A Cluster Analysis Method for Diagnosing the Composite Trend Characteristics and Intensity of Climate Change Based on Multi-Source Data

Abstract

This invention discloses a cluster analysis method for diagnosing the composite trend characteristics and intensity of climate change based on multi-source data. The method includes: selecting a study area, identifying the study objects, and acquiring multi-source data using big data technology; analyzing the changing trends of the study objects from each data source using the Sen slope estimation method; evaluating the reliability of the changing trends of the study objects from each data source using the Mann-Kendall test; selecting composite changing trend evaluation indicators for the study objects based on whether they pass the reliability test; distinguishing the changing trend characteristics of the study objects from each data source under different conditions, classifying and summarizing the composite changing trend characteristics of the study objects, calculating the ratio of composite changing trend characteristics of the study objects, and diagnosing the composite changing trend characteristics and intensity of the study objects. This method is based on cluster analysis theory and uses multiple data sources to determine the composite changing trends of hydrological and meteorological elements under specific conditions. It can be directly applied to the composite diagnosis of climate change in the hydrological and meteorological field.

Description

Technical Field

[0001] This invention belongs to the field of climate change assessment in hydrometeorology, specifically involving a cluster analysis method for diagnosing the composite trend characteristics and intensity of climate change from multiple sources. Background Technology

[0002] The spatiotemporal evolution of hydrometeorological elements is a hot topic in climate change research. For regions with no or insufficient data, due to harsh climate conditions, complex topography, and a scarcity of existing hydrometeorological observation stations, long-term, large-scale hydrometeorological data are scarce, making it difficult to accurately diagnose the spatiotemporal evolution characteristics of climate change in these areas. Numerical simulations suffer from uncertainty due to the lack of observational data validation. However, the development of satellite remote sensing technology and the improvement of numerical simulation techniques have provided conditions for obtaining long-term hydrometeorological data in data-scarce areas. Sen slope estimation is a non-parametric trend calculation method that is efficient and robust to noise. The Mann-Kendall test is a non-parametric time series trend test method that is unaffected by outliers and does not require the data to follow a normal distribution. Currently, Sen slope estimation and the Mann-Kendall test are widely used in the analysis of hydrometeorological element trends. However, how to fully utilize big data resources to extract the composite characteristics and intensity of the spatiotemporal evolution of climate elements from multiple data sources remains unresearched. Summary of the Invention

[0003] Purpose of the invention: In order to effectively assess climate change trends and address the problems of insufficient existing observational data and inconsistent climate change trends among various data sources, this invention proposes a cluster analysis method for diagnosing the composite trend characteristics and intensity of climate change from multiple data sources.

[0004] Technical Solution: To achieve the objectives of this invention, the technical solution adopted is: a multi-source data-based cluster analysis method for diagnosing the composite trend characteristics and intensity of climate change, comprising the following steps:

[0005] Step 1: Determine the study area, select hydrological and meteorological elements that represent regional climate change as the research object, and obtain the data source for the selected research object;

[0006] Step 2: Use the Sen slope estimation method to obtain the changing trend of the research object at each grid point of each data source;

[0007] Step 3: Use the Mann-Kendall test to assess the reliability of the changing trends of the research objects at each grid point of each data source;

[0008] Step 4: Evaluate the composite trend of selected research subjects based on whether they pass the credibility test;

[0009] Step 5: Distinguish the changing trend characteristics of the research objects in different data sources under different conditions;

[0010] Step 6: Classify and summarize the composite change trend characteristics of the research objects at each grid point;

[0011] Step 7: Calculate the ratio of composite trend characteristics of each grid point of the research object using cluster analysis.

[0012] Step 8: Calculate the composite change trend characteristics and intensity index of each grid point research object using cluster analysis;

[0013] Step 9: Diagnose the climate change characteristics of the study area based on the composite change trend characteristics and intensity indicators of the research subjects.

[0014] Furthermore, step 2, which uses the Sen slope estimation method to analyze the changing trends of the research objects at each grid point of each data source, specifically includes:

[0015] Using formula (1), the Sen slope member with a fixed time step is obtained for the research object in the research period of the p-th grid point of the q-th data source;

[0016] Using formula (2), the median function in MATLAB is used to obtain the Sen slope of the research object at a fixed time step during the research period of the p-th grid point of the q-th data source, i.e. the trend of change;

[0017]

[0018] β(X p,q ) = median(X k,p,q (2)

[0019] Among them, X k,p,q X is the k-th Sen slope member with a fixed time step within the research period of the p-th grid point of the q-th data source, X. j,p,q and X i,p,q Let β(X) represent the values ​​of the research object X at time j and time i, respectively, at the p-th grid point of the q-th data source, where j>i, n is the total time series length, and N is the total number of Sen slope members at each grid point; p,q Let be the Sen slope of the research object X at the p-th grid point of the q-th data source;

[0020] β(X p,q β(X) > 0 indicates that the number of research subjects showed a positive growth trend during the research period, and vice versa; β(X) > 0 indicates a negative growth trend. p,q The magnitude of the value represents the extent of the growth, that is, the strength of the trend of change.

[0021] Furthermore, in step 3, the Mann-Kendall method is used to obtain the reliability index Z of the trend of the research object X at the p-th grid point of the q-th data source. S (X p,q Specifically, this includes:

[0022] The cumulative trend statistic S(X) of the research object X at the p-th grid point of the q-th data source is calculated using the sign function. p,q The formula is as follows:

[0023]

[0024]

[0025] Where, sign(X) j,p,q -X i,p,q The relationship between X and the p-th grid point in the q-th data source is the trend of the changes at time j and time i. p,q S(X) represents the research object X at the p-th grid point of the q-th data source. p,q () represents the cumulative change trend of the research object X at the p-th grid point of the q-th data source;

[0026] Calculate the Mann-Kendall confidence index Z for research object X at the p-th grid point of the q-th data source. S (X p,q The formula is as follows:

[0027]

[0028]

[0029] In the formula, V(S(X) p,q The variance of the Mann-Kendall confidence test for the research object X at the p-th grid point of the q-th data source is given.

[0030] Choose a confidence level α, if |Z S (X p,q )|>Z 1-α / 2 If the Sen trend of the research object at the p-th grid point of the q-th data source passes the Mann-Kendall confidence test at confidence level α, then Z 1-α / 2 This represents the value when the confidence level is α.

[0031] Furthermore, step 4, based on whether to consider credibility testing, selects composite trend evaluation indicators for the research subjects, specifically including:

[0032] 1) If reliability testing is not considered, the Sen slope of each grid point of each data source is used to assess the composite trend characteristics and intensity of the research objects;

[0033] If the Sen slope β(X) of the research object X at the p-th grid point of the q-th data source p,q β(X) > 0 indicates that the number of research subjects showed a positive growth trend during the research period, and vice versa; β(X) > 0 indicates a negative growth trend. p,q The magnitude of the value represents the extent of the growth, that is, the strength of the trend of change;

[0034] 2) If credibility testing is considered, the Sen slope and Mann-Kendall credibility test index of each grid point of each data source are used together to evaluate the composite trend characteristics and intensity of the research objects;

[0035] Given a confidence level α, if the Sen slope β(X) of the study object X at the p-th grid point of the q-th data source... p,q )>0 and |Z S (X p,q )|>Z 1-α / 2 This indicates that the research object X at the p-th grid point of the q-th data source shows a significant positive growth trend during the research period; β(X p,q )>0 but|Z S (X p,q )|≤Z 1-α / 2 This indicates that the research object X at the p-th grid point of the q-th data source shows a non-significant positive growth trend during the research period;

[0036] If β(X) p,q )<0 and |Z S (X p,q )|>Z 1-α / 2 This indicates that the research object X at the p-th grid point of the q-th data source shows a significant negative growth trend during the research period; if β(X p,q )<0 but|Z S (X p,q )| <Z 1-α / 2 This indicates that the research object X at the p-th grid point of the q-th data source shows a non-significant negative growth trend during the research period;

[0037] β(X p,q The magnitude of β(X) represents the magnitude of the increase, i.e., the strength of the trend; if β(X) p,q If ) = 0, it indicates that the research object X at the p-th grid point of the q-th data source has no trend change during the research period.

[0038] Furthermore, step 5, distinguishing the changing trend characteristics of the research objects in each data source, specifically includes:

[0039] 1) Define the trend characteristic MI(X) of the research object X at the p-th grid point of the q-th data source. p,q );

[0040] 2) Without considering reliability testing, based on the Sen slope β(X) of the research object X at the p-th grid point of the q-th data source. p,q Using formula (7), the trend characteristic MI(X) of the research object X at the p-th grid point of the q-th data source is defined. p,q ):

[0041]

[0042] If the Sen slope β(X) of the research object X at the p-th grid point of the q-th data source p,q )>0, MI(X) p,q A value of 1 indicates that the number of research subjects showed a positive growth trend during the research period;

[0043] If β(X) p,q )<0, MI(X) p,q A value of -1 indicates that the number of research subjects showed a negative growth trend during the research period;

[0044] If β(X) p,q ) = 0, MI(X) p,q A value of 0 indicates that the research subjects showed no trend of change during the research period;

[0045] 3) Considering the reliability test, based on the Sen slope β(X) of the research object X at the p-th grid point of the q-th data source. p,q ) and the Mann-Kendall credibility test index Z S (X p,q Using formula (8), we define the trend characteristic MI(X) of the research object X at the p-th grid point of the q-th data source. p,q ):

[0046]

[0047] If the Sen slope β(X) of the research object X at the p-th grid point of the q-th data source p,q )>0 and |Z S (X p,q )|>Z 1-α / 2 MI(X) p,q The value of 1 indicates that the research object X at the p-th grid point of the q-th data source shows a significant positive growth during the research period;

[0048] If β(X) p,q )<0 and |Z S (X p,q )|>Z 1-α / 2MI(X) p,q The value of -1 indicates that the research object X at the p-th grid point of the q-th data source shows a significant negative growth during the research period;

[0049] If |Z S (X p,q )|≤Z 1-α / 2 MI(X) p,q If the value is 0, it indicates that the trend of change of the research object X at the p-th grid point of the q-th data source is not significant during the research period.

[0050] Furthermore, step 6, based on whether or not reliability testing is considered, categorizes and summarizes the composite change trend characteristics of research object X at the p-th grid point, specifically including:

[0051] 1) Define the consistency evaluation index (CI) of the composite change trend of research object X at the p-th grid point. p );

[0052] 2) Based on the changing trend characteristics of the research object X at the p-th grid point of the q-th data source under different conditions, MI(X) p,q Using formula (9), the consistency index CI(X) of the composite change trend of the research object X at the p-th grid point is calculated. p );

[0053]

[0054] Among them, X p Let MI(X) be the composite trend of the research object X at the p-th grid point, q be the q-th data source, and m be the total number of data sources to be used. Under conditions of no reliability testing and with reliability testing, MI(X) p,q The values ​​are obtained by formula (7) and formula (8) respectively.

[0055] Furthermore, step 7, depending on whether credibility testing is considered, calculates the ratio of the composite change trend characteristics of the research object X at the p-th grid point under different conditions, specifically including:

[0056] The ratio (X) of different trends of each grid point under different conditions is calculated using formulas (10) and (11). p );

[0057]

[0058]

[0059] In the formula, Nic(X) p N1 is a classification statistic of the trend characteristics of the research object X at the p-th grid point. p N2 p and N3p N1 represents the statistical variables in the data source where the research object X shows positive growth, negative growth, and no significant growth at the p-th grid point, respectively. p (X p N2 p (X p ) and N3 p (X p Ratio(X) represents the total number of research objects X in the data source that show positive growth, negative growth, and no significant growth at the p-th grid point, respectively; p Ro1(X) is the ratio of different trend characteristics of the research object X at the p-th grid point. p Ro2(X) p ) and Ro3(X p ) represent the percentages of research object X in the data source that show positive growth, negative growth, and no significant growth at the p-th grid point, respectively.

[0060] Furthermore, step 8, depending on whether reliability testing is considered, calculates the composite change trend characteristics and intensity of each grid point under different conditions, specifically including:

[0061] 1) Define the composite trend evaluation index CCP(X) for the research object X at the p-th grid point. p ) and CCS(X p ); where CCP(X) p CCS(X) represents the composite trend characteristics of the research object. p () represents the intensity of the complex change trend;

[0062] 2) Calculate the composite change trend characteristics (CCP(X)) of each grid point under different conditions using formula (12). p );

[0063]

[0064] 3) Calculate the composite trend intensity CCS(X) of each grid point under different conditions using formula (13). p );

[0065]

[0066] In the formula, NaN value represents a null value, indicating that the research object X has no significant trend of change at the p-th grid point.

[0067] Beneficial effects: Compared with the prior art, the technical solution of the present invention has the following beneficial technical effects:

[0068] This invention employs a multi-source data-based cluster analysis method for diagnosing the composite trend characteristics and intensity of climate change. Utilizing large-scale multi-source data, it uses cluster analysis based on Sen slope estimation and the Mann-Kendall test to cluster similar information from different data sources to diagnose the composite trend characteristics and intensity of the research object. This method is highly efficient, noise-resistant, simple to operate, and has low computational cost, effectively diagnosing composite climate change characteristics at different spatiotemporal scales.

[0069] "Clustering" refers to weighting and averaging the precipitation trends and intensities of various data sources in a specific way to obtain a quantitative value, which represents the composite precipitation trend and intensity of multiple data sources. "Diagnosis" refers to analyzing the characteristics of the research object through this composite trend feature and intensity index, such as a precipitation composite trend showing an increasing trend and an intensity reaching 10 mm / year.

[0070] The multi-source data-based cluster analysis method for diagnosing the composite trend characteristics and intensity of climate change proposed in this invention is based on cluster analysis theory, utilizes big data technology, and employs data from multiple data sources to comprehensively diagnose the spatiotemporal composite evolution characteristics and intensity of hydrological and meteorological elements. It can directly serve the diagnosis of climate change trends and has broad application prospects. Attached Figure Description

[0071] Figure 1 This is a flowchart for diagnosing composite trend characteristics and intensity from multiple data sources;

[0072] Figure 2 This represents the spatial distribution of the annual average temperature Sen slope from various data sources.

[0073] Figure 3 This is the spatial distribution of the Mann-Kendall confidence test index for annual average temperature from various data sources;

[0074] Figure 4 It is the spatial distribution of the composite variation trend characteristics of annual average temperature from various data sources;

[0075] Figure 5 It represents the spatial distribution of the composite trend intensity of annual average temperature from various data sources. Detailed Implementation

[0076] The technical solution of the present invention will be further described below with reference to the accompanying drawings and embodiments. The following embodiments are only used to more clearly illustrate the technical solution of the present invention, and should not be used to limit the scope of protection of the present invention.

[0077] This embodiment contains existing temperature data sequences at different spatiotemporal scales, including CRU, CMFD, ERA5, and GLDAS-v2, covering the Qinghai-Tibet Plateau region from January 1981 to December 2018. Based on... Figure 1The present invention, as described in the flowchart, employs a multi-source data-based cluster analysis method for diagnosing the composite trend characteristics and intensity of climate change in the Qinghai-Tibet Plateau from 1981 to 2018. The process is as follows:

[0078] (1) The Qinghai-Tibet Plateau was selected as the study area, and temperature was taken as the research object (the research object refers to hydro-meteorological elements (such as precipitation, temperature, runoff, etc.)). Big data technology was used to obtain temperature data sequences of CRU, CMFD, ERA5 and GLDAS-v2 data products at different spatiotemporal scales from January 1981 to December 2018 (Table 1). The weighted average method was used to unify the data with different spatiotemporal resolutions to the same spatiotemporal scale. In this embodiment, all data were uniformly converted to an annual average spatial resolution of 0.25°.

[0079] Table 1

[0080]

[0081] (2) The Sen slope estimation method is used to analyze the changing trend of annual average temperature at each grid point of each data source. Using formula (1), the Sen slope members of annual average temperature at each grid point of each data source are obtained; using formula (2), the Sen slope of annual average temperature at each grid point of each data source is obtained using the median function in MATLAB, that is, the changing trend.

[0082]

[0083] β(X p,q ) = median(X k,p,q (2)

[0084] Among them, X j,p,q and X i,p,q Let X be the annual average temperature X of the q-th data source, at the p-th grid point, for the j-th and i-th years (j>i). n is the total time series length; i ranges from 1 to n-1; j is greater than i and ranges from i+1 to n; N is the total number of Sen slope members for each grid point. p,q β(X) represents the Sen slope of the annual average temperature at the p-th grid point of the q-th data source. p,q β(X) > 0 indicates a positive growth trend in the number of research subjects during the research period, while β(X) > 0 indicates a negative growth trend. p,q The magnitude of the value represents the extent of the growth, that is, the strength of the trend of change.

[0085] Figure 2 This represents the spatial distribution of the annual average temperature trends across the Qinghai-Tibet Plateau. Figure 2It can be seen that the annual average temperature based on CRU and ERA5 data showed a warming trend during the study period of 1981-2018, with the warming range ranging from [0 to 0.07 K / 5a]. The annual average temperature based on CMFD and GLDAS data showed a stronger trend in some areas during the study period of 1981-2018, and the trend even showed negative growth in some regions. For example, GLDAS data showed negative temperature growth in parts of the western and southeastern Tibetan Plateau, with a cooling range of [-0.21 K / 5a to 0.07 K / 5a]. In summary, the annual average temperature based on all data sources showed a warming trend in most areas of the Tibetan Plateau during the period of 1981-2018.

[0086] (3) The Mann-Kendall method was used to obtain the reliability test index of the annual average temperature change trend of each grid point in each data source. The sign function of formula (3) and formula (4) was used to calculate S(X) of each grid point in each data source. p,q The Mann-Kendall confidence index Z for the research subject X at the p-th grid point of the q-th data source is calculated using formulas (5) and (6). S (X p,q In this embodiment, a 5% confidence level test is selected, that is, if |Z S (X p,q If |> 1.96, then the trend of the annual average temperature at the p-th grid point of the q-th data source is considered to have passed the 5% confidence level test. (Z) 1-α / 2 The values ​​at the 0.01 and 0.05 confidence levels are 2.58 and 1.96, respectively; if |Z S (X p,q If Sen > 1.96, it indicates that the trend of X at the p-th grid point of the q-th data source passed the Mann-Kendall significance test with a confidence level of 5%.

[0087]

[0088]

[0089]

[0090]

[0091] Where, sign(X) j,p,q -X i,p,q The relationship between X and the p-th grid point in the q-th data source is the trend of the changes at time j and time i. p,q S(X) represents the research object X at the p-th grid point of the q-th data source. p,q V(S(X)) represents the cumulative change trend of the research object X at the p-th grid point of the q-th data source;p,q The variance of the Mann-Kendall confidence test for the research object X at the p-th grid point of the q-th data source is calculated.

[0092] Figure 3 This figure shows the spatial distribution of the Mann-Kendall confidence test index for the annual average temperature trends from various data sources between 1981 and 2018. As shown in the figure, the Sen slope of the annual average temperature based on CRU data passed the 5% confidence test in the Tibetan Plateau study area, i.e., |Z|. S (X p,q The annual average temperature variation trend based on CMFD and ERA5 passed the 5% confidence level test in most of the study areas. The annual average temperature variation trend based on GLDAS also passed the 5% confidence level test in most areas, but the variation trend showed a significant negative growth trend in the western and southeastern parts, and the temperature variation trend was not significant at the boundary between the positive and negative variation trends, such as at the boundary of the Qiangtang Plateau.

[0093] (4) Select the composite trend evaluation index of annual average temperature based on whether it passes the reliability test; distinguish the trend characteristics of annual average temperature MI(X) in each data source under different conditions. p,q If reliability testing is not considered, based on the Sen slope β(X) of the annual average temperature at each grid point of each data source... p,q Using formula (7), the changing trend characteristics of annual average temperature MI(X) in each data source are defined. p,q ).

[0094]

[0095] Considering the reliability test, based on the Sen slope β(X) of the annual average temperature at each grid point of each data source... p,q ) and the Mann-Kendall credibility test index Z S (X p,q Using formula (8), the changing trend characteristics of annual average temperature MI(X) in each data source are defined. p,q ).

[0096]

[0097] (5) Classify and summarize the composite trend characteristics of annual average temperature at each grid point. Use formula (9) to calculate the consistency index CI(X) of the composite trend of annual average temperature at each grid point under different conditions. p ).

[0098]

[0099] Among them, X pMI(X) represents the composite trend of the annual average temperature at the p-th grid point, q is the q-th data source, and m is the total number of data sources to be used. Under conditions of no reliability testing and with reliability testing, MI(X) p,q The values ​​are obtained by formula (7) and formula (8) respectively.

[0100] (6) Calculate the ratio (X) of the different trends in annual average temperature at each grid point using formulas (10) and (11). p ).

[0101]

[0102]

[0103] In the formula, Nic(X) p N1 is a categorical statistic describing the trend of annual average temperature at the p-th grid point. p N2 p and N3 p N1 represents the statistical variable in the data source where the annual average temperature shows positive growth, negative growth, and no significant growth at the p-th grid point, respectively. p (X p N2 p (X p ) and N3 p (X p Ratio(X) represents the total amount of annual average temperature in the data source that shows positive growth, negative growth, and no significant growth at the p-th grid point, respectively; p Ro1(X) is the ratio of different trend characteristics of annual average temperature at the p-th grid point. p Ro2(X) p ) and Ro3(X p ) represent the percentages of the annual average temperature in the data source that show positive growth, negative growth, and no significant growth at the p-th grid point, respectively.

[0104] (7) Calculate the composite trend assessment index (CCP(X)) for annual average temperature. p ) and CCS(X p Among them, CCP(X) p ) represents the characteristics of complex changing trends, CCS(X) p The composite trend intensity is represented by ) . The composite trend characteristics of annual average temperature at each grid point under different conditions (CCP(X)) are calculated using formula (12). p The composite trend intensity (CCS) of the annual average temperature at each grid point under different conditions is calculated using formula (13). p Based on CCP and CCS, the composite trend characteristics and intensity of annual average temperature changes at each grid point were diagnosed.

[0105]

[0106]

[0107] Figure 4 It is the spatial distribution of the composite variation trend characteristics of annual average temperature of various data sources under different conditions (CCP). Figure 5 This represents the spatial distribution of the composite trend intensity (CCS) of annual average temperature from various data sources under different conditions. As shown in the figure, without confidence testing, the composite trend of temperature from various data sources shows a warming trend in most areas of the Tibetan Plateau. Figure 4 The warming intensity is in the range of [0 0.09K / 5a]( Figure 5 Under the Mann-Kendall confidence test, the composite temperature trends from various data sources all show a significant warming trend in the northeastern and southeastern Yellow River-Yangtze River-Lancang River source areas and the southern Yarlung Tsangpo River basin, as well as at some boundary grid points. Figure 4 The strength is in the range of [0.03K / 5a 0.09K / 5a]. Figure 5 ).

[0108] In summary, the multi-data source composite trend feature and intensity diagnostic cluster analysis method proposed in this invention is based on cluster analysis theory. Utilizing large-scale, multi-source data, it uses Sen slope estimation and the Mann-Kendall test to cluster similar information from different data sources to diagnose the composite trend features and intensity of research objects. This method can effectively assess climate change characteristics at different spatiotemporal scales. This invention's multi-data source composite trend feature and intensity diagnostic cluster analysis method is highly efficient, noise-resistant, simple to operate, and computationally inexpensive, making it directly applicable to climate change trend diagnosis and showing broad application prospects.

[0109] The above description represents the preferred embodiments of the present invention. It should be noted that those skilled in the art can make various improvements and modifications without departing from the technical principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A cluster analysis method for diagnosing the composite trend characteristics and intensity of climate change based on multi-source data, characterized in that, The method comprises the following steps: Step 1, determining a research area, selecting hydro-meteorological elements representing regional climate change as research objects, and obtaining data sources of the selected research objects; Step 2, obtaining the change trend of the research objects at each grid point of each data source by using a Sen slope estimation method; Step 3, evaluating the reliability of the change trend of the research objects at each grid point of each data source by using a Mann-Kendall test; Step 4, selecting a composite change trend evaluation index of the research objects according to whether the reliability test is passed, specifically including: 1) if the reliability test is not considered, using the Sen slope of each grid point of each data source to evaluate the composite change trend characteristics and intensity of the research objects; If the research object X has a Sen slope β(X p,q ) > 0 at the pth grid point of the qth data source, it indicates that the research object has a positive growth trend during the research period, otherwise, it has a negative growth trend; the magnitude of β(X p,q ) represents the growth amplitude, i.e. the change trend strength. 2) if the reliability test is considered, using the Sen slope and the Mann-Kendall reliability test index of each grid point of each data source to evaluate the composite change trend characteristics and intensity of the research objects; Select a confidence level α, if the study object X in the qth data source pth grid Sen slope β(X p,q )>0 and |Z S (X p,q )|>Z 1-α / 2 , it indicates that the qth data source pth grid study object X in the study period shows a significant positive growth trend, wherein Z S (X p,q ) is the reliability test index of the change trend of the study object X in the qth data source pth grid, and Z 1-α / 2 is the value of the confidence level α. if β(x p,q )>0 but |z S (x p,q )|≤z 1-α / 2 , then it indicates that the qth data source and the pth grid point of the research object X show a significant positive growth trend in the research period. if β(X p,q )<0 and |Z S (X p,q )|>Z 1-α / 2 , it indicates that the qth data source and the pth grid point of the research object X show a significant negative growth trend in the research period. if β(X p,q )<0 but |Z S (X p,q )|<Z 1-α / 2 , it indicates that the research object X at the pth grid point of the qth data source presents a non-significant negative growth trend in the research period; β(X p,q ) represents the magnitude of the growth, i.e. the strength of the trend; if β(X p,q ) = 0, it indicates that the research object X at the pth grid point of the qth data source has no trend change in the research period. Step 5, distinguishing the change trend characteristics of the research objects in each data source under different conditions; Step 6, classifying and summarizing the composite change trend characteristics of the research objects at each grid point; Step 7, calculating the composite change trend characteristic ratio of the research objects at each grid point by using a clustering analysis method; Step 8, calculating the composite change trend characteristics and intensity index of the research objects at each grid point by using a clustering analysis method; Step 9, diagnosing the climate change characteristics of the research area according to the composite change trend characteristics and intensity index of the research objects. 2.The multi-source data climate change composite trend feature and intensity diagnosis cluster analysis method according to claim 1, characterized in that, In the step 2, the Sen slope estimation method is used to analyze the change trend of the research objects at each grid point of each data source, specifically including: Formula (1) is used to obtain the Sen slope member of the research objects at a fixed time step in the research period at the pth grid point of the qth data source; Formula (2) is used to obtain the Sen slope of the research objects at a fixed time step in the research period at the pth grid point of the qth data source, that is, the change trend, by using the median function in MATLAB. β(X p,q ) = median(X k,p,q ) (2) wherein X k,p,q is the kth Sen slope member of the research object X at the fixed time step in the research period at the pth grid point of the qth data source, X j,p,q and X i,p,q are the jth and ith moments of the research object X at the pth grid point of the qth data source, respectively, j>i, n is the total time series length, N is the total number of Sen slope members at each grid point; and β(X p,q ) is the Sen slope of the research object X at the pth grid point of the qth data source. β(X p,q ) > 0 indicates that the research object has a positive growth trend during the research period, otherwise a negative growth trend; the magnitude of β(X p,q ) represents the magnitude of growth, i.e. the strength of the change trend. 3.The multi-source data climate change composite trend feature and intensity diagnosis cluster analysis method according to claim 2, characterized in that, The step 3 uses the Mann-Kendall method to obtain a reliability test index Z of a change trend of the research object X at a pth grid point of a qth data source s (X p,q ), and specifically comprises: The cumulative trend statistic S(X) of the research object X at the p-th grid point of the q-th data source is calculated using the sign function. p,q The formula is as follows: wherein sign(X j,p,q ) is the change trend of the research object X at the jth and ith moment at the pth grid point of the qth data source; X i,p,q represents the research object X at the pth grid point of the qth data source, and S(X p,q ) is the cumulative change trend of the research object X at the pth grid point of the qth data source. p,q ​ The Mann-Kendall reliability test index Z of the research object X at the pth grid point of the qth data source S (X p,q ), the formula is as follows: In the formula, V(S(X p,q )) is the Mann-Kendall credibility test variance of the research object X at the pth grid point of the qth data source. Select the confidence level α, if |Z S (X p,q )|>Z 1-α / 2 , the Sen trend of the study object at the pth grid point of the qth data source passes the Mann-Kendall credibility test at the confidence level α, Z 1-α / 2 is the value of the confidence level α.

4. The multi-source data climate change composite trend feature and intensity diagnosis cluster analysis method according to claim 3, characterized in that, In the step 5, the change trend characteristics of the research objects in each data source are distinguished, specifically including: 1) defining a change trend feature MI(X p,q ) of the research object X at the pth grid point of the qth data source; 2) Without considering the credibility test, according to the Sen slope β(X p,q ) of the research object X at the pth grid point of the qth data source, the change trend feature MI(X p,q ) of the research object X at the pth grid point of the qth data source is defined by formula (7): If the study object X has a Sen slope β(X p,q )>0 at the pth grid point of the qth data source, and MI(X p,q ) equals 1, it indicates that the study object has a positive growth trend during the study period. If β(X p,q ) < 0, MI(X p,q ) equals -1, indicating that the subject under study is in a negative growth trend during the study period; If β(x p,q ) = 0, MI(x p,q ) equals 0, indicating that the subject under study has no trend of change during the study period; 3) Considering the reliability test, based on the Sen slope β(X) of the research object X at the p-th grid point of the q-th data source. p,q ) and the Mann-Kendall credibility test index Z S (X p,q Using formula (8), we define the trend characteristic MI(X) of the research object X at the p-th grid point of the q-th data source. p,p ): if the Sen slope β(X p,q )>0 and |Z S (X p,q )|>Z 1-α / 2 , MI(X p,q ) equals 1, indicating that the research object X at the pth grid point of the qth data source is significantly positively growing in the research period; if β(x p,q )<0 and |z S (x p,q )|>z 1-α / 2 , mi(x p,q ) equals -1, indicating that the subject x has a significant negative growth at the pth grid point of the qth data source during the study period; If |Z s (X p,q ) is less than or equal to Z 1-α / 2 , MI(X p,q ) equals 0, indicating that the research subject X is not significant in the trend of change in the research period at the pth grid point of the qth data source.

5. The multi-source data climate change composite trend feature and intensity diagnosis cluster analysis method according to claim 4, characterized in that, In the step 6, the composite change trend characteristics of the research objects X at the pth grid point are classified and summarized according to whether the reliability test is considered, specifically including: 1) define the consistency evaluation index CI(Xp) of the compound change trend of the study object X at the pth lattice point p ); 2) According to the change trend characteristics MI(X p,q ) of the research object X at the pth grid point of the qth data source under different conditions, the composite change trend consistency index CI(X p ) of the research object X at the pth grid point is calculated by using formula (9); wherein X p is the composite trend of the research object X at the pth lattice point, q is the qth data source, and m is the total amount of data sources to be adopted; under the condition of not considering the reliability test and considering the reliability test, MI(X p,q ) is calculated by formula (7) and formula (8), respectively.

6. The multi-source data climate change composite trend feature and intensity diagnosis cluster analysis method according to claim 5, characterized in that, In the step 7, the composite change trend characteristic ratio of the research objects X at the pth grid point under different conditions is calculated according to whether the reliability test is considered, specifically including: The ratio Ratio(X) of different change trends of each grid point under different conditions is calculated by using formula (10) and formula (11) p ); In the formula, Nic(X p ) is the change trend characteristic classification statistic of the research object X at the pth lattice point, N1 p , N2 p , and N3 p are the statistical variables of the research object X at the pth lattice point in the data source, respectively showing positive growth, negative growth, and no significant growth; N1 p (X p ), N2 p (X p ), and N3 p (X p ) are the total amounts of the research object X at the pth lattice point in the data source, respectively showing positive growth, negative growth, and no significant growth; Ratio(X p ) is the ratio of different change trend characteristics of the research object X at the pth lattice point, Ro1(X p ), Ro2(X p ), and Ro3(X p ) represent the ratios of the research object X at the pth lattice point in the data source, respectively showing positive growth, negative growth, and no significant growth.

7. The multi-source data climate change composite trend feature and intensity diagnosis cluster analysis method according to claim 6, characterized in that, In the step 8, the composite change trend characteristics and intensity of each grid point under different conditions are calculated according to whether the reliability test is considered, specifically including: 1) Define the composite trend evaluation index CCP(X) for the research object X at the p-th grid point. p ) and CCS(X p ); where CCP(X) p CCS(X) represents the composite trend characteristics of the research object. p () represents the intensity of the complex change trend; 2) Calculate the composite change trend characteristics CCP(Xi, t) of each grid point under different conditions using formula (12) p ); 3) Calculate the composite change trend intensity CCS(X, Y, t) of each grid point under different conditions using formula (13) p ); In the formula, NaN represents a null value, representing that the research objects X have no significant change trend at the pth grid point.

Images