WO2023240509A1 - 一种基于降水预报与遥相关对应关系的空间概率分析方法及系统 - Google Patents

一种基于降水预报与遥相关对应关系的空间概率分析方法及系统 Download PDF

Info

Publication number
WO2023240509A1
WO2023240509A1 PCT/CN2022/099021 CN2022099021W WO2023240509A1 WO 2023240509 A1 WO2023240509 A1 WO 2023240509A1 CN 2022099021 W CN2022099021 W CN 2022099021W WO 2023240509 A1 WO2023240509 A1 WO 2023240509A1
Authority
WO
WIPO (PCT)
Prior art keywords
forecast
coefficient
teleconnection
observation
precipitation
Prior art date
Application number
PCT/CN2022/099021
Other languages
English (en)
French (fr)
Inventor
赵铜铁钢
陈浩玲
Original Assignee
中山大学
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by 中山大学 filed Critical 中山大学
Priority to PCT/CN2022/099021 priority Critical patent/WO2023240509A1/zh
Publication of WO2023240509A1 publication Critical patent/WO2023240509A1/zh

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
    • G06F17/10—Complex mathematical operations
    • G06F17/15—Correlation function computation including computation of convolution operations
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
    • G06F17/10—Complex mathematical operations
    • G06F17/16—Matrix or vector computation, e.g. matrix-matrix or matrix-vector multiplication, matrix factorization
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
    • G06F17/10—Complex mathematical operations
    • G06F17/18—Complex mathematical operations for evaluating statistical data, e.g. average values, frequency distributions, probability functions, regression analysis
    • 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"

Definitions

  • the present invention relates to the technical field of precipitation forecast analysis, and more specifically, to a spatial probability analysis method and system based on the correspondence between precipitation forecast and teleconnection.
  • ENSO and precipitation teleconnection coefficients provide important support for global seasonal precipitation forecasts.
  • Multiple analytical studies have shown that the main source of predictability information for seasonal precipitation forecasts is the ENSO signal, and pointed out that the strength of the teleconnection coefficient between ENSO and regional precipitation corresponds to the accuracy of precipitation forecasts. Therefore, the ability of global meteorological models to capture and characterize the ENSO-precipitation teleconnection coefficient provides a key entry point for the assessment of the applicability of global seasonal precipitation forecasts.
  • the similarity of forecast accuracy and teleconnection intensity spatial distribution is often directly compared, making it difficult to give a quantitative description of the corresponding relationship between the two.
  • the present invention provides a method based on Spatial probability analysis method and system for the correspondence between precipitation forecast and teleconnection.
  • a spatial probability analysis method based on the correspondence between precipitation forecast and teleconnection including the following steps:
  • the forecast-observation correlation coefficient and meteorological factor-observed precipitation teleconnection coefficient of each grid are calculated respectively, and based on the significance of the forecast-observation correlation coefficient and the meteorological factor-observed precipitation teleconnection coefficient, each network is evaluated Grid classification;
  • the spatial consistency probability that the forecast-observation correlation coefficient is significantly positive is calculated.
  • the present invention also proposes a spatial probability analysis system based on the correspondence between precipitation forecast and teleconnection, applying the above-mentioned spatial probability analysis method based on the correspondence between precipitation forecast and teleconnection. It includes a data acquisition module, a correlation coefficient calculation module, a classification module, a significance judgment module, a spatial weight coefficient calculation module, and a spatial consistency probability analysis module.
  • the data acquisition module is used to obtain the sample sequence of precipitation forecasts to be analyzed, as well as the corresponding sample sequence of observed precipitation and meteorological factors;
  • the correlation coefficient calculation module is used to calculate each element in the target area based on the obtained sample sequence.
  • the classification module is used to analyze the significance of the forecast-observation correlation coefficient and the meteorological factor-observed precipitation teleconnection coefficient, and conduct analysis on each grid based on the analysis results.
  • the significance judgment module is used to judge the correspondence between the forecast-observation correlation coefficient and the teleconnection coefficient based on the grid classification results
  • the spatial weight coefficient calculation module is used to calculate the spatial weight coefficient based on the spatial coordinates of the grid to obtain the spatial weight coefficient matrix
  • the spatial consistency probability analysis module is used to calculate the spatial consistency probability that the forecast-observation correlation coefficient is a significant positive correlation, and the forecast-observation correlation coefficient based on the spatial weight coefficient matrix and the corresponding relationship between the forecast-observation correlation coefficient and the teleconnection coefficient. The spatial consistency probability of the corresponding relationship with different teleconnection coefficients respectively.
  • the beneficial effects of the technical solution of the present invention are: by combining the spatial relationship and probability of forecast precipitation, the present invention quantifies the spatial consistency probability that the correlation coefficient between forecast and observation is significantly positive, and can decompose it It is the spatial consistency probability of different corresponding relationships with teleconnection effects, thereby providing a reference for the evaluation and selection of precipitation forecast products.
  • Figure 1 is a flow chart of a spatial probability analysis method based on the correspondence between precipitation forecast and teleconnection according to an embodiment of the present invention.
  • Figure 2 is a schematic diagram of the significance classification results of DJF’s forecast-observation correlation coefficient.
  • Figure 3 is a schematic diagram of the significance classification results of DJF’s teleconnection coefficients.
  • Figure 4 shows the spatial consistency probability distribution diagram with a significant positive correlation between the forecast-observation correlation coefficient.
  • Figure 5 shows the spatial consistency probability distribution diagram with significant positive correlation between the forecast-observation correlation coefficient and the teleconnection coefficient.
  • Figure 6 shows the spatial consistency probability distribution diagram where the forecast-observation correlation coefficient is a significant positive correlation and the teleconnection coefficient is not significant.
  • Figure 7 shows the spatial consistency probability distribution diagram where the forecast-observation correlation is significantly positive and the teleconnection coefficient is significantly negative.
  • Figure 8 is an architectural diagram of a spatial probability analysis system based on the correspondence between precipitation forecast and teleconnection according to an embodiment of the present invention.
  • Figure 1 is a flow chart of the spatial probability analysis method based on the correspondence between precipitation forecast and teleconnection in this embodiment.
  • the spatial consistency probability that the correlation coefficient between forecast and observation is significantly positive is quantified, and it can be decomposed into spatial consistency probability with different corresponding relationships with teleconnection effects. , thus providing a reference for the evaluation and selection of precipitation forecast products.
  • the steps of calculating the forecast-observation correlation coefficient and the meteorological factor-observation precipitation teleconnection coefficient of each grid according to the obtained sample sequence include:
  • o k represents the observed precipitation data in the k-th year
  • f k represents the forecast precipitation data in the k-th year; represent the mean value of historical observed precipitation data and the mean value of historical forecast precipitation data respectively.
  • ⁇ k represents the meteorological factor index in the kth year, Indicates the average value of historical meteorological factor index.
  • the average value of historical observed precipitation data It is calculated based on the sample sequence of historical observed precipitation, and its expression is as follows:
  • K is the total number of years of the historical sample sequence.
  • the steps for classifying each grid include:
  • the probability density function of the correlation coefficient r is estimated using the Beta function:
  • r is the correlation coefficient
  • B is the Beta function
  • n is the number of forecast and observation samples used to calculate the correlation coefficient.
  • the cumulative probability density function corresponding to this probability density function is denoted as F.
  • ⁇ the 100 ⁇ (1- ⁇ /2) quantile of the correlation coefficient r is expressed as r 1- ⁇ /2 :
  • the 100 ⁇ ( ⁇ /2) quantile of the correlation coefficient r is expressed as r ⁇ /2 :
  • forecast-observation correlation coefficient r(o,f) is less than or equal to r 1- ⁇ /2 and greater than r ⁇ /2 , it is determined to be insignificant;
  • forecast-observation correlation coefficient r(o,f) is less than r ⁇ /2 , it is determined to be a significant negative correlation.
  • r(o, ⁇ ) If the teleconnection coefficient r(o, ⁇ ) is less than or equal to r 1- ⁇ /2 and greater than r ⁇ /2 , it is determined to be insignificant;
  • each grid is divided into significantly positive correlation (significantly positive, P), non-significant (non-significant, ns) and significantly negative correlation (significantly negative) according to the significance of its correlation coefficient. , N) three categories.
  • the forecast-observation correlation coefficient r(o,f) can be divided into three categories:
  • the steps of judging the corresponding relationship between the forecast-observation correlation coefficient and the teleconnection coefficient based on the grid classification results include: judging grid by grid whether the forecast-observation correlation is significantly positive and the meteorological factor-observation precipitation teleconnection is significantly positive. Correlation, the forecast-observation correlation is a significant positive correlation and the meteorological factor-observed precipitation teleconnection is not significant, or the forecast-observation correlation is a significant positive correlation and the meteorological factor-observed precipitation teleconnection is a significant negative correlation, and the correspondence is constructed through Boolean numbers Relationship vector.
  • the corresponding relationship between the forecast-observation correlation coefficient and the teleconnection coefficient is first judged when the forecast-observation correlation coefficient r(o,f) is significantly positive, and a corresponding relationship vector is constructed by combining Boolean numbers. Its expression is as follows:
  • N is the total number of grids in the target area.
  • b(P AC &P ENSO ) is a Boolean vector that simultaneously has a significant positive correlation between forecast and observation P AC and a significant positive correlation between meteorological factors and observed precipitation teleconnection P ENSO .
  • grid i satisfies P AC &P ENSO , then The value of x i is 1, otherwise the value of x i is 0.
  • b(P AC &ns ENSO ) represents a Boolean vector that simultaneously has a significant positive correlation P AC between the forecast-observation correlation and an insignificant ns ENSO meteorological factor-observed precipitation teleconnection.
  • b(P AC &N ENSO ) represents a Boolean vector that simultaneously has a significant positive correlation between forecast and observation P AC and a significant negative correlation between meteorological factors and observed precipitation teleconnection N ENSO .
  • grid i satisfies P AC &N ENSO , then The value of x i is 1, otherwise the value of x i is 0.
  • the spatial weight coefficient is calculated according to the spatial coordinates of the grid, and the step of obtaining the spatial weight coefficient matrix includes:
  • d ij is the Euclidean distance between the grid points (u i , vi ) and grid points (u j , v j ) of any grid i and grid j.
  • d is the weight coefficient bandwidth value.
  • the weight coefficient bandwidth value d is 5.
  • N is the total number of grids in the target area.
  • row normalization is performed on the spatial weight coefficient matrix A, and row normalization is performed on each spatial weight coefficient; its expression is as follows:
  • This embodiment performs row standardization on each weight coefficient to ensure that the sum of the weight coefficients of each row is equal to 1.
  • the steps of calculating the spatial consistency probability that the forecast-observation correlation coefficient is a significant positive correlation include:
  • b(P AC ) is a Boolean vector indicating that the forecast-observation correlation coefficient of the grid is significantly positive, and p i represents the spatial consistency probability that the forecast-observation correlation coefficient of grid i is significantly positive.
  • the Boolean number vector b(P AC ) whose forecast-observation correlation coefficient is a significant positive correlation includes the corresponding grid that simultaneously belongs to the Boolean where the forecast-observation correlation is a significant positive correlation and the meteorological factor-observation precipitation teleconnection is a significant positive correlation.
  • the forecast-observation correlation is a significant positive correlation and the meteorological factor-observed precipitation teleconnection is insignificant Boolean number vector b (P AC &ns ENSO ), and the forecast-observation correlation is a significant positive correlation
  • the meteorological factor-observed precipitation teleconnection is a Boolean vector b(P AC &N ENSO ) with significant negative correlation.
  • the spatial consistency probability P(P AC ) calculated by the forecast-observation correlation coefficient is a significant positive correlation, including that the corresponding grid belongs to both the forecast-observation correlation and the meteorological factor-observation precipitation teleconnection is significant.
  • P(P AC ) P(P AC &P ENSO )+P(P AC &ns ENSO )+P(P AC &N ENSO )
  • this embodiment can calculate the spatial consistency probability P (ns AC ) that the forecast-observation correlation coefficient is insignificant according to the spatial weight coefficient matrix A and the correspondence between the forecast-observation correlation coefficient and the teleconnection coefficient.
  • the forecast-observation correlation coefficient is the spatial consistency probability P(N AC ) with significant negative correlation, which further provides a reference for the evaluation and selection of precipitation forecast products.
  • Example 1 the spatial probability analysis method based on the correspondence between precipitation forecast and teleconnection proposed in Example 1 is tested.
  • the monthly scale grid precipitation data set CPC-URD (Climate Prediction Center global daily Unified Raingauge Database) of the National Oceanic and Atmosphere Administration (NOAA) from 1982 to 2010 is used.
  • the monthly precipitation data is the observation data, and the observed precipitation for three consecutive months is accumulated to obtain the seasonal precipitation;
  • the second-generation climate forecast system CFSv2 of the U.S. Centers for Environmental Prediction (NCEP) is used as the forecast precipitation data;
  • NCEP the second-generation climate forecast system
  • the 3.4 index represents the ENSO phenomenon.
  • CFSv2 forecast precipitation uses seasonal forecast precipitation with a 0-month forecast period. Take December-January-February (DJF) precipitation as an example. The spatial resolution of observed precipitation and forecast precipitation is both 1° ⁇ 1°.
  • Figure 4 shows the spatial consistency probability distribution diagram with a significant positive correlation between the forecast-observation correlation coefficient. It can be seen that the probability of a significantly positive correlation coefficient is higher in southern North America, northern and southeastern South America, eastern and southern Africa, northeastern Asia, southern my country, Southeast Asia, southern Australia, and Europe. It shows that precipitation forecast has better effect in these areas.
  • Figure 5 shows the spatial consistency probability distribution diagram in which the forecast-observation correlation coefficient of the grid is significantly positive and the teleconnection coefficient is significantly positive. It can be seen that the probability is higher in southern North America, southeastern South America, eastern Africa, central Asia and southern my country. This result reflects that in these areas, there is a strong correspondence between forecast precipitation and teleconnection intensity.
  • Figure 6 shows the spatial consistency probability when the forecast-observation correlation coefficient is significantly positive, but the teleconnection coefficient is not significant. This type of situation has a higher probability of spatial consistency in northern Eurasia, southern Australia, and northwestern Africa. This result reflects that forecasts still have good results in areas with weak teleconnection intensity.
  • Figure 7 shows the spatial consistency probability when the forecast-observation correlation coefficient is significantly positive and the teleconnection coefficient is significantly negative. Probability values are higher in northeastern South America, southern Africa, Southeast Asia, and northeastern Asia.
  • the spatial consistency probability shown in Figure 5, Figure 6 and Figure 7 decomposes the spatial consistency probability shown in Figure 4 into three parts to facilitate the analysis of different possible influencing factors on the forecast effect.
  • the above experimental results show that the spatial probability analysis method proposed by the present invention can effectively quantify the spatial consistency probability when the forecast-observation correlation coefficient is significantly positive, and at the same time decompose it into the spatial consistency probability corresponding to different teleconnection relationships, and can intuitively It shows the spatial distribution of different corresponding relationships, which can provide reference for the business use of forecasts.
  • This embodiment proposes a spatial probability analysis system based on the correspondence between precipitation forecast and teleconnection, and applies the spatial probability analysis method based on the correspondence between precipitation forecast and teleconnection proposed in Embodiment 1.
  • FIG 8 it is an architectural diagram of the spatial probability analysis system based on the correspondence between precipitation forecast and teleconnection in this embodiment.
  • Data acquisition module 1 is used to obtain the sample sequence of precipitation forecast to be analyzed, and the corresponding sample sequence of observed precipitation and meteorological factors.
  • Correlation coefficient calculation module 2 is used to calculate the forecast-observation correlation coefficient and meteorological factor-observation precipitation teleconnection coefficient of each grid in the target area based on the obtained sample sequence.
  • Classification module 3 is used to analyze the significance of forecast-observation correlation coefficients and meteorological factors-observation precipitation teleconnection coefficients, and classify each grid based on the analysis results.
  • the significance judgment module 4 is used to judge the correspondence between the forecast-observation correlation coefficient and the teleconnection coefficient based on the grid classification results.
  • the spatial weight coefficient calculation module 5 is used to calculate the spatial weight coefficient according to the spatial coordinates of the grid and obtain the spatial weight coefficient matrix.
  • the spatial consistency probability analysis module 6 is used to calculate the spatial consistency probability that the forecast-observation correlation coefficient is a significant positive correlation, and the forecast-observation correlation based on the spatial weight coefficient matrix and the corresponding relationship between the forecast-observation correlation coefficient and the teleconnection coefficient. The spatial consistency probability of the corresponding relationship between coefficients and different teleconnection coefficients.
  • the classification module 3 determines the significance of each grid in the target area based on the preset significance level ⁇ , as well as the forecast-observation correlation coefficient and meteorological factor-observation precipitation teleconnection coefficient of each grid. And classification:
  • forecast-observation correlation coefficient r(o,f) is less than or equal to r 1- ⁇ /2 and greater than r ⁇ /2 , it is determined to be insignificant;
  • r(o, ⁇ ) If the teleconnection coefficient r(o, ⁇ ) is less than or equal to r 1- ⁇ /2 and greater than r ⁇ /2 , it is determined to be insignificant;
  • the significance judgment module 4 determines whether the forecast-observation correlation is a significant positive correlation and the meteorological factor- The observed precipitation teleconnection is a significant positive correlation, the forecast-observation correlation is a significant positive correlation and the meteorological factor-observed precipitation teleconnection is not significant, or the forecast-observation correlation is a significant positive correlation and the meteorological factor-observed precipitation teleconnection is a significant negative correlation. , and construct the corresponding relationship vector through Boolean numbers.
  • the spatial weight coefficient calculation module 5 also includes row normalization for each spatial weight coefficient, the spatial weight coefficient matrix A.
  • the spatial consistency probability analysis module 6 calculates that the forecast-observation correlation coefficient of each grid is a significant positive correlation based on the corresponding relationship between the forecast-observation correlation coefficient and the teleconnection coefficient of each grid in the target area. Boolean vector, and then multiply the Boolean vector with the spatial weight coefficient corresponding to the grid to calculate the spatial consistency probability that the forecast-observation correlation coefficient of the corresponding grid is significantly positive.
  • b(P AC ) is a Boolean vector indicating that the forecast-observation correlation coefficient of the grid is significantly positive, and p i represents the spatial consistency probability that the forecast-observation correlation coefficient of grid i is significantly positive.
  • the Boolean number vector b(P AC ) in which the forecast-observation correlation coefficient of each grid is a significant positive correlation, including the corresponding grid that simultaneously belongs to the forecast-observation correlation is a significant positive correlation and the meteorological factor-observation precipitation teleconnection is significant
  • the Boolean vector b(P AC &N ENSO ) is a significant positive correlation and the meteorological factor-observed precipitation teleconnection is a significant negative correlation.

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Mathematical Physics (AREA)
  • Data Mining & Analysis (AREA)
  • Mathematical Optimization (AREA)
  • Computational Mathematics (AREA)
  • Pure & Applied Mathematics (AREA)
  • Mathematical Analysis (AREA)
  • Databases & Information Systems (AREA)
  • Software Systems (AREA)
  • Business, Economics & Management (AREA)
  • General Engineering & Computer Science (AREA)
  • Algebra (AREA)
  • Strategic Management (AREA)
  • Computing Systems (AREA)
  • Operations Research (AREA)
  • Human Resources & Organizations (AREA)
  • Economics (AREA)
  • Game Theory and Decision Science (AREA)
  • Probability & Statistics with Applications (AREA)
  • Development Economics (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Evolutionary Biology (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Marketing (AREA)
  • Quality & Reliability (AREA)
  • Tourism & Hospitality (AREA)
  • General Business, Economics & Management (AREA)
  • Image Analysis (AREA)

Abstract

一种基于降水预报与遥相关对应关系的空间概率分析方法及系统，涉及降水预报分析技术领域。其中方法包括以下步骤:获取待分析的降水预报的样本序列,以及对应的观测降水和气象因子的样本序列（S1）；根据获取的样本序列，分别计算各个网格的预报-观测相关系数和气象因子-观测降水遥相关系数，并根据预报-观测相关系数和气象因子-观测降水遥相关系数的显著性，对各个网格进行分类（S2）；根据网格分类结果判断预报-观测相关系数和遥相关系数的对应关系（S3）；根据网格的空间坐标计算空间权重系数，获得空间权重系数矩阵（S4）；根据空间权重系数矩阵以及预报-观测相关系数和遥相关系数的对应关系，计算预报-观测相关系数为显著正相关的空间一致性概率（S5）。

Description

一种基于降水预报与遥相关对应关系的空间概率分析方法及系统 技术领域
本发明涉及降水预报分析技术领域，更具体地，涉及一种基于降水预报与遥相关对应关系的空间概率分析方法及系统。
背景技术
科学准确的季节降水预报在防洪减灾、洪水资源化利用和水库调度等方面有重要的应用价值。发源于赤道中东太平洋的厄尔尼诺-南方涛动(El
Figure PCTCN2022099021-appb-000001
-SouthernOscillation，ENSO)现象对于全球季节降水具有重要的指示作用。大量的观测数据和分析研究表明，ENSO通过大气遥相关作用，可以对区域甚至全球降水产生重要影响。因此，一些国家和地区的业务预报中心，专门针对ENSO事件进行观测和预测，并以此为依据提供未来降水异常的统计预报。另一方面，全球气象模型(Global Climate Models，GCMs)在近年来得到稳步发展，提供了丰富的降水和气温等气象驱动数据。这些预报数据具有相当的精度和较长的预见期，被逐渐应用到业务降水预报中。
对ENSO和降水遥相关系数的认识和研究为全球季节降水的预报提供了重要的支持。多项分析研究表明，季节降水预报的可预报性信息的主要来源是ENSO信号，并指出ENSO和区域降水遥相关系数的强度和降水预报精度的对应性。因此，全球气象模型对ENSO-降水遥相关系数的捕捉刻画能力，为全球季节降水预报适用性评估提供了关键切入点。在实际的降水预报评估分析中，往往直接对比预报精度和遥相关强度空间分布的相似性，难以给出两者对应关系的定量描述。此外，相邻区域的相关系数通常并非独立，而是存在较强的关联关系，而传统的降水预报评估分析方法一般针对单个网格，忽略了变量的空间属性，导致无法准确评估降水预报的预报精度。
发明内容
本发明为克服上述现有的降水预报评估分析中难以给出预报精度和遥相关强度空间分布的定量描述，且忽略了变量的空间属性，导致无法准确评估降水预报精度的缺陷，提供一种基于降水预报与遥相关对应关系的空间概率分析方法及 系统。
为解决上述技术问题，本发明的技术方案如下：
一种基于降水预报与遥相关对应关系的空间概率分析方法，包括以下步骤：
获取待分析的降水预报的样本序列，以及对应的观测降水和气象因子的样本序列；
根据获取的样本序列，分别计算各个网格的预报-观测相关系数和气象因子-观测降水遥相关系数，并根据预报-观测相关系数和气象因子-观测降水遥相关系数的显著性，对各个网格进行分类；
根据网格分类结果判断预报-观测相关系数和遥相关系数的对应关系；
根据网格的空间坐标计算空间权重系数，获得空间权重系数矩阵；
根据空间权重系数矩阵以及预报-观测相关系数和遥相关系数的对应关系，计算预报-观测相关系数为显著正相关的空间一致性概率。
进一步地，本发明还提出了一种基于降水预报与遥相关对应关系的空间概率分析系统，应用上述提出的基于降水预报与遥相关对应关系的空间概率分析方法。其中包括数据采集模块、相关系数计算模块、分类模块、显著性判断模块、空间权重系数计算模块、和空间一致性概率分析模块。
本技术方案中，数据采集模块用于获取待分析的降水预报的样本序列，以及对应的观测降水和气象因子的样本序列；相关系数计算模块用于根据获取的样本序列，分别计算目标区域内各个网格的预报-观测相关系数和气象因子-观测降水遥相关系数；分类模块用于分析预报-观测相关系数和气象因子-观测降水遥相关系数的显著性，并根据分析结果对各个网格进行分类；显著性判断模块用于根据网格分类结果判断预报-观测相关系数和遥相关系数的对应关系；空间权重系数计算模块用于根据网格的空间坐标计算空间权重系数，获得空间权重系数矩阵；空间一致性概率分析模块用于根据空间权重系数矩阵以及预报-观测相关系数和遥相关系数的对应关系，计算预报-观测相关系数为显著正相关的空间一致性概率，以及预报-观测相关系数分别与不同遥相关系数的对应关系的空间一致性概率。
与现有技术相比，本发明技术方案的有益效果是：本发明通过将预报降水的空间关系和概率结合，量化了预报和观测相关系数显著为正的空间一致性概率，且可以将其分解为和遥相关作用不同对应关系的空间一致性概率，从而为降水预 报产品的评估和选择提供参考。
附图说明
图1为本发明实施例的基于降水预报与遥相关对应关系的空间概率分析方法的流程图。
图2为DJF的预报-观测相关系数显著性分类结果示意图。
图3为DJF的遥相关系数显著性分类结果示意图。
图4为预报-观测相关系数为显著正相关的空间一致性概率分布图。
图5为预报-观测相关系数和遥相关系数均为显著正相关的空间一致性概率分布图。
图6为预报-观测相关系数为显著正相关、遥相关系数不显著的空间一致性概率分布图。
图7为预报-观测相关系为显著正相关、遥相关系数为显著负相关的空间一致性概率分布图。
图8为本发明实施例的基于降水预报与遥相关对应关系的空间概率分析系统的架构图。
具体实施方式
附图仅用于示例性说明，不能理解为对本专利的限制；
对于本领域技术人员来说，附图中某些公知结构及其说明可能省略是可以理解的。
下面结合附图和实施例对本发明的技术方案做进一步的说明。
实施例1
本实施例提出一种基于降水预报与遥相关对应关系的空间概率分析方法，如图1所示，为本实施例的基于降水预报与遥相关对应关系的空间概率分析方法的流程图。
本实施例提出的基于降水预报与遥相关对应关系的空间概率分析方法中，包括以下步骤：
S1、获取待分析的降水预报的样本序列，以及对应的观测降水和气象因子的样本序列.
S2、根据获取的样本序列，分别计算各个网格的预报-观测相关系数和气象因子-观测降水遥相关系数，并根据预报-观测相关系数和气象因子-观测降水遥相 关系数的显著性，对各个网格进行分类。
S3、根据网格分类结果判断预报-观测相关系数和遥相关系数的对应关系。
S4、根据网格的空间坐标计算空间权重系数，获得空间权重系数矩阵。
S5、根据空间权重系数矩阵以及预报-观测相关系数和遥相关系数的对应关系，计算预报-观测相关系数为显著正相关的空间一致性概率。
本实施例中，通过将预报降水的空间关系和概率结合，量化了预报和观测相关系数显著为正的空间一致性概率，且可以将其分解为和遥相关作用不同对应关系的空间一致性概率，从而为降水预报产品的评估和选择提供参考。
在一可选实施例中，根据获取的样本序列，分别计算各个网格的预报-观测相关系数和气象因子-观测降水遥相关系数的步骤包括：
S2.1、根据获取的样本序列，提取目标区域网格的预报降水数据和观测降水数据。
S2.2、逐网格计算预报-观测相关系数r(o,f)；其表达式如下：
Figure PCTCN2022099021-appb-000002
式中，o k表示第k年的观测降水数据，f k表示第k年的预报降水数据；
Figure PCTCN2022099021-appb-000003
分别表示历史观测降水数据均值和历史预报降水数据均值。
S2.3、逐网格计算气象因子-观测降水遥相关系数r(o,η)；其表达式如下：
Figure PCTCN2022099021-appb-000004
式中，η k表示第k年的气象因子指数，
Figure PCTCN2022099021-appb-000005
表示历史气象因子指数均值。
其中，历史观测降水数据均值
Figure PCTCN2022099021-appb-000006
根据历史观测降水的样本序列计算得到，其表达式如下：
Figure PCTCN2022099021-appb-000007
式中，K为历史样本序列的总年数。
同理，历史降水预报数据均值
Figure PCTCN2022099021-appb-000008
根据历史降水预报的样本序列计算得到，其 表达式如下：
Figure PCTCN2022099021-appb-000009
本实施例中，对于厄尔尼诺-南方涛动(El
Figure PCTCN2022099021-appb-000010
-Southern Oscillation)遥相关而言，常用气象因子为
Figure PCTCN2022099021-appb-000011
3.4指数。
进一步地，根据预报-观测相关系数和气象因子-观测降水遥相关系数的显著性，对各个网格进行分类的步骤包括：
S2.4、根据预设的显著性水平α，以及各个网格的预报-观测相关系数和气象因子-观测降水遥相关系数，对目标区域各个网格进行显著性判断及分类。
其中相关系数r的概率密度函数利用Beta函数进行估计：
Figure PCTCN2022099021-appb-000012
式中，r即相关系数，B为Beta函数，n为用于计算相关系数的预报和观测样本数。该概率密度函数对应的累积概率密度函数记为F。则显著性水平α下，相关系数r的100×(1-α/2)分位数表示为r 1-α/2：
r 1-α/2＝F -1(1-α/2)
相关系数r的100×(α/2)分位数则表示为r α/2：
r α/2＝F -1(α/2)。
则对于预报-观测相关系数r(o,f)：
若预报-观测相关系数r(o,f)大于r 1-α/2，则判定为显著正相关；
若预报-观测相关系数r(o,f)小于或等于r 1-α/2，且大于r α/2，则判定为不显著；
若预报-观测相关系数r(o,f)小于r α/2，则判定为显著负相关。
对于气象因子-观测降水遥相关系数r(o,η)：
若遥相关系数r(o,η)大于r 1-α/2，则判定为显著正相关；
若遥相关系数r(o,η)小于或等于r 1-α/2，且大于r α/2，则判定为不显著；
若遥相关系数r(o,η)小于r α/2，则判定为显著负相关。
其中，在给定显著性水平α下，各网格根据其相关系数的显著性，分为显著正相关(significantly positive，P)，不显著(non-significant，ns)以及显著负相 关(significantly negative，N)三种类别。
即对于预报-观测相关系数r(o,f)可以分为三类：
Figure PCTCN2022099021-appb-000013
对于气象因子-观测降水遥相关系数r(o,η)可以分为三类：
Figure PCTCN2022099021-appb-000014
进一步地，根据网格分类结果判断预报-观测相关系数和遥相关系数的对应关系的步骤包括：逐个网格判断同时属于预报-观测相关为显著正相关且气象因子-观测降水遥相关为显著正相关、预报-观测相关为显著正相关且气象因子-观测降水遥相关为不显著，或预报-观测相关为显著正相关且气象因子-观测降水遥相关为显著负相关，并通过布尔数构建对应关系向量。
本实施例首先针对预报-观测相关系数r(o,f)为显著正相关的情况对预报-观测相关系数和遥相关系数的对应关系进行判断，并结合布尔数构建对应关系向量。其表达式如下：
b(P AC&P ENSO)＝[x i] N×1
b(P AC&ns ENSO)＝[x i] N×1
b(P AC&N ENSO)＝[x i] N×1
式中，N为目标区域网格的总数量。
b(P AC&P ENSO)为同时属于预报-观测相关为显著正相关P AC且气象因子-观测降水遥相关为显著正相关P ENSO的布尔数向量，当网格i满足P AC&P ENSO，则x i取值为1，否则x i取值为0。
b(P AC&ns ENSO)表示同时属于预报-观测相关为显著正相关P AC且气象因子-观测降水遥相关为不显著ns ENSO的布尔数向量，当网格i满足P AC&ns ENSO，则x i取值为1，否则x i取值为0。
b(P AC&N ENSO)表示同时属于预报-观测相关为显著正相关P AC且气象因子-观测降水遥相关为显著负相关N ENSO的布尔数向量，当网格i满足P AC&N ENSO，则x i取值为1，否则x i取值为0。
在一可选实施例中，根据网格的空间坐标计算空间权重系数，获得空间权重系数矩阵的步骤包括：
S4.1、以目标区域左上角作为原点对目标区域网格进行坐标标记。
S4.2、通过距离的二次衰减函数对任意两个网格坐标进行空间权重系数计算，其表达式如下：
Figure PCTCN2022099021-appb-000015
Figure PCTCN2022099021-appb-000016
Figure PCTCN2022099021-appb-000017
式中，d ij为任意网格i和网格j的网格点(u i,v i)和网格点(u j,v j)之间的欧氏距离。d为权重系数带宽值，可选地，权重系数带宽值d取值为5。
S4.3、根据任意两个网格坐标的空间权重系数构建空间权重系数矩阵，其表达式如下：
W＝[w ij] N×N
式中，N为目标区域网格的总数量。
进一步地，对空间权重系数矩阵A，进行行标准化处理，对每个空间权重系数取行标准化；其表达式如下：
Figure PCTCN2022099021-appb-000018
本实施例通过对每个权重系数取行标准化，以保证每一行的权重系数加和等于1。
在一可选实施例中，根据空间权重系数矩阵A以及预报-观测相关系数和遥相关系数的对应关系，计算预报-观测相关系数为显著正相关的空间一致性概率的步骤包括：
根据目标区域内各个网格的预报-观测相关系数和遥相关系数的对应关系，计算各个网格的预报-观测相关系数为显著正相关的布尔数向量，然后将布尔数 向量与该网格对应的空间权重系数相乘，计算得到相应网格的预报-观测相关系数为显著正相关的空间一致性概率。其表达式如下：
P(P AC)＝A·b(P AC)＝[p i] N×1
式中，b(P AC)为网格的预报-观测相关系数为显著正相关的布尔数向量，p i表示网格i的预报-观测相关系数为显著正相关的空间一致性概率。
其中，预报-观测相关系数为显著正相关的布尔数向量b(P AC)中，包括相应网格同时属于预报-观测相关为显著正相关且气象因子-观测降水遥相关为显著正相关的布尔数向量b(P AC&P ENSO)、预报-观测相关为显著正相关且气象因子-观测降水遥相关为不显著的布尔数向量b(P AC&ns ENSO)，和预报-观测相关为显著正相关且气象因子-观测降水遥相关为显著负相关的布尔数向量b(P AC&N ENSO)。
由此计算得到的预报-观测相关系数为显著正相关的空间一致性概率P(P AC)中，包括相应网格同时属于预报-观测相关为显著正相关且气象因子-观测降水遥相关为显著正相关的空间一致性概率P(P AC&P ENSO)，预报-观测相关为显著正相关且气象因子-观测降水遥相关为不显著的空间一致性概率P(P AC&ns ENSO)，和预报-观测相关为显著正相关且气象因子-观测降水遥相关为显著负相关的空间一致性概率P(P AC&N ENSO)。其表达式如下：
P(P AC)＝P(P AC&P ENSO)+P(P AC&ns ENSO)+P(P AC&N ENSO)
＝A·b(P AC&P ENSO)+A·b(P AC&ns ENSO)+A·b(P AC&N ENSO)
＝[p i] N×1
可选地，同理本实施例可根据空间权重系数矩阵A以及预报-观测相关系数和遥相关系数的对应关系，计算预报-观测相关系数为不显著的空间一致性概率P(ns AC)以及预报-观测相关系数为显著负相关的空间一致性概率P(N AC)，进一步为降水预报产品的评估和选择提供参考。
实施例2
本实施例将实施例1提出的基于降水预报与遥相关对应关系的空间概率分析方法进行试验。
本实施例中，以美国国家海洋和大气管理局(National Oceanic and Atmosphere Administration,NOAA)气候预报中心的月尺度网格降水数据集CPC-URD(Climate Prediction Center global daily Unified Raingauge Database)1982-2010年的月降水数据为观测数据，将连续三个月的观测降水累加得到季节 降水；以美国环境预测中心(NCEP)的第二代气候预报系统CFSv2为预报降水数据；以
Figure PCTCN2022099021-appb-000019
3.4指数表示ENSO现象。
其中，CFSv2预报降水采用0个月预见期的季节预报降水。以12月-1月-2月(December-January-February，DJF)降水为例。观测降水和预报降水空间分辨率均为1°×1°。
如图2所示，为DJF的预报-观测相关系数显著性分类结果。其中，灰度中等的网格表示相关系数显著为正，表示预报对观测降水具有一定的指示作用。灰度较浅的网格表示相关系数不显著，灰度较深的网格则表示预报和观测的相关系数显著为负，这两类网格表明预报效果不够理想。可以看到，大部分的网格为预报和观测的相关系数显著为负的，其次是相关系数显著为正。
如图2所示，为DJF的遥相关系数显著性分类结果。其中，灰度中等的网格表示相关系数显著为正，灰度较浅的网格表示相关系数不显著，灰度较深的网格则表示预报和观测的相关系数显著为负。相比于预报-观测相关系数，可以看到遥相关系数的显著性分类结果在全球的分布更加连续。对比图2和图3，可以看到部分地区显著系数分布的相似性。例如，在北美洲南部、南美洲北部、非洲东部等地，预报-观测相关系数显著为正，同时遥相关强度较大。
图2和图3的空间分布图给出了预报效果较好的区域分布，但是无法反映这些网格在空间上的集聚程度。同时，图2和图3的对应性无法进一步给出对应程度的强度以及这种对应性空间上的一致性。
在图2的基础之上，首先计算预报-观测相关系数显著为正的空间一致性概率。对每个网格及以其为中心的5°范围内的所有网格统计相关系数分类，根据空间权重系数矩阵，得到显著为正网格的空间一致性概率。
图4为预报-观测相关系数为显著正相关的空间一致性概率分布图。可以看到相关系数显著为正的情况在北美洲南部、南美洲北部和东南部、非洲东部和南部、亚洲东北部、我国华南地区、东南亚、澳大利南部以及欧洲概率较高。说明降水预报在这些区域有较好的效果。
进一步的，建立起图2和图3的对应关系。具体的，图5中给出了网格的预报-观测相关系数显著为正，同时遥相关系数显著为正的空间一致性概率分布图。可以看到，在北美洲南部、南美洲东南部、非洲东部、亚洲中部和我国华南地区的概率较高。这个结果反映了，在这些地区，预报降水和遥相关强度具有较强的 对应关系。图6则给出了预报-观测相关系数显著为正，而遥相关系数不显著的空间一致性概率。这一类情况在欧亚大陆北部、澳大利亚南部和非洲西北部的空间一致性概率较高，这个结果反映了在遥相关强度较弱的地区，预报依然具有较好的效果。图7给出了预报-观测相关系数显著为正，同时遥相关系数显著为负的空间一致性概率。在南美洲东北部、非洲南部、东南亚和亚洲东北部，概率数值较高。
图5、图6和图7的空间一致性概率将图4所示的空间一致性概率分解为三个部分，便于分析预报效果可能的不同影响因素。以上实验结果表明，本发明提出的空间概率分析方法，能够有效量化出预报-观测相关系数显著为正的空间一致性概率，同时分解为与不同遥相关关系对应关系的空间一致性概率，能够直观的展示出不同的对应关系空间分布情况，能够为预报的业务使用提供参考。
实施例3
本实施例提出一种基于降水预报与遥相关对应关系的空间概率分析系统，应用实施例1提出的基于降水预报与遥相关对应关系的空间概率分析方法。
如图8所示，为本实施例的基于降水预报与遥相关对应关系的空间概率分析系统的架构图。
本实施例提出的基于降水预报与遥相关对应关系的空间概率分析系统中，包括：
数据采集模块1，用于获取待分析的降水预报的样本序列，以及对应的观测降水和气象因子的样本序列。
相关系数计算模块2，用于根据获取的样本序列，分别计算目标区域内各个网格的预报-观测相关系数和气象因子-观测降水遥相关系数。
分类模块3，用于分析预报-观测相关系数和气象因子-观测降水遥相关系数的显著性，并根据分析结果对各个网格进行分类。
显著性判断模块4，用于根据网格分类结果判断预报-观测相关系数和遥相关系数的对应关系。
空间权重系数计算模块5，用于根据网格的空间坐标计算空间权重系数，获得空间权重系数矩阵。
空间一致性概率分析模块6，用于根据空间权重系数矩阵以及预报-观测相关系数和遥相关系数的对应关系，计算预报-观测相关系数为显著正相关的空间一 致性概率，以及预报-观测相关系数分别与不同遥相关系数的对应关系的空间一致性概率。
在一可选实施例中，分类模块3根据预设的显著性水平α，以及各个网格的预报-观测相关系数和气象因子-观测降水遥相关系数，对目标区域各个网格进行显著性判断及分类：
对于预报-观测相关系数r(o,f)：
若预报-观测相关系数r(o,f)大于r 1-α/2，则判定为显著正相关；
若预报-观测相关系数r(o,f)小于或等于r 1-α/2，且大于r α/2，则判定为不显著；
若预报-观测相关系数r(o,f)小于r α/2，则判定为显著负相关；
对于气象因子-观测降水遥相关系数r(o,η)：
若遥相关系数r(o,η)大于r 1-α/2，则判定为显著正相关；
若遥相关系数r(o,η)小于或等于r 1-α/2，且大于r α/2，则判定为不显著；
若遥相关系数r(o,η)小于r α/2，则判定为显著负相关。
在一可选实施例中，显著性判断模块4针对预报-观测相关系数r(o,f)为显著正相关的情况，逐个网格判断同时属于预报-观测相关为显著正相关且气象因子-观测降水遥相关为显著正相关、预报-观测相关为显著正相关且气象因子-观测降水遥相关为不显著，或预报-观测相关为显著正相关且气象因子-观测降水遥相关为显著负相关，并通过布尔数构建对应关系向量。
在一可选实施例中，空间权重系数计算模块5还包括对每个空间权重系数取行标准化，空间权重系数矩阵A。
在一可选实施例中，空间一致性概率分析模块6根据目标区域内各个网格的预报-观测相关系数和遥相关系数的对应关系，计算各个网格的预报-观测相关系数为显著正相关的布尔数向量，然后将布尔数向量与该网格对应的空间权重系数相乘，计算得到相应网格的预报-观测相关系数为显著正相关的空间一致性概率。
其表达式如下：
P(P AC)＝A·b(P AC)＝[p i] N×1
式中，b(P AC)为网格的预报-观测相关系数为显著正相关的布尔数向量，p i表示网格i的预报-观测相关系数为显著正相关的空间一致性概率。
其中，各个网格的预报-观测相关系数为显著正相关的布尔数向量b(P AC)中，包括相应网格同时属于预报-观测相关为显著正相关且气象因子-观测降水遥相 关为显著正相关的布尔数向量b(P AC&P ENSO)、预报-观测相关为显著正相关且气象因子-观测降水遥相关为不显著的布尔数向量b(P AC&ns ENSO)，和预报-观测相关为显著正相关且气象因子-观测降水遥相关为显著负相关的布尔数向量b(P AC&N ENSO)。
显然，本发明的上述实施例仅仅是为清楚地说明本发明所作的举例，而并非是对本发明的实施方式的限定。对于所属领域的普通技术人员来说，在上述说明的基础上还可以做出其它不同形式的变化或变动。这里无需也无法对所有的实施方式予以穷举。凡在本发明的精神和原则之内所作的任何修改、等同替换和改进等，均应包含在本发明权利要求的保护范围之内。

Claims (10)

  1. 一种基于降水预报与遥相关对应关系的空间概率分析方法，其特征在于，包括以下步骤：
    获取待分析的降水预报的样本序列，以及对应的观测降水和气象因子的样本序列；
    根据获取的样本序列，分别计算各个网格的预报-观测相关系数和气象因子-观测降水遥相关系数，并根据预报-观测相关系数和气象因子-观测降水遥相关系数的显著性，对各个网格进行分类；
    根据网格分类结果判断预报-观测相关系数和遥相关系数的对应关系；
    根据网格的空间坐标计算空间权重系数，获得空间权重系数矩阵；
    根据空间权重系数矩阵以及预报-观测相关系数和遥相关系数的对应关系，计算预报-观测相关系数为显著正相关的空间一致性概率。
  2. 根据权利要求1所述的基于降水预报与遥相关对应关系的空间概率分析方法，其特征在于，根据获取的样本序列，分别计算各个网格的预报-观测相关系数和气象因子-观测降水遥相关系数的步骤包括：
    根据获取的样本序列，提取目标区域网格的预报降水数据和观测降水数据；
    逐网格计算预报-观测相关系数r(o,f)；其表达式如下：
    Figure PCTCN2022099021-appb-100001
    式中，o k表示第k年的观测降水数据，f k表示第k年的预报降水数据；
    Figure PCTCN2022099021-appb-100002
    分别表示历史观测降水数据均值和历史预报降水数据均值；
    逐网格计算气象因子-观测降水遥相关系数r(o,η)；其表达式如下：
    Figure PCTCN2022099021-appb-100003
    式中，η k表示第k年的气象因子指数，
    Figure PCTCN2022099021-appb-100004
    表示历史气象因子指数均值。
  3. 根据权利要求1所述的基于降水预报与遥相关对应关系的空间概率分析方法，其特征在于，根据预报-观测相关系数和气象因子-观测降水遥相关系数的 显著性，对各个网格进行分类的步骤包括：
    根据预设的显著性水平α，以及各个网格的预报-观测相关系数和气象因子-观测降水遥相关系数，对目标区域各个网格进行显著性判断及分类：
    对于预报-观测相关系数r(o,f)：
    若预报-观测相关系数r(o,f)大于r 1-α/2，则判定为显著正相关；
    若预报-观测相关系数r(o,f)小于或等于r 1-α/2，且大于r α/2，则判定为不显著；
    若预报-观测相关系数r(o,f)小于r α/2，则判定为显著负相关；
    对于气象因子-观测降水遥相关系数r(o,η)：
    若遥相关系数r(o,η)大于r 1-α/2，则判定为显著正相关；
    若遥相关系数r(o,η)小于或等于r 1-α/2，且大于r α/2，则判定为不显著；
    若遥相关系数r(o,η)小于r α/2，则判定为显著负相关；
    其中，r 1-α/2为相关系数r的100×(1-α/2)分位数，r α/2为相关系数r的100×(α/2)分位数。
  4. 根据权利要求3所述的基于降水预报与遥相关对应关系的空间概率分析方法，其特征在于，根据网格分类结果判断预报-观测相关系数和遥相关系数的对应关系的步骤包括：
    针对预报-观测相关系数r(o,f)为显著正相关的情况，逐个网格判断同时属于预报-观测相关为显著正相关且气象因子-观测降水遥相关为显著正相关、预报-观测相关为显著正相关且气象因子-观测降水遥相关为不显著，或预报-观测相关为显著正相关且气象因子-观测降水遥相关为显著负相关，并通过布尔数构建对应关系向量；其表达式如下：
    b(P AC&P ENSO)＝[x i] N×1
    b(P AC&ns ENSO)＝[x i] N×1
    b(P AC&N ENSO)＝[x i] N×1
    式中，N为目标区域网格的总数量；b(P AC&P ENSO)为同时属于预报-观测相关为显著正相关P AC且气象因子-观测降水遥相关为显著正相关P ENSO的布尔数向量，当网格i满足P AC&P ENSO，则x i取值为1，否则x i取值为0；
    b(P AC&ns ENSO)表示同时属于预报-观测相关为显著正相关P AC且气象因子-观测降水遥相关为不显著ns ENSO的布尔数向量，当网格i满足P AC&ns ENSO，则x i取值为1，否则x i取值为0；
    b(P AC&N ENSO)表示同时属于预报-观测相关为显著正相关P AC且气象因子-观测降水遥相关为显著负相关N ENSO的布尔数向量，当网格i满足P AC&N ENSO，则x i取值为1，否则x i取值为0。
  5. 根据权利要求1所述的基于降水预报与遥相关对应关系的空间概率分析方法，其特征在于，根据网格的空间坐标计算空间权重系数，获得空间权重系数矩阵的步骤包括：
    以目标区域左上角作为原点对目标区域网格进行坐标标记；
    通过距离的二次衰减函数对任意两个网格坐标进行空间权重系数计算，其表达式如下：
    Figure PCTCN2022099021-appb-100005
    Figure PCTCN2022099021-appb-100006
    Figure PCTCN2022099021-appb-100007
    式中，d ij为网格点(u i,v i)和网格点(u j,v j)之间的欧氏距离，d为权重系数带宽值；
    根据任意两个网格坐标的空间权重系数构建空间权重系数矩阵，其表达式如下：
    W＝[w ij] N×N
    式中，N为网格总数。
  6. 根据权利要求5所述的基于降水预报与遥相关对应关系的空间概率分析方法，其特征在于，还包括以下步骤：对空间权重系数矩阵A，进行行标准化处理，对每个空间权重系数取行标准化；其表达式如下：
    Figure PCTCN2022099021-appb-100008
  7. 根据权利要求1～6任一项所述的基于降水预报与遥相关对应关系的空间概率分析方法，其特征在于，根据空间权重系数矩阵A以及预报-观测相关系数 和遥相关系数的对应关系，计算预报-观测相关系数为显著正相关的空间一致性概率的步骤包括：
    根据目标区域内各个网格的预报-观测相关系数和遥相关系数的对应关系，计算各个网格的预报-观测相关系数为显著正相关的布尔数向量，然后将布尔数向量与该网格对应的空间权重系数相乘，计算得到相应网格的预报-观测相关系数为显著正相关的空间一致性概率；其表达式如下：
    P(P AC)＝A·b(P AC)＝[p i] N×1
    式中，b(P AC)为网格的预报-观测相关系数为显著正相关的布尔数向量，p i表示网格i的预报-观测相关系数为显著正相关的空间一致性概率。
  8. 根据权利要求7所述的基于降水预报与遥相关对应关系的空间概率分析方法，其特征在于，各个网格的预报-观测相关系数为显著正相关的布尔数向量b(P AC)中，包括相应网格同时属于预报-观测相关为显著正相关且气象因子-观测降水遥相关为显著正相关的布尔数向量b(P AC&P ENSO)、预报-观测相关为显著正相关且气象因子-观测降水遥相关为不显著的布尔数向量b(P AC&ns ENSO)，和预报-观测相关为显著正相关且气象因子-观测降水遥相关为显著负相关的布尔数向量b(P AC&N ENSO)。
  9. 根据权利要求7所述的基于降水预报与遥相关对应关系的空间概率分析方法，其特征在于，还包括以下步骤：根据空间权重系数矩阵A以及预报-观测相关系数和遥相关系数的对应关系，计算预报-观测相关系数为不显著以及预报-观测相关系数为显著负相关的空间一致性概率。
  10. 一种基于降水预报与遥相关对应关系的空间概率分析系统，应用权利要求1～9任一项所述的基于降水预报与遥相关对应关系的空间概率分析方法，其特征在于，包括：
    数据采集模块，用于获取待分析的降水预报的样本序列，以及对应的观测降水和气象因子的样本序列；
    相关系数计算模块，用于根据获取的样本序列，分别计算目标区域内各个网格的预报-观测相关系数和气象因子-观测降水遥相关系数；
    分类模块，用于分析预报-观测相关系数和气象因子-观测降水遥相关系数的显著性，并根据分析结果对各个网格进行分类；
    显著性判断模块，用于根据网格分类结果判断预报-观测相关系数和遥相关 系数的对应关系；
    空间权重系数计算模块，用于根据网格的空间坐标计算空间权重系数，获得空间权重系数矩阵；
    空间一致性概率分析模块，用于根据空间权重系数矩阵以及预报-观测相关系数和遥相关系数的对应关系，计算预报-观测相关系数为显著正相关的空间一致性概率，以及预报-观测相关系数分别与不同遥相关系数的对应关系的空间一致性概率。
PCT/CN2022/099021 2022-06-15 2022-06-15 一种基于降水预报与遥相关对应关系的空间概率分析方法及系统 WO2023240509A1 (zh)

Priority Applications (1)

Application Number Priority Date Filing Date Title
PCT/CN2022/099021 WO2023240509A1 (zh) 2022-06-15 2022-06-15 一种基于降水预报与遥相关对应关系的空间概率分析方法及系统

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
PCT/CN2022/099021 WO2023240509A1 (zh) 2022-06-15 2022-06-15 一种基于降水预报与遥相关对应关系的空间概率分析方法及系统

Publications (1)

Publication Number Publication Date
WO2023240509A1 true WO2023240509A1 (zh) 2023-12-21

Family

ID=89192750

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2022/099021 WO2023240509A1 (zh) 2022-06-15 2022-06-15 一种基于降水预报与遥相关对应关系的空间概率分析方法及系统

Country Status (1)

Country Link
WO (1) WO2023240509A1 (zh)

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20180188417A1 (en) * 2017-01-03 2018-07-05 UBIMET GmbH Method for increasing the spatial resolution of a weather forecast
CN110472762A (zh) * 2018-05-11 2019-11-19 中国电力科学研究院有限公司 一种网格化降水的订正方法及系统
CN111797131A (zh) * 2020-06-09 2020-10-20 武汉大学 一种基于遥感降水产品的极端降水区域频率分析方法
CN113219555A (zh) * 2020-09-30 2021-08-06 国家气候中心 一种基于统计降尺度技术的短期气候预测方法
CN113496104A (zh) * 2021-07-16 2021-10-12 中科技术物理苏州研究院 基于深度学习的降水预报订正方法及系统
WO2022011728A1 (zh) * 2020-07-16 2022-01-20 中山大学 一种面向遥相关模式的空间自相关聚类方法

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20180188417A1 (en) * 2017-01-03 2018-07-05 UBIMET GmbH Method for increasing the spatial resolution of a weather forecast
CN110472762A (zh) * 2018-05-11 2019-11-19 中国电力科学研究院有限公司 一种网格化降水的订正方法及系统
CN111797131A (zh) * 2020-06-09 2020-10-20 武汉大学 一种基于遥感降水产品的极端降水区域频率分析方法
WO2022011728A1 (zh) * 2020-07-16 2022-01-20 中山大学 一种面向遥相关模式的空间自相关聚类方法
CN113219555A (zh) * 2020-09-30 2021-08-06 国家气候中心 一种基于统计降尺度技术的短期气候预测方法
CN113496104A (zh) * 2021-07-16 2021-10-12 中科技术物理苏州研究院 基于深度学习的降水预报订正方法及系统

Similar Documents

Publication Publication Date Title
Wang et al. A new copula-based standardized precipitation evapotranspiration streamflow index for drought monitoring
CN110459329B (zh) 一种登革热风险综合评估方法
Nourani et al. Estimation of prediction interval in ANN-based multi-GCMs downscaling of hydro-climatologic parameters
Han et al. Drought forecasting based on the remote sensing data using ARIMA models
CN111709646A (zh) 空气污染暴露风险评价方法及系统
CN113836808A (zh) 一种基于重污染特征约束的pm2.5深度学习预测方法
KR101440932B1 (ko) 실시간 앙상블 가뭄전망정보 관리방법
CN115575601A (zh) 基于水汽通量散度的植被旱灾指数评估方法和系统
Galán et al. Airborne fungal spore monitoring: between analyst proficiency testing
CN116310858A (zh) 一种基于多源遥感数据的高寒高海拔地表水体自动识别监测方法
CN115598027A (zh) 一种基于遥感和机器学习技术的pm2.5反演方法
CN113095579B (zh) 一种耦合伯努利-伽马-高斯分布的日尺度降水预报校正方法
Gao et al. Leveraging high-resolution weather information to predict hail damage claims: A spatial point process for replicated point patterns
Pei et al. Analysis of spring drought in Northeast China from the perspective of atmosphere, snow cover, and soil
WO2023240509A1 (zh) 一种基于降水预报与遥相关对应关系的空间概率分析方法及系统
CN115292667A (zh) 一种基于降水预报与遥相关对应关系的空间概率分析方法及系统
Miller et al. A preliminary assessment of using spatiotemporal lightning patterns for a binary classification of thunderstorm mode
CN111507514A (zh) 大气气溶胶数据预测方法
CN115510945B (zh) 基于主成分和Logistic分析地质灾害概率预报方法
CN107066786A (zh) 基于神经网络的气溶胶光学厚度反演算法
CN114707079B (zh) 一种登革热蚊媒适生区及主要影响因素的识别方法和系统
CN116739131A (zh) 一种基于智能网格预报产品的农业干旱预测方法及系统
CN118624540B (zh) 一种基于遥感气象数据协同的病虫害监测方法
CN117973703B (zh) 一种森林生态环境的分级定损方法及系统
CN112632407B (zh) 一种顾及地理环境异质性的空间抽样方法

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 22946203

Country of ref document: EP

Kind code of ref document: A1