en) | Mixed time series elevator operation parameter prediction method based on big data | |
| Beamonte et al. | Retrospective analysis of the dwelling price by means of STAR models with neighborhood effects | |
| KR20200092469A (en) | Apparatus and method for providing fashion coordination knowledge based on a neural network having explicit memory | |
| Panda et al. | Machine learning using exploratory analysis to predict taxi fare |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20211109 Address after: 550000 No. 17 Binhe Road, Nanming District, Guiyang City, Guizhou Province Patentee after: Guizhou Power Grid Co.,Ltd. Address before: 550002 4th floor, No.5 Baba street, Guiyang City, Guizhou Province Patentee before: ELECTRICITY TRANSMISSION RUN OVERHAUL BRANCH, GUIZHOU POWER GRID CO.,LTD. Patentee before: GUIZHOU ELECTRIC POWER DESIGN INSTITUTE |
|
| TR01 | Transfer of patent right |