CN106251022A - A kind of Short-term Climate Forecast method based on polyfactorial multiparameter similar set - Google Patents
A kind of Short-term Climate Forecast method based on polyfactorial multiparameter similar set Download PDFInfo
- Publication number
- CN106251022A CN106251022A CN201610644070.XA CN201610644070A CN106251022A CN 106251022 A CN106251022 A CN 106251022A CN 201610644070 A CN201610644070 A CN 201610644070A CN 106251022 A CN106251022 A CN 106251022A
- Authority
- CN
- China
- Prior art keywords
- prediction
- factor
- timeliness
- month
- key
- 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.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 47
- 230000002159 abnormal effect Effects 0.000 claims abstract description 14
- 238000005096 rolling process Methods 0.000 claims description 11
- 238000000354 decomposition reaction Methods 0.000 claims description 4
- 230000001550 time effect Effects 0.000 claims description 2
- PEDCQBHIVMGVHV-UHFFFAOYSA-N Glycerine Chemical compound OCC(O)CO PEDCQBHIVMGVHV-UHFFFAOYSA-N 0.000 claims 3
- 230000008520 organization Effects 0.000 claims 3
- 241000208340 Araliaceae Species 0.000 claims 1
- 235000005035 Panax pseudoginseng ssp. pseudoginseng Nutrition 0.000 claims 1
- 235000003140 Panax quinquefolius Nutrition 0.000 claims 1
- 239000000470 constituent Substances 0.000 claims 1
- 235000008434 ginseng Nutrition 0.000 claims 1
- 238000007689 inspection Methods 0.000 claims 1
- 238000010189 synthetic method Methods 0.000 claims 1
- 238000012360 testing method Methods 0.000 description 7
- 238000009826 distribution Methods 0.000 description 4
- 230000001932 seasonal effect Effects 0.000 description 4
- 230000015572 biosynthetic process Effects 0.000 description 3
- 238000004364 calculation method Methods 0.000 description 3
- 239000002131 composite material Substances 0.000 description 3
- 238000002790 cross-validation Methods 0.000 description 3
- 238000005259 measurement Methods 0.000 description 3
- 238000001556 precipitation Methods 0.000 description 3
- 238000003786 synthesis reaction Methods 0.000 description 3
- 238000004458 analytical method Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000011156 evaluation Methods 0.000 description 2
- 238000013277 forecasting method Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 230000000877 morphologic effect Effects 0.000 description 2
- 238000012353 t test Methods 0.000 description 2
- 230000005856 abnormality Effects 0.000 description 1
- 230000004931 aggregating effect Effects 0.000 description 1
- 238000012271 agricultural production Methods 0.000 description 1
- 238000012937 correction Methods 0.000 description 1
- 230000001186 cumulative effect Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 230000007774 longterm Effects 0.000 description 1
- 230000005906 menstruation Effects 0.000 description 1
- 230000010355 oscillation Effects 0.000 description 1
- 238000004379 similarity theory Methods 0.000 description 1
- 238000004088 simulation Methods 0.000 description 1
Classifications
-
- 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"
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Human Resources & Organizations (AREA)
- Economics (AREA)
- Strategic Management (AREA)
- Marketing (AREA)
- Game Theory and Decision Science (AREA)
- Entrepreneurship & Innovation (AREA)
- Development Economics (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Tourism & Hospitality (AREA)
- Physics & Mathematics (AREA)
- General Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
本发明公开了一种基于多因子的多参数相似集合的短期气候预测方法。本发明预测方法根据考虑不同因子时效的组合及其独立性，综合应用反映预测样本间空间距离和形态相似的多种相似参数，选取最佳相似年及相反年份，通过对多个最佳样本对应的要素进行合成集合，并进行交叉检验和逐月滚动，最终对目标进行预测。本发明在相似参数的选择及使用上有改进。其优点在于能针对可能发生异常变化的区域进行预测，不局限于因子与预报量之间线性部分的预测。相似方法的非线性使该方法在不同尺度预测领域存在更好的应用前景，也可用于模式产品的统计释用以提高模式预报水平。
The invention discloses a short-term climate prediction method based on multi-factor and multi-parameter similar sets. The prediction method of the present invention considers the combination of different factors and their independence, comprehensively applies a variety of similar parameters that reflect the spatial distance and shape similarity between prediction samples, selects the best similar year and the opposite year, and corresponds to a plurality of best samples. The elements are synthesized and aggregated, cross-checked and rolled month by month, and finally the target is predicted. The present invention improves upon the selection and use of similar parameters. Its advantage is that it can predict the region where abnormal changes may occur, and is not limited to the prediction of the linear part between the factor and the predictor. The nonlinearity of similar methods makes this method have better application prospects in different scale prediction fields, and it can also be used for statistical interpretation of model products to improve the level of model forecasting.
Description
技术领域technical field
本发明属于大气科学预测技术领域，具体涉及一种基于多因子的多参数相似集合的短期气候预测方法。The invention belongs to the technical field of atmospheric science prediction, and in particular relates to a short-term climate prediction method based on a multi-factor multi-parameter similar set.
背景技术Background technique
气候异常对国民经济及农业生产有很大影响，国家领导和各级政府都很重视。但由于气候系统空间尺度大、时间尺度长，气候异常及其成因相当复杂；而要对气候异常进行预测则相当困难。Climate anomalies have a great impact on the national economy and agricultural production, and national leaders and governments at all levels attach great importance to them. However, due to the large spatial scale and long time scale of the climate system, climate anomalies and their causes are quite complex; and it is quite difficult to predict climate anomalies.
我国已开展的气候预测业务主要集中在月、季、年尺度，即短期气候预测。现在业务中使用的主要方法有气候模式和各种统计及动力统计相结合的方法。当前，气候系统模式能合理预测东亚季风多时间尺度变率、主要空间模态分布、季风与其它气候系统尤其是与厄尔尼诺-南方涛动(ENSO)的关系，等等。虽然气候模式预测系统已能够对全球和区域气候特征尤其是ENSO、季风等主要气候现象展现出合理的预测技巧，且对我国气候尤其是极端气候异常事件的预测能力明显提升。但是，受分辨率有限、初始条件不确定、物理过程不完善、气候现象本身的可预报性有限等多种因素影响，气候模式对东亚气候的预测存在明显误差及巨大不确定性，特别是在东亚副热带季风区的预报能力很弱。总体看，我国短期气候预测水平仍十分有限。The climate prediction business that has been carried out in my country is mainly concentrated on the monthly, seasonal, and annual scales, that is, short-term climate prediction. The main methods used in business now are the combination of climate models and various statistics and dynamic statistics. At present, climate system models can reasonably predict the multi-timescale variability of the East Asian monsoon, the distribution of main spatial modes, the relationship between the monsoon and other climate systems, especially the El Niño-Southern Oscillation (ENSO), and so on. Although the climate model prediction system has been able to demonstrate reasonable prediction skills for global and regional climate characteristics, especially major climate phenomena such as ENSO and monsoon, and its ability to predict my country's climate, especially extreme climate anomalies, has been significantly improved. However, due to various factors such as limited resolution, uncertain initial conditions, imperfect physical processes, and limited predictability of climate phenomena, there are obvious errors and huge uncertainties in the prediction of East Asian climate by climate models, especially in the Forecast capability in the subtropical monsoon region of East Asia is weak. Overall, the level of short-term climate prediction in my country is still very limited.
为了提高预测水平，动力统计相结合的预测方法成为现阶段较为关注、又行之有效的预测方法之一。如通过发展集合预报技术、利用统计与动力相结合的降尺度方法等来提高短期气候预测技巧。其中相似理论及方法在各类预测领域及模式误差订正等方面被广泛应用。我们知道若要制作技巧水平高的预报，不仅需要较好地把握有意义的物理因子，还要有对好的方法的合理使用。相似预报效果的好坏除了受预测因子影响外，还主要受相似性度量的方法的影响。统计学上已有的度量相似的参数有十余种，但实际应用过程中的不同会导致差异，即使是同一种度量参数也可能有不同的结果。此外，不同相似度量参数的特征不同，在使用的时候最好加以区分。In order to improve the forecasting level, the forecasting method combined with dynamic statistics has become one of the more concerned and effective forecasting methods at the present stage. For example, improve short-term climate prediction skills by developing ensemble forecasting techniques and using downscaling methods combining statistics and dynamics. Among them, similarity theory and methods are widely used in various forecasting fields and model error correction. We know that in order to make forecasts with a high level of skill, not only a good grasp of meaningful physical factors is required, but also a reasonable use of good methods. The quality of the similarity forecast is not only affected by the predictors, but also mainly affected by the method of similarity measurement. There are more than ten parameters that measure similarity in statistics, but differences in the actual application process will lead to differences, and even the same measurement parameter may have different results. In addition, the characteristics of different similarity measurement parameters are different, and it is best to distinguish them when using them.
发明内容Contents of the invention
本发明的目的是为了解决现有技术中存在的缺陷，提供一种综合形态和距离相似的多参数相似预测方法。The object of the present invention is to solve the defects in the prior art, and provide a multi-parameter similarity prediction method that integrates shape and distance similarity.
为了达到上述目的，本发明提供了一种基于多因子的多参数相似集合的短期气候预测方法，该方法通过获取关键因子集，优化组合关键因子集，然后进行相似预测；其中，相似预测采用多因子多参数相似预测，具体方法如下：按照预测时效，利用对关键因子集优化组合得到的相应的组合因子，以预测年因子，运用欧式距离、相似系数、相似离度、海明距离反映样本间空间距离或者形态相似的多个相似参数，不同类型的相似参数两两组合使用，并依照先形态后距离或者先距离后形态相似的顺序，选取最佳相似年和相反年各若干年，通过合成方法对最佳相似年与相反年计算其合成值，并以统计t检验达到90％显著可信的区域站点为预测将发生明显异常的站点，以相似年合成值、相反年合成值及合成差值的加权平均作为目标预测值，对区域气候进行预测。In order to achieve the above object, the present invention provides a short-term climate prediction method based on multi-factor multi-parameter similar sets, the method obtains key factor sets, optimizes and combines key factor sets, and then performs similar prediction; wherein, similar prediction adopts multiple Factor multi-parameter similarity prediction, the specific method is as follows: according to the prediction timeliness, use the corresponding combination factors obtained by optimizing the combination of key factor sets to predict the year factor, and use the Euclidean distance, similarity coefficient, similarity distance, and Hamming distance to reflect the difference between samples. Multiple similarity parameters with similar spatial distance or shape are used in pairs of different types of similarity parameters, and according to the order of shape first and then distance or first distance and then shape similarity, select the best similar year and the number of opposite years respectively, through the synthesis The method calculates the combined value of the best similar year and the opposite year, and takes the regional stations with 90% significant reliability of the statistical t test as the stations that will have obvious anomalies, and uses the combined values of the similar years, the combined values of the opposite years and the combined difference The weighted average of the values is used as the target prediction value to predict the regional climate.
本发明短期气候预测方法的具体步骤如下：The concrete steps of short-term climate prediction method of the present invention are as follows:
(1)基本因子场关键区域选择：根据历史资料计算500hPa流函数、200hPa势函数及850hPa势函数作为基本因子场，针对预测目标，选择基本因子场关键区域；(1) Selection of key areas of the basic factor field: Calculate the 500hPa flow function, 200hPa potential function and 850hPa potential function as the basic factor field according to historical data, and select the key area of the basic factor field for the prediction target;
(2)关键区域因子主分量获得：对所述各基本因子场关键区域依照不同的预测时效分别进行经验正交函数分解，得到相应关键区域因子主分量；(2) Obtaining the principal components of the key regional factors: performing empirical orthogonal function decomposition on the key regions of the basic factor fields according to different prediction timeliness to obtain the corresponding key regional factor principal components;
(3)关键因子集获取：按照预测时效，进一步计算不同关键区域因子主成分与预测目标的相关系数，根据相关显著性选取相应预测时效的预测因子，获取关键因子集；(3) Acquisition of key factor sets: According to the forecast timeliness, further calculate the correlation coefficient between the principal components of different key regional factors and the forecast target, and select the predictors corresponding to the forecast timeliness according to the significance of the correlation to obtain the key factor set;
(4)优化组合关键因子集：按照预测时效，通过经验正交函数展开对关键因子集进行优化组合得到相应的组合因子，并保证各预测因子间的相对独立性；(4) Optimizing the combination of key factor sets: according to the forecast timeliness, the key factor sets are optimized and combined through empirical orthogonal function expansion to obtain the corresponding combination factors, and the relative independence of each predictor is guaranteed;
(5)多因子多参数相似预测：按照预测时效，利用对关键因子集优化组合得到的相应的组合因子，以预测年因子，运用欧式距离、相似系数、相似离度、海明距离反映样本间空间距离或者形态相似的多个相似参数，不同类型的相似参数两两组合使用，并依照先形态后距离或者先距离后形态相似的顺序，选取最佳相似年和相反年各若干年，通过合成方法对最佳相似年与相反年计算其合成差值，并以统计t检验达到90％显著可信的区域站点为预测将发生明显异常的站点，以相似年合成值、相反年合成值及合成差值的加权平均作为目标预测值，对区域气候进行预测；其中预测因子个数通过以下方法确定：预测因子个数根据因子组合模态所解释总体方差的贡献超过95％为标准，即步骤(4)对因子集进行经验正交函数展开累积方差贡献达到95％之前的模态个数为最终选定的因子个数；(5) Multi-factor and multi-parameter similarity prediction: According to the prediction timeliness, use the corresponding combination factors obtained by optimizing the combination of key factor sets to predict the year factor, and use Euclidean distance, similarity coefficient, similarity distance, and Hamming distance to reflect the sample distance. Multiple similarity parameters with similar spatial distance or shape are used in pairs of different types of similarity parameters, and according to the order of shape first and then distance or first distance and then shape similarity, select the best similar year and the number of opposite years respectively, through the synthesis The method calculates the combined difference between the best similar year and the opposite year, and takes the regional stations with a 90% significant credible statistical t test as the stations that will have obvious anomalies. The weighted average of the difference is used as the target forecast value to predict the regional climate; the number of predictors is determined by the following method: the number of predictors is based on the standard that the contribution of the overall variance explained by the factor combination mode exceeds 95%, that is, the step ( 4) The number of modes before the empirical orthogonal function expansion of the factor set reaches 95% of the cumulative variance contribution is the final number of selected factors;
(6)交叉检验集合预测：对所选取的历史资料依照预测时效进行交叉检验，同时根据步骤(5)对独立预测时段预测目标进行多次预测，通过集合给出预测目标的预测结果。(6) Cross-checking set forecasting: cross-checking the selected historical data according to the forecast timeliness, and making multiple forecasts for the forecast target in the independent forecast period according to step (5), and giving the forecast result of the forecast target through the set.
(7)逐月滚动集合预测：针对预测目标，按照预测时效逐月经步骤(5)、步骤(6)分别得到起报月前各月的多参数相似合成预测值及交叉检验集合预测结果，在此基础上，对起报月之前更长预测时效的可预测区域站点的有效预测结果滚动至对应预测时效的预测结果中，通过对不同预测时效的预测结果进行集合，得到预测目标的逐月滚动集合预测值；所述有效预测的确定方法如下：根据步骤(6)多次交叉检验集合预测具有有效的预测值的站点。(7) Month-by-month rolling ensemble forecast: For the forecast target, according to the forecast timeliness, step (5) and step (6) of menstruation are obtained respectively to obtain the multi-parameter similar synthetic forecast value and the cross-check set forecast result of each month before the reporting month. On this basis, the effective forecast results of sites in predictable areas with longer forecast timeliness before the start of the reporting month are rolled to the forecast results of the corresponding forecast timeliness, and the monthly rolling of the forecast target is obtained by aggregating the forecast results of different forecasting timeliness Aggregate predictive value; the method for determining the effective prediction is as follows: according to step (6) multiple times of cross-validation set prediction of sites with effective predictive value.
本发明相比现有技术具有以下优点：本发明气候预测方法运用多个独立因子，综合运用包含形态和距离相似的相似参数，通过计算在长时间历史资料库中选取针对预测年的最佳相似和相反样本，并对上述样本进行统计合成，对通过统计显著性达到90％的站点要素进行预测；同时，利用交叉检验集合、逐月滚动集合增加有效可预报站点数及预测结果的稳定性。本发明气候预测方法，综合使用多种参数，通过多因子多参数相似的计算选取最佳相似/相反样本，再结合统计合成与显著性检验方法，针对要素异常变化显著的区域站点进行预测，具有预测时效长、计算相对模式简单、预测结果较稳定，特别是对异常区域的预测具有较高的评分，该方法不仅仅局限于因子与预报量之间线性部分的预测。相似方法的非线性使该方法在不同尺度预测领域存在更好的应用前景，也可用于模式产品的统计释用以提高模式预报准确率水平。Compared with the prior art, the present invention has the following advantages: the climate prediction method of the present invention uses multiple independent factors, comprehensively uses similar parameters including shape and distance similarity, and selects the best similarity for the forecast year in the long-term historical database through calculation. And the opposite sample, and statistically synthesize the above samples, and predict the station elements that pass the statistical significance of 90%; at the same time, use the cross-validation set and monthly rolling set to increase the number of effective predictable stations and the stability of the prediction results. The climate prediction method of the present invention uses multiple parameters comprehensively, selects the best similar/contrary sample through multi-factor and multi-parameter similar calculations, and then combines statistical synthesis and significance testing methods to predict regional sites with significant abnormal changes in elements. The prediction time is long, the calculation is relatively simple, and the prediction results are relatively stable, especially the prediction of abnormal areas has a high score. This method is not limited to the prediction of the linear part between the factor and the predictor. The nonlinearity of similar methods makes this method have better application prospects in different scale prediction fields, and it can also be used for statistical interpretation of model products to improve the accuracy of model forecasts.
附图说明Description of drawings
图1为本发明气候预测方法的流程图；Fig. 1 is the flowchart of climate prediction method of the present invention;
图2是基于该方法进行拟合预测的有效可预报站点数随预测时效的分布；Figure 2 is the distribution of the number of effective predictable stations according to the fitting prediction based on this method with the prediction time;
图3是针对中国冬季气温拟合预测的Ps评分；Figure 3 is the Ps score for the fitting prediction of winter temperature in China;
图4是针对中国冬季气温拟合预测的Ts评分；Figure 4 is the Ts score for the fitting prediction of winter temperature in China;
图5是针对中国冬季气温拟合预测的ACC评分；Figure 5 is the ACC score for the fitting prediction of winter temperature in China;
图6是针对中国冬季气温拟合预测的Sk评分；Figure 6 is the Sk score for the fitting prediction of winter temperature in China;
图7 2015/2016年冬季对应的前期1月因子及相似年因子的演变特征。Figure 7 The evolution characteristics of the previous January factors and similar year factors corresponding to the winter of 2015/2016.
图中，ENC1为多因子多参数相似集合预测结果；ENC2为多因子多参数相似集合+交叉检验集合预测结果；ENC3为多因子多参数相似集合+交叉检验集合+逐月滚动集合预测结果；In the figure, ENC1 is the forecast result of multi-factor and multi-parameter similarity set; ENC2 is the forecast result of multi-factor multi-parameter similarity set + cross-check set; ENC3 is the forecast result of multi-factor multi-parameter similar set + cross-check set + month-by-month rolling set;
图3-6中a为针对2010/2011-2014/2015年进行独立预测的结果；b为针对1979/1980-2008/2009年冬季气温进行拟合的结果；In Figure 3-6, a is the result of independent prediction for 2010/2011-2014/2015; b is the result of fitting for the winter temperature of 1979/1980-2008/2009;
图7中包含分别按照仅考虑空间距离相似(E_D)、仅考虑空间形态相似(C_D)以及既考虑空间距离相似，又考虑空间形态相似(EC_D)选取的前期因子的演变特征；图中横坐标是因子序数，纵坐标是因子距平值。Figure 7 contains the evolution characteristics of the early factors selected according to the consideration of only spatial distance similarity (E_D), only spatial shape similarity (C_D), and both spatial distance similarity and spatial shape similarity (EC_D); the abscissa in the figure is the factor ordinal, and the ordinate is the factor anomaly.
具体实施方式detailed description
下面结合附图对本发明进行详细说明。The present invention will be described in detail below in conjunction with the accompanying drawings.
各相似参数及评估参数如下：The similar parameters and evaluation parameters are as follows:
1、相似参数1. Similar parameters
(1)相似系数(1) similarity coefficient
(2)欧式距离(2) Euclidean distance
(3)相似离度(3) similarity distance
Xijk＝Xik-Xjk X ijk ＝X ik -X jk
(4)海明距离(4) Hamming distance
Hij＝|Xik-Xjk|H ij ＝|X ik -X jk |
其中，X为因子场。k＝1,m为入选因子的维数；i为试预报年的序数；j为因子样本序数。其中相似参数(1)相似系数注重样本间的形态相似，(2)欧氏距离及(4)海明距离注重样本间的距离相似，(3)相似离度有说既注重形态相似，又注重距离相似。但也有文献证实(3)其实质注重空间距离的相似(罗阳等，2011)。本发明实施例以(1)-(2)、(1)-(3)、(1)-(4)两两组合为3组，每组计算得到2个最佳相似样本和2个最佳相反样本，这样针对某年的预测一共可得到6个最佳相似年和6个最佳相反样本。2、预测结果的评估参数Among them, X is the factor field. k=1, m is the dimension of the selected factor; i is the ordinal number of the test forecast year; j is the ordinal number of the factor sample. Among them, the similarity parameter (1) the similarity coefficient pays attention to the morphological similarity between samples, (2) the Euclidean distance and (4) the Hamming distance pays attention to the similar distance between samples, and (3) the similarity distance is said to focus on both morphological similarity and The distances are similar. But there are also literatures confirming (3) that its essence focuses on the similarity of spatial distance (Luo Yang et al., 2011). In the embodiment of the present invention, (1)-(2), (1)-(3), and (1)-(4) are combined into three groups, and each group is calculated to obtain 2 best similar samples and 2 best samples. Contrary samples, so the forecast for a certain year can get 6 best similar years and 6 best opposite samples. 2. Evaluation parameters of prediction results
(1)相关系数用下式表示(ACC)：(1) The correlation coefficient is expressed by the following formula (ACC):
式中，ΔRf、为降水距平百分率(或平均气温距平)的预报值及其平均值：ΔR0、为相应观测值：N为评分总站数。In the formula, ΔR f , is the forecast value of the percentage of precipitation anomaly (or mean temperature anomaly) and its mean value: ΔR 0 , is the corresponding observed value: N is the total number of scoring stations.
(2)技巧评分用下式表示(Sk)：(2) The skill score is represented by the following formula (S k ):
式中，Nα、N分别为预报准确的站数和参加评分的总站数：N′为基于某种无技巧预报期望能预报准确(这里仅指距平符号报对；预报和实况出现零距平均视为正距平)的站数。由于气候差异，不同地区、不同季节的温度、降水气侯概率会有所不同，这里定义随机预报的准确率，F＝(P1×P1+P2×P2)/(P1+P2)。In the formula, N α and N are the number of stations with accurate forecast and the total number of stations participating in scoring respectively: N′ is based on some kind of unskilled forecast and expects to be able to predict accurately (here only refers to the correctness of the anomaly symbol; the zero distance between the forecast and the actual The number of stations considered as positive anomalies on average). Due to climate differences, the climate probability of temperature and precipitation in different regions and seasons will be different. Here, the accuracy rate of random forecast is defined, F=(P 1 ×P 1 +P 2 ×P 2 )/(P 1 +P 2 ).
(3)异常级评分用下式表示(Ts)：(3) Abnormal score is expressed by the following formula (T s ):
异常级评分主要用来评估预测异常级的能力，通常指达到二级或一级异常，用下式表示：The abnormal grade score is mainly used to evaluate the ability to predict the abnormal grade, which usually refers to reaching the second or first grade abnormality, which is expressed by the following formula:
式中，Nf、NO分布表示预测和实况达到异常级的站数，Nc报对的异常级站数。TS评分表示报对的异常级站数占预测和实况异常级总站数的比，同时考虑了报错的影响。In the formula, the N f and N O distributions represent the number of stations that reach the abnormal level in the prediction and the actual situation, and the number of abnormal level stations reported by N c is correct. The TS score indicates the ratio of the number of abnormal stations reported correctly to the total number of predicted and live abnormal stations, and the impact of error reports is also taken into account.
(4)预报评分用下式表示(Ps)：(4) The prediction score is expressed by the following formula (P s ):
式中，N0为距平符号报对的以及预报与实况虽距平符号不同但都属正常级(各级标准见表1)的站数；N为参加评分范围内的总站数：n1、f1和n2、f2分别为一级异常报对和二级异常报对的站数和权重系数。一级和二级权重系数fi＝1/pi，fi与月或季的降水距平百分率达到一级或二级异常出现的气候概率(pi)成反比，称之为反比权重系数，根据1951—1995年历史资料按月、季分别统计得到(pi各月或季的值略)。为方便起见，在实际使用时取月或季平均(表2)的整数值作为固定权重系数，即：季节预报f1＝5，f2＝2。In the formula, N 0 is the number of stations where the anomaly symbols are correct, and the anomaly symbols of the forecast and the actual situation are different, but both belong to the normal level (see Table 1 for the standards of all levels); N is the total number of stations participating in the scoring range: n 1 , f 1 and n 2 , and f 2 are the number of stations and weight coefficients for the first-level abnormal reporting and the second-level abnormal reporting respectively. The first-level and second-level weight coefficients f i =1/p i , and f i is inversely proportional to the climate probability (p i ) when the monthly or seasonal precipitation anomaly reaches the first-level or second-level anomaly, which is called the inverse weight coefficient , obtained according to the historical data from 1951 to 1995 by month and quarter respectively (the values of p i for each month or quarter are omitted). For convenience, the integer value of the monthly or seasonal average (Table 2) is used as the fixed weight coefficient in actual use, namely: seasonal forecast f 1 =5, f 2 =2.
表1平均气温距平分级标准Table 1 Classification criteria for average temperature anomalies
表2各月、季一级或二级异常权重系数(f1或f2)计算值Table 2 Calculated values of first-level or second-level anomaly weight coefficients (f 1 or f 2 ) for each month and season
本发明短期气候预测以前期1月起报，针对当年12月-次年2月冬季气温进行预测，示例该方法的总体思路和效果。The short-term climate prediction of the present invention is reported from January in the previous period, and the winter temperature is predicted from December of the current year to February of the next year, and the general idea and effect of the method are exemplified.
首先，基于1979/1980-2008/2009年30年1月全国160站月平均气温资料和NCEP/NCAR逐月格点风场资料计算的500hPa高度场、200hPa与850hPa势函数资料。对1979/1980-2008/2009年冬季全国160站气温距平进行EOF(Empirical Orthogonal Function，经验正交函数)分解，并通过投影得到2009/2010-2014/2015年冬季气温距平对应的时间系数值。通过计算气温时间系数与不同超前时间的三层次因子场的全球相关，选取因子场的关键区域。逐月势函数(30°S～30°N)、逐月500hPa高度场(30°N～85°N)。本发明预测目标和预测因子均先选取因子增量与预测目标增量之间的关系对其年际增量进行预测，然后通过简单的和运算得到预测年的距平预测值。First, the 500hPa height field, 200hPa and 850hPa potential function data are calculated based on the monthly average temperature data of 160 stations across the country in January 1979/1980-2008/2009 and the monthly grid wind field data of NCEP/NCAR. EOF (Empirical Orthogonal Function, Empirical Orthogonal Function) decomposition was performed on the temperature anomalies of 160 stations across the country in the winter of 1979/1980-2008/2009, and the time coefficient corresponding to the temperature anomalies in the winter of 2009/2010-2014/2015 was obtained by projection value. By calculating the global correlation between the temperature time coefficient and the three-level factor field with different lead times, the key regions of the factor field are selected. Monthly potential function (30°S～30°N), monthly 500hPa height field (30°N～85°N). Both the forecast target and the forecast factor of the present invention first select the relationship between the factor increment and the forecast target increment to predict the inter-annual increment, and then obtain the anomaly forecast value of the forecast year through simple sum operation.
如图1所示，本发明提供的气候预测方法，包括如下步骤：As shown in Figure 1, the climate prediction method provided by the present invention comprises the following steps:
步骤1，根据超前时间组合相应的因子原场，对1979/1980-2008/2009年冬季全国160站气温距平对应的前期因子原场分别按关键区域范围进行EOF展开，得到对应的时间系数。这里预测时效从前一年的1月开始到次年1月；同样对独立试验的2010/2011-2014/2015年相应因子原场也通过投影得到其时间系数；Step 1: Combining the original field of corresponding factors according to the advanced time, the original field of previous factors corresponding to the temperature anomalies of 160 stations across the country in the winter of 1979/1980-2008/2009 were respectively expanded by EOF according to the scope of key regions, and the corresponding time coefficients were obtained. Here, the forecast time limit starts from January of the previous year to January of the next year; similarly, the original field of the corresponding factors of independent experiments in 2010/2011-2014/2015 is also projected to obtain its time coefficient;
步骤2，运用关键区域因子EOF分解得到的时间系数作为初选因子集，这里根据步骤1选择850hPa(30°S-30°N，0°-360°E)、200hPa势函数场(30°S-30°N，0°-360°E)、500hPa高度场(30°S-85°N，0°-360°E)的年际增量分别进行EOF分解，得到相应的时间系数；选取500hPa高度场、850hPa势函数场、200hPa势函数场时间系数分别为前12个、前5个、前2个共20个因子；再对上述因子集进行EOF分解，得到组合因子主成分作为最终的预测因子，第一因子即为组合因子的第一主成分；上述因子均按照不同预测时效进行分解组合，即前期2月起报的因子为前期2月因子场分析得到。Step 2, use the time coefficient obtained from the factor EOF decomposition of the key area as the primary selection factor set, here select 850hPa (30°S-30°N, 0°-360°E), 200hPa potential function field (30°S -30°N, 0°-360°E) and 500hPa height field (30°S-85°N, 0°-360°E) interannual increments are respectively decomposed by EOF to obtain corresponding time coefficients; select 500hPa The height field, the 850hPa potential function field, and the 200hPa potential function field time coefficients are the first 12, the first 5, and the first 2, a total of 20 factors; then the above-mentioned factor sets are decomposed by EOF, and the principal components of the combined factors are obtained as the final prediction Factors, the first factor is the first principal component of the combined factors; the above factors are decomposed and combined according to different forecasting timeliness, that is, the factors reported from February in the previous period are obtained from the factor field analysis in February in the previous period.
步骤3，对不同时效入选的20个因子集合按照时效进行组合并进行EOF展开，选取时间系数为最终的预测因子，预测因子个数以EOF展开模态所解释的方差贡献达到95％的个数为准。这里前12个因子模态可解释95.4％的总方差，所以最终选择的因子个数为12。Step 3: Combine the 20 factor sets selected according to the timeliness according to the timeliness and perform EOF expansion, select the time coefficient as the final predictor, and the number of predictors will account for 95% of the variance explained by the EOF expansion mode prevail. Here the first 12 factor modes can explain 95.4% of the total variance, so the number of factors finally selected is 12.
步骤4，依照预测时效，按照多因子的多参数相似得到针对拟合预测年的最佳相似和相反样本，并通过合成分析得到相应的预测值，多参数集合取多个最佳样本的合成预测的平均值(这里以相似年合成值、相反年合成值及合成差值的加权平均得到，权重系数依次分别取0.25、0.25和0.5。)；Step 4: According to the forecast timeliness, according to the multi-factor multi-parameter similarity, the best similar and opposite samples for the fitting forecast year are obtained, and the corresponding forecast values are obtained through synthetic analysis, and the multi-parameter set takes multiple best samples for synthetic forecasting (here obtained by the weighted average of the composite value of similar years, the composite value of opposite years and the composite difference, and the weight coefficients are respectively 0.25, 0.25 and 0.5 respectively.);
步骤5，依照预测时效，针对1979/1980-2008/2009年的资料进行交叉检验，每次检验建模时去掉其中的一年，这样一共进行30次检验和相似预测。每进行一次检验预测的同时对独立预测年(2010/2011-2014/2015年)都进行一次相似预测，通过这30次结果的集合得到目标的交叉检验集合预测值；通过步骤4-5得到的结果就为多因子多参数相似集合+交叉检验集合的预测结果。Step 5, according to the forecast timeliness, conduct cross-checks on the data from 1979/1980-2008/2009, and remove one of the years when modeling each test, so that a total of 30 tests and similar predictions are performed. Every time a test prediction is made, a similar prediction is made for the independent forecast year (2010/2011-2014/2015), and the cross-test set prediction value of the target is obtained through the collection of these 30 results; the obtained through steps 4-5 The result is the prediction result of the multi-factor multi-parameter similarity set + cross-validation set.
步骤6，针对预测目标依照不同时效的逐月预测经步骤4-5得到的多参数相似集合+交叉检验集合的预测结果，对起报月之前月份具有有效预测结果的站点的预测值滚动集合到该时效的预测值中，可得到多因子多参数相似集合+交叉检验集合+逐月滚动集合的预测结果。Step 6: According to the monthly prediction of the prediction target according to different timeliness, the prediction results of the multi-parameter similarity set + cross-checking set obtained in steps 4-5, the rolling collection of the prediction values of the sites with valid prediction results in the month before the reporting month to In the forecast value of this time effect, the prediction results of multi-factor and multi-parameter similarity set + cross-check set + month-by-month rolling set can be obtained.
如图2所示：由图2可以看出如果只考虑逐月的多因子多参数相似集合预测或者多因子多参数相似集合+交叉检验集合预测，而不实行逐月滚动，则从前一年1月开始对当年冬季的气温进行预测，其可以进行预报的有效站点随着预测时效的分布没有明显的增加或者减少趋势，总体可预测站点数较少。但如果进行逐月滚动，则可以进行预报的有效站点随着预测时效的变短而增加，自前期4月起可实现预测站点的全覆盖。As shown in Figure 2: It can be seen from Figure 2 that if only the monthly multi-factor multi-parameter similarity set forecast or the multi-factor multi-parameter similarity set + cross-test set forecast is considered without rolling month by month, then from the previous year 1 The winter temperature of the current year has been predicted since January, and the effective stations that can make predictions have no obvious increase or decrease trend with the distribution of prediction time, and the overall number of predictable stations is small. However, if the month-by-month rolling is carried out, the effective stations that can be forecasted will increase as the forecast time becomes shorter, and the full coverage of the forecasted stations can be realized from April in the previous period.
从图3-图6可以看出，预测时效从前一年的1月到当年1月。最长预测时效为13个月。From Figures 3 to 6, we can see that the time limit for forecasting is from January of the previous year to January of the current year. The longest forecast period is 13 months.
多因子多参数相似集合预测(ENC1)、多因子多参数相似集合+交叉检验集合预测(ENC2)和多因子多参数相似集合+交叉检验集合+逐月滚动集合预测(ENC3)三种方法的拟合预测评分Ps、距平相关系数ACC、基于随机预报的技巧评分Sk、异常级评分Ts分别为：91.3/91.8/91.8、0.58/0.61/0.61、0.50/0.53/0.53、0.62/0.68/0.68；相应的独立预测评估分别为：88.0/88.1/88.2、0.56/0.58/0.59、0.26/0.29/0.28、0.52/0.53/0.53。总体看，各种评分均显示ENC2较ENC1的评分要高，而ENC2同ENC3相比，虽然评分上没有显示出明显的优越性，但是由图2可知，ENC3较ENC2的可预报站点数明显增加，其预测效果还能够保持在较高的水平，且相对稳定。The simulation of three methods of multi-factor multi-parameter similarity ensemble prediction (ENC1), multi-factor multi-parameter similarity ensemble + cross-check ensemble forecast (ENC2) and multi-factor multi-parameter similarity ensemble + cross-check ensemble + month-by-month rolling ensemble forecast (ENC3) Combined prediction score P s , anomaly correlation coefficient ACC, skill score S k based on stochastic forecast, and abnormal grade score T s were: 91.3/91.8/91.8, 0.58/0.61/0.61, 0.50/0.53/0.53, 0.62/0.68 /0.68; the corresponding independent predictive assessments were: 88.0/88.1/88.2, 0.56/0.58/0.59, 0.26/0.29/0.28, 0.52/0.53/0.53, respectively. Overall, various scores show that ENC2 has a higher score than ENC1. Compared with ENC3, although ENC2 does not show obvious superiority in score, it can be seen from Figure 2 that the number of predictable stations of ENC3 is significantly higher than that of ENC2. , its prediction effect can also be maintained at a relatively high level and is relatively stable.
由图7可见，其中既考虑空间距离相似，又考虑空间形态相似(EC_D)选取的前期因子的演变特征与预测年因子的演变最相似。It can be seen from Figure 7 that the evolution characteristics of the early factors selected considering both the similarity in spatial distance and the similarity in spatial form (EC_D) are most similar to the evolution of the predicted year factors.
本发明实施例中涉及的一些技术包括：不排除在包括所述要素的过程、方法中还存在另外的相同要素。Some technologies involved in the embodiments of the present invention include: it is not excluded that there are other same elements in the process and method including the above elements.
以上所述仅是本发明的优选实施方式，应当指出，对于本技术领域的普通技术人员来说，在不脱离本发明原理的前提下，还可以做出若干改进和润饰，这些改进和润饰也应视本发明的保护范围。The above is only a preferred embodiment of the present invention, it should be pointed out that, for those of ordinary skill in the art, without departing from the principle of the present invention, some improvements and modifications can also be made, and these improvements and modifications can also be made. It should be regarded as the protection scope of the present invention.
Claims (5)
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610644070.XA CN106251022A (en) | 2016-08-08 | 2016-08-08 | A kind of Short-term Climate Forecast method based on polyfactorial multiparameter similar set |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610644070.XA CN106251022A (en) | 2016-08-08 | 2016-08-08 | A kind of Short-term Climate Forecast method based on polyfactorial multiparameter similar set |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN106251022A true CN106251022A (en) | 2016-12-21 |
Family
ID=58077940
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610644070.XA Pending CN106251022A (en) | 2016-08-08 | 2016-08-08 | A kind of Short-term Climate Forecast method based on polyfactorial multiparameter similar set |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106251022A (en) |
Cited By (15)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106953766A (en) * | 2017-03-31 | 2017-07-14 | 北京奇艺世纪科技有限公司 | A kind of alarm method and device |
| CN107103396A (en) * | 2017-06-13 | 2017-08-29 | 南京大学 | The Chinese seasonal climate Forecasting Methodology modeled based on main SVD mode |
| CN108957590A (en) * | 2018-05-22 | 2018-12-07 | 南京信息工程大学 | A kind of extracting method based on the real-time index of EEOF quasi-biweekly oscillation |
| CN109426886A (en) * | 2017-08-29 | 2019-03-05 | 北京思湃德信息技术有限公司 | A kind of climatic prediction system |
| CN110245773A (en) * | 2019-03-26 | 2019-09-17 | 国家气象中心 | A method for extracting multi-source live spatio-temporal predictors and incorporating them into model interpretation |
| CN110458338A (en) * | 2019-07-23 | 2019-11-15 | 天津大学 | A Medium- and Long-Term Statistical Prediction Method of Sea Surface Height Based on Reanalysis Data |
| CN110555554A (en) * | 2019-08-28 | 2019-12-10 | 向波 | intelligent climate prediction technology based on objective quantification |
| CN110990459A (en) * | 2019-12-04 | 2020-04-10 | 上海眼控科技股份有限公司 | Method and device for statistically analyzing historical fog-out process of airport based on weather data |
| CN111815041A (en) * | 2020-06-28 | 2020-10-23 | 青岛哈船海智科技有限公司 | Seawater temperature field prediction method based on improved EOF |
| CN112036617A (en) * | 2020-08-17 | 2020-12-04 | 国电大渡河流域水电开发有限公司 | A Dynamic-Statistical Objective Quantitative Climate Prediction Method and System |
| CN112381331A (en) * | 2020-12-02 | 2021-02-19 | 京津冀环境气象预报预警中心 | Atmospheric forecasting method using time weight-based similarity set algorithm |
| CN112748483A (en) * | 2020-12-24 | 2021-05-04 | 北京思湃德信息技术有限公司 | Air temperature forecast deviation correction method and device based on deep learning |
| CN113762618A (en) * | 2021-09-07 | 2021-12-07 | 中国水利水电科学研究院 | A lake water level prediction method based on multi-factor similarity analysis |
| CN113807583A (en) * | 2021-09-18 | 2021-12-17 | 南京信息工程大学 | Low-frequency increment space-time coupling-based power statistics combined sub-season prediction method |
| CN114330132A (en) * | 2021-12-30 | 2022-04-12 | 中国科学院大气物理研究所 | ENSO diversity prediction method based on artificial intelligence |
-
2016
- 2016-08-08 CN CN201610644070.XA patent/CN106251022A/en active Pending
Cited By (26)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106953766A (en) * | 2017-03-31 | 2017-07-14 | 北京奇艺世纪科技有限公司 | A kind of alarm method and device |
| CN107103396B (en) * | 2017-06-13 | 2019-02-19 | 南京大学 | A Seasonal Climate Prediction Method for China Based on Main SVD Mode Modeling |
| CN107103396A (en) * | 2017-06-13 | 2017-08-29 | 南京大学 | The Chinese seasonal climate Forecasting Methodology modeled based on main SVD mode |
| CN109426886A (en) * | 2017-08-29 | 2019-03-05 | 北京思湃德信息技术有限公司 | A kind of climatic prediction system |
| CN108957590B (en) * | 2018-05-22 | 2020-10-09 | 南京信息工程大学 | Extraction method based on EEOF quasi-bi-periodic oscillation real-time index |
| CN108957590A (en) * | 2018-05-22 | 2018-12-07 | 南京信息工程大学 | A kind of extracting method based on the real-time index of EEOF quasi-biweekly oscillation |
| CN110245773A (en) * | 2019-03-26 | 2019-09-17 | 国家气象中心 | A method for extracting multi-source live spatio-temporal predictors and incorporating them into model interpretation |
| CN110245773B (en) * | 2019-03-26 | 2021-06-25 | 国家气象中心 | A method for multi-source real-time spatiotemporal forecast factor extraction and incorporation into model interpretation application |
| CN110458338B (en) * | 2019-07-23 | 2023-01-03 | 天津大学 | Sea surface height medium-long term statistical prediction method based on reanalysis data |
| CN110458338A (en) * | 2019-07-23 | 2019-11-15 | 天津大学 | A Medium- and Long-Term Statistical Prediction Method of Sea Surface Height Based on Reanalysis Data |
| CN110555554A (en) * | 2019-08-28 | 2019-12-10 | 向波 | intelligent climate prediction technology based on objective quantification |
| CN110990459A (en) * | 2019-12-04 | 2020-04-10 | 上海眼控科技股份有限公司 | Method and device for statistically analyzing historical fog-out process of airport based on weather data |
| CN111815041A (en) * | 2020-06-28 | 2020-10-23 | 青岛哈船海智科技有限公司 | Seawater temperature field prediction method based on improved EOF |
| CN111815041B (en) * | 2020-06-28 | 2024-07-19 | 青岛哈船海智科技有限公司 | Seawater temperature field prediction method based on improved EOF |
| CN112036617A (en) * | 2020-08-17 | 2020-12-04 | 国电大渡河流域水电开发有限公司 | A Dynamic-Statistical Objective Quantitative Climate Prediction Method and System |
| CN112036617B (en) * | 2020-08-17 | 2021-10-29 | 国能大渡河流域水电开发有限公司 | A Dynamic-Statistical Objective Quantitative Climate Prediction Method and System |
| CN112381331A (en) * | 2020-12-02 | 2021-02-19 | 京津冀环境气象预报预警中心 | Atmospheric forecasting method using time weight-based similarity set algorithm |
| CN112381331B (en) * | 2020-12-02 | 2021-11-16 | 京津冀环境气象预报预警中心 | Atmospheric forecasting method using time weight-based similarity set algorithm |
| CN112748483B (en) * | 2020-12-24 | 2022-08-19 | 北京思湃德信息技术有限公司 | Air temperature forecast deviation correction method and device based on deep learning |
| CN112748483A (en) * | 2020-12-24 | 2021-05-04 | 北京思湃德信息技术有限公司 | Air temperature forecast deviation correction method and device based on deep learning |
| CN113762618A (en) * | 2021-09-07 | 2021-12-07 | 中国水利水电科学研究院 | A lake water level prediction method based on multi-factor similarity analysis |
| CN113762618B (en) * | 2021-09-07 | 2022-03-01 | 中国水利水电科学研究院 | Lake water level forecasting method based on multi-factor similarity analysis |
| CN113807583A (en) * | 2021-09-18 | 2021-12-17 | 南京信息工程大学 | Low-frequency increment space-time coupling-based power statistics combined sub-season prediction method |
| CN113807583B (en) * | 2021-09-18 | 2023-05-05 | 南京信息工程大学 | Subseasonal prediction method based on dynamic statistics combination of low-frequency increment space-time coupling |
| CN114330132A (en) * | 2021-12-30 | 2022-04-12 | 中国科学院大气物理研究所 | ENSO diversity prediction method based on artificial intelligence |
| CN114330132B (en) * | 2021-12-30 | 2022-07-01 | 中国科学院大气物理研究所 | An AI-based ENSO Diversity Forecast Method |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106251022A (en) | A kind of Short-term Climate Forecast method based on polyfactorial multiparameter similar set | |
| Shiru et al. | Performance evaluation of CMIP6 global climate models for selecting models for climate projection over Nigeria | |
| Abbasian et al. | Performance of the general circulation models in simulating temperature and precipitation over Iran | |
| CN103294848B (en) | Based on the satellite sun energy cell array life-span prediction method of mixture autoregressive moving average model | |
| Gualtieri et al. | Extrapolating wind speed time series vs. Weibull distribution to assess wind resource to the turbine hub height: A case study on coastal location in Southern Italy | |
| CN104932035A (en) | Typhoon intensity prediction method and system | |
| CN114461697B (en) | Urban carbon emission accounting method based on night light remote sensing and improved STIRPAT | |
| CN105740991A (en) | Climate change prediction method and system for fitting various climate modes based on modified BP neural network | |
| CN111027175A (en) | Method for evaluating social and economic influences of flood based on coupling model integrated simulation | |
| CN107563554A (en) | A kind of screening technique for counting the NO emissions reduction model prediction factor | |
| Shahriari et al. | Using the analog ensemble method as a proxy measurement for wind power predictability | |
| Cheng et al. | Statistical downscaling of hourly and daily climate scenarios for various meteorological variables in south-central Canada | |
| CN113779760A (en) | Dynamic-Statistical Combined Seasonal Climate Prediction Method Based on Predictable Climate Mode | |
| CN105974495A (en) | Method for pre-judging future average cloud amount of target area by using classification fitting method | |
| CN119067263B (en) | Regional sub-season wind energy resource prediction method based on dynamic climate mode, electronic equipment, storage medium and computer program product | |
| CN115438848A (en) | Long-term PM2.5 Concentration Prediction Method Based on Deep Mixed Graph Neural Network | |
| Abdolmanafi et al. | Evaluation of global ensemble prediction models for forecasting medium to heavy precipitations | |
| CN105678420A (en) | Cross-year and long-effectiveness climate prediction method | |
| CN117114176A (en) | Land utilization change prediction method and system based on data analysis and machine learning | |
| Vindel et al. | Discontinuity in the production rate due to the solar resource intermittency | |
| CN112380778B (en) | Weather drought prediction method based on sea temperature | |
| CN106227706A (en) | A kind of many climatic models output aggregation of data correction and uncertain appraisal procedure | |
| Massetti | Analysis and estimation of the effects of missing values on the calculation of monthly temperature indices | |
| Bhend et al. | Evaluation of simulated recent climate change in Australia | |
| Feldmann et al. | Statistical postprocessing of ensemble forecasts for temperature: The importance of spatial modeling |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20161221 |
|
| RJ01 | Rejection of invention patent application after publication |