Extreme rainfall change diagnosis attribution and estimation method, equipment and storage medium

Abstract

The invention discloses an extreme rainfall change diagnosis attribution and estimation method, equipment and a storage medium, and the method comprises the steps: employing a plurality of statistical analysis and attribution methods, carrying out the depiction of extreme rainfall from a plurality of dimensions, such as frequency, intensity, duration, and the like, and analyzing the time-space evolution trend of the extreme rainfall; correlativity among extreme rainfall change, global warming and large-scale circulation factors is disclosed; a multi-mode and multi-method integrated non-consistency multi-dimensional deviation correction method is constructed, the spatial correlation and the change trend of a precipitation field are effectively reserved, and a basis is provided for construction of an extreme precipitation scene in the future.

Description

Technical Field

[0001] The present invention relates to the technical field of climatology, and in particular to a method, device and storage medium for diagnosing, attributing and estimating extreme precipitation changes. Background Art

[0002] In recent years, climate change has altered the global hydrological cycle, directly impacting hydrological elements such as precipitation and runoff, leading to spatiotemporal redistribution of water resources and changes in total water resources. Within the context of large-scale circulation systems, dynamic factors directly influence the triggering conditions and maintenance mechanisms of precipitation processes by altering atmospheric stability and water vapor transport pathways, thereby affecting the frequency, intensity, and duration of extreme precipitation. At the regional and river basin scales, extreme precipitation is influenced by a combination of climate change, large-scale circulation systems, and local factors, making its spatiotemporal evolution increasingly complex. Under the influence of global warming and intensive human activities, the frequency, intensity, spatial extent, and duration of extreme precipitation events are shifting, leading to changes in their flooding effects. This poses new challenges to flood control and disaster reduction in river basins and poses a significant threat to people's lives and property. Therefore, the attribution of spatiotemporal changes in extreme precipitation and the projection of future scenarios have profound implications for flood control and disaster reduction in river basins and the optimal allocation of water resources.

[0003] Global climate models (GCMs) are a key method for studying the mechanisms of climate change and a crucial tool for projecting future changes in the climate system. Studies of the impact of global climate change on regional hydrological cycle elements generally rely on precipitation data from the international Coupled Model Intercomparison Project (CMIP) GCM ensemble as input to hydrological models. Currently, global climate models have reached CMIP6. However, due to multiple factors, including the complexity of the climate system and the uncertainty of climate models, the CMIP6 models exhibit uncertainty in their ability to simulate precipitation in specific regions or basins. Bias correction of CMIP6 data is necessary to improve simulation accuracy and meet the requirements of basin-wide hydrological simulations. Existing diagnostic, attribution, and projection methods for extreme precipitation fail to fully account for the spatiotemporal correlations and inconsistencies of precipitation fields. Therefore, developing diagnostic, attribution, and projection methods that can comprehensively account for these spatiotemporal correlations and inconsistencies is crucial for improving the accuracy and reliability of extreme precipitation forecasts. Summary of the Invention

[0004] Purpose of the invention: The present invention provides a method, device and storage medium for diagnosing, attributing and estimating extreme precipitation changes, which can effectively preserve the spatial correlation and change trend of the precipitation field and provide a basis for constructing future extreme precipitation scenarios.

[0005] Technical solution: The method for diagnosing, attributing, and estimating extreme precipitation changes described in the present invention includes the following steps:

[0006] Step 1, collecting daily scale station and grid precipitation observation data, atmospheric circulation index, global average near-surface air temperature data, global climate model data in the basin;

[0007] Step 2, data preprocessing, missing data imputation and data spatial interpolation are performed on the collected data;

[0008] Step 3, extreme precipitation index calculation: based on the daily scale precipitation observation data measured by the meteorological station, the annual scale time series of the extreme precipitation index is calculated and obtained;

[0009] Step 4, analysis of the temporal and spatial variation trend of the extreme precipitation index, comparison of the climatic characteristics of the precipitation index in different periods, and test of the temporal and spatial variation trend of the extreme precipitation index based on the annual scale time series of the extreme precipitation index obtained in step 3;

[0010] Step 5, attribution analysis of the trend change of extreme precipitation, analysis of the correlation between the circulation factor, global warming and the extreme precipitation index based on the atmospheric circulation index, global average near-surface air temperature data obtained in step 1 and the annual scale time series of the extreme precipitation index obtained in step 3;

[0011] Step 6, future extreme precipitation scenario estimation, combining the daily scale station and grid precipitation observation data obtained in step 1 and the global climate model output data, using the non-uniform multi-dimensional bias correction method to correct the bias of the large-scale climate model output data, and constructing the future extreme precipitation scenario.

[0012] Further, in step 1, the daily scale station and grid precipitation observation data include daily scale precipitation observation data covering multiple stations in multiple years and the China CN05.1 grid daily observation data set; the atmospheric circulation index includes atmospheric / ocean circulation index (Arctic Oscillation, North Atlantic Oscillation, Atlantic Multidecadal Oscillation, Pacific Decadal Oscillation, El Nino and Southern Oscillation), summer monsoon index (East Asian summer monsoon index, South Asian summer monsoon index, South China Sea summer monsoon index); the global average near-surface air temperature data is the global average near-surface air temperature anomaly; the global climate model data is the latest CMIP6 multi-model ensemble historical simulation and future scenario data.

[0013] Further, in step 2, the missing records of the daily scale station precipitation observation data are filled using daily climatology values; based on the first-order conservative interpolation algorithm in the Climate Data Operator (CDO) toolbox, the precipitation data of the climate model is uniformly interpolated to the same 0.25°×0.25° spatial resolution grid as the observation data to ensure the consistency of the spatial resolution.

[0014] Further, in step 3, the R software package "climdex.pcic" is used to obtain the extreme precipitation index sequence.

[0015] Further, in step 4, the probability density curve of each extreme precipitation index is drawn based on kernel density estimation (KDE) to show the probability distribution of each extreme precipitation index; the Kolmogorov-Smirnov test is used to determine whether the probability distributions of the extreme precipitation index sequences in different periods are significantly different; and the iterative Mann-Kendall trend test method is used to test the long-term change trend of each extreme precipitation index.

[0016] Further, in step 5, the extreme precipitation trend change attribution analysis specifically includes the following steps:

[0017] Step 51, calculate the regional average extreme precipitation index on an annual scale; convert the atmospheric circulation index and the global average near-surface air temperature anomaly into an annual scale;

[0018] Step 52, use the Spearman rank correlation analysis method to calculate the correlation coefficient between the annual scale regional average extreme precipitation index and the annual scale atmospheric circulation index and the global average near-surface air temperature anomaly, and perform a significance test thereon, to preliminarily identify the statistical correlation between the extreme precipitation index and the circulation index and global warming:

[0019]

[0020] In the formula: The difference between the rank of the ith observation in the two variables n represents the sample size; Xi and Yi represent the rank of the ith observation in variable X and Y, respectively; p s ∈[-1,1] represents the correlation coefficient, and the result is positive for positive correlation and negative for negative correlation, and the absolute value is closer to 1, the stronger the correlation;

[0021] For a large sample (usually n>30), if there is a significant rank correlation between two variables, the t distribution can be approximated:

[0022]

[0023] Further, in step 6, the future extreme precipitation scenario prediction specifically includes the following steps:

[0024] Step 61, extract CMIP6 multi-model ensemble precipitation data, unify the time range (baseline period, future scenario period), and use spatial interpolation method to spatially interpolate the precipitation data output by the climate model, so that the spatial resolution is consistent with the observation value;

[0025] Step 62, using multi-dimensional bias correction method, considering the "non-uniformity" problem of correlation structure, bias correction is carried out on large-scale climate model to obtain gridded future precipitation data set under different greenhouse gas emission scenarios with spatial and temporal correlation.

[0026] Further, in step 62, quantile delta mapping (QDM) bias correction method is used to perform one-dimensional bias correction on climate variables in each model grid to make them consistent with observation data in statistical characteristics as much as possible and retain the change trend of climate variables; R2D2 method based on rank reordering is used to reconstruct the correlation structure of precipitation data between model grids, which includes the following steps:

[0027] Step 621, based on the cumulative distribution function F (t) m,p , the cumulative probability τ m,p (t) of the future time t model simulation value x m,p (t) is calculated, and the quantile position of the model simulation value in the CDF is determined:

[0028] τ m,p (t)=F (t) m,p [x m,p (t)],τ m,p (t)∈{0,1}

[0029] Step 622, the change ratio Δ m,p (t) of the model simulation value x m (t) in the future relative to the historical period is calculated:

[0030]

[0031] In the formula: is the inverse function of the CDF of the original model data in the historical period;

[0032] Step 623, the cumulative probability τ m,p (t) is substituted into the inverse function of the historical CDF of the observation data to correct the systematic bias:

[0033]

[0034] Step 624, the historical bias correction value x o:m,h:p (t) is multiplied by the relative change Δ m (t) to obtain the corrected model data x m,c (t), which ensures that the relative trend of the model to future climate change is retained while the bias is corrected:

[0035] x m,c ​(t) = x o:m,h:p (t)Δ m (t)

[0036] Step 625: Using the rank matrix of the observed data as a reference, use the R2D2 method based on rank reordering to reorder the model data that has been corrected in step 624 to reconstruct the correlation structure of precipitation data between model grids and obtain future precipitation data sets for different scenarios with spatiotemporal correlation. The R2D2 rank reordering method is based on the R function package "MBC".

[0037] Accordingly, a device for diagnosing, attributing, and estimating extreme precipitation changes includes: one or more processors;

[0038] a storage device for storing one or more programs and user data;

[0039] When the one or more programs are executed by one or more processors, the one or more processors implement a method for diagnosing, attributing, and predicting extreme precipitation changes.

[0040] Accordingly, a storage medium for diagnosing, attributing and estimating extreme precipitation changes stores a computer program, which, when executed by a processor, implements a method for diagnosing, attributing and estimating extreme precipitation changes.

[0041] Beneficial effects: Compared with the existing technology, the present invention has the following significant advantages: it adopts a variety of statistical analysis and attribution methods to characterize extreme precipitation from multiple dimensions such as frequency, intensity, duration, and analyze its temporal and spatial evolution trend, revealing the correlation between extreme precipitation changes and global warming and large-scale circulation factors; it constructs a non-consistent multidimensional deviation correction method integrating multiple models and multiple methods, effectively retains the spatial correlation and change trend of the precipitation field, and provides a basis for the construction of future extreme precipitation scenarios. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Figure 1 Schematic diagram of the method of the present invention.

[0043] Figure 2 This is a schematic diagram of the spatial distribution of the long-term trend of extreme precipitation indices at all stations in the study area obtained by the iterative Mann-Kendall trend test in the present invention.

[0044] Figure 3 This is a schematic diagram of the correlation matrix between the average extreme precipitation index in the study area and the atmospheric circulation pattern and the global average near-surface temperature anomaly obtained by the Spearman rank correlation analysis method in the present invention.

[0045] Figure 4Schematic diagram of the multi-year average precipitation simulation results (top) and deviations (bottom) for the study area based on the CMIP6 multi-model ensemble obtained by the present invention through inconsistency bias correction.

[0046] Figure 5 This is a schematic diagram of the change in multi-year average precipitation in the study area from 2021 to 2050 under the SSP126 scenario predicted by the CMIP6 models in this invention. DETAILED DESCRIPTION

[0047] A method for diagnosing, attributing, and predicting extreme precipitation changes includes the following steps:

[0048] Step 1: Collect daily-scale station and grid precipitation observation data within the basin, atmospheric circulation index, global mean near-surface temperature data, and global climate model data;

[0049] Step 2: Data preprocessing: missing data interpolation and data space interpolation of the collected data;

[0050] Step 3: Calculation of extreme precipitation index: Based on the daily precipitation observation data measured at the meteorological station, the annual time series of extreme precipitation index is calculated;

[0051] Step 4: Analyze the spatiotemporal variation trend of the extreme precipitation index. Combined with the annual time series of the extreme precipitation index obtained in step 3, the climatological characteristics of the precipitation index in different periods are compared to test the spatiotemporal variation trend of the extreme precipitation index.

[0052] Step 5: Attribution analysis of extreme precipitation trend changes. Combine the atmospheric circulation index and global mean near-surface temperature data obtained in step 1 with the annual time series of extreme precipitation index obtained in step 3 to analyze the correlation between circulation factors, global warming, and extreme precipitation index.

[0053] Step 6: Forecast future extreme precipitation scenarios. Combining the daily-scale station and grid precipitation observation data obtained in step 1 with the global climate model output data, the non-uniform multidimensional bias correction method is used to perform bias correction on the large-scale climate model output data to construct future extreme precipitation scenarios.

[0054] This embodiment of the present invention uses daily precipitation observation data from 176 meteorological stations in the Yangtze River Basin as an example. The following process is implemented using the R language platform, and the method steps are detailed with reference to the accompanying figures. This embodiment is intended to illustrate the present invention but is not intended to limit its scope of application. It is equally applicable to different regions or time periods.

[0055] The method of the present invention is a method for diagnosing, attributing and predicting extreme precipitation changes. The implementation flow chart is as follows: Figure 1 The specific steps are as follows:

[0056] Step 1, Precipitation data were obtained from the China National Meteorological Information Center (https: / / data.cma.cn / en), and global average surface air temperature data were obtained from the NASA GISS (http: / / data.giss.nasa.gov / gistemp / ). Five oceanic / atmospheric circulation indices, including the Atlantic Multidecadal Oscillation (AMO), Arctic Oscillation (AO), North Atlantic Oscillation (NAO), Pacific Decadal Oscillation (PDO), and Oceanic Niño Index (ONI) (characterizing El Niño-Southern Oscillation, ENSO), were obtained from the NOAA Physical Sciences Laboratory (https: / / esrl.noaa.gov / psd / data / climateindices / list / ). Three summer monsoon indices, including the East Asian Summer Monsoon (EASM), South Asian Summer Monsoon (SASM), and South China Sea Summer Monsoon (SCSM), were obtained from the research group of Professor Li Jianping (http: / / lijianping.cn / dct / page / 1). CMIP6 multi-model ensemble historical simulation and future scenario data were obtained from the CMIP6 data public website https: / / esgf-node.llnl.gov / search / cmip6 / h. The time series of extreme precipitation indices were from 176 meteorological stations in the Yangtze River Basin from 1960 to 2019. The daily precipitation data simulated by 24 climate models in CMIP6 included historical scenario data from 1961 to 2014 and data from 2021 to 2050 under two future projection scenarios (Shared Socioeconomic Pathway SSP126 and SSP585). Table 1 shows the detailed information of the 24 CMIP6 global climate models used.

[0057] Table 1. Information of CMIP6 global climate models

[0058]

[0059] Step 2, Data preprocessing: The daily precipitation observation data from 176 stations in the Yangtze River Basin from 1960 to 2019 were used, with the total missing data percentage less than 1% and the annual missing data percentage less than 20%. The missing precipitation data were filled using daily climatology values. Based on the first-order conservative interpolation algorithm in the Climate Data Operator (CDO) toolbox, the precipitation data from climate models were uniformly interpolated to the same 0.25°x0.25° spatial resolution grid as the observation data, ensuring the consistency of spatial resolution.

[0060] Step 3: Calculation of the Extreme Precipitation Index: The frequency, intensity, and duration of extreme precipitation in the Yangtze River Basin are measured using the 12 extreme precipitation indices defined by the Expert Team on Climate Change Detection and Indices (ETCCDI). These indices include: maximum one-day precipitation (Rx1day), maximum five-day precipitation (Rx5day), heavy precipitation (R95p), ​​extreme heavy precipitation (R99p), number of precipitation days (R1mm), number of heavy precipitation days (R10mm), number of extremely heavy precipitation days (R20mm), number of extremely heavy precipitation days (R50mm), maximum consecutive dry days (CDD), maximum consecutive wet days (CWD), total precipitation (PRCPTOT), and mean daily precipitation intensity (SDII). Daily precipitation observations are processed into an annual-scale extreme precipitation index time series. Table 2 provides detailed information on the extreme precipitation indices used.

[0061] Table 2 Extreme precipitation index information

[0062]

[0063] Step 4: Analysis of spatiotemporal trend of extreme precipitation index: Combined with the annual time series of extreme precipitation index obtained in step 3, the climatological characteristics of precipitation index in different periods are compared to test the spatiotemporal trend of extreme precipitation index;

[0064] Based on kernel density estimation (KDE), probability density curves of each extreme precipitation indices were drawn to show the long-term climatological characteristics of each extreme precipitation indices in the Yangtze River Basin (upper, middle and lower reaches). The Kolmogorov-Smirnov test was applied to determine whether the probability distributions of extreme precipitation index series in two different periods were significantly different. The iterative Mann-Kendall trend test method was used to test the long-term change trends of each extreme precipitation indices in the Yangtze River Basin.

[0065] To eliminate the influence of serial autocorrelation on the significance test results of the traditional Mann-Kendall method, an improved Mann-Kendall method—the iterative MK (IMK) trend test—is used to further test trend significance. Given that trends and autocorrelation often coexist in a time series, the IMK test iteratively estimates the trend and first-order autocorrelation coefficient. After completing the entire iterative process, the Mann-Kendall test is performed on the final pre-whitening time series to determine the significance of the trend. Because the Mann-Kendall method lacks the ability to provide the degree of trend, the Sen's slope estimation method is used to quantify the degree of trend. The significance level α is set to 0.05, and the magnitude of the trend (i.e., the slope) is calculated using the Sen's slope estimation method.

[0066] Kolmogorov-Smirnov test is a non-parametric test method for statistical analysis, mainly used to assess whether a single sample conforms to a specific distribution, or to determine whether the distribution functions of two independent samples are the same. In the two-sample test, this method mainly measures the difference between the two samples by comparing their empirical cumulative distribution functions, and the test method is sensitive to the shape and location changes in the distribution function. The significance level a is set to 0.05.

[0067] In this embodiment, based on step 4, it is obtained that: although the total precipitation (PRCPTOT) in the middle reaches of the Yangtze River is about 10% more than that in the lower reaches, the climate average of the average daily precipitation intensity (SDII) of the two is similar. The extreme precipitation occurrence frequency in the middle and lower reaches of the Yangtze River is higher than that in the upper reaches of the Yangtze River (reflected by R10mm, R20mm and R50mm), and the intensity is greater (reflected by Rx1day, Rx5day, R95p and R99p). If the influence of spatial difference is considered, the occurrence frequency and intensity of extreme precipitation in the middle and lower reaches are basically similar. In contrast, the maximum continuous drought days (CDD is 44 days / year) in the upper reaches is greater than that in the middle and lower reaches (CDD is about 30 days / year), and the maximum continuous precipitation days (CWD is about 8 days / year) in the upper, middle and lower reaches are similar.

[0068] The iterative Mann-Kendall trend test method is used to detect the long-term change trend of extreme precipitation in the Yangtze River Basin, and five types of trends are considered here: significant increasing trend (increasing at 0.05 significance level), significant decreasing trend (decreasing at 0.05 significance level), increasing trend, decreasing trend and no trend. As shown in Table 2, the extreme precipitation in the Yangtze River Basin has a significant increasing trend in the past 60 years, and the increasing trend of the upper reaches is more significant than that of the middle and lower reaches. Figure 2As shown, most stations in the Yangtze River Basin showed increasing trends in Rx1day (74.4%), Rx5day (64.8%), R95p (76.7%), R10mm (42.6%), R20mm (41.5%), PRCPTOT (62.5%), and SDII (76.7%). For R99p and R50mm, although most stations did not show a clear trend, 14 and 9 stations, respectively, showed significant increasing trends. Some stations had R50mm values ​​consistently at 0 throughout the study period, resulting in no detectable trend at these stations. In contrast, more stations showed decreasing trends than increasing trends for R1mm (55.1%), CDD (47.2%), and CWD (32.4%). These results indicate that both average precipitation intensity (represented by SDII) and extreme precipitation intensity (represented by Rx1day, R95p, and R99p) have increased significantly in the Yangtze River Basin over the past 60 years. Furthermore, despite a significant decrease in the number of precipitation days (R1 mm) and the number of longest continuous precipitation days (CWD), the frequency of extreme heavy precipitation (R50 mm) increased significantly, particularly in the southwestern region of the midstream. This will lead to increases in both average and extreme precipitation intensity in the region. PRCPTOT is significantly correlated with almost all EPIs (except CWD), indicating that total annual precipitation is closely related to extreme precipitation events. In contrast, CWD has weaker correlations with other EPIs.

[0069] Step 5: Attribution analysis of extreme precipitation trend changes: Combine the atmospheric circulation index obtained in step 1, the global mean near-surface temperature data, and the annual time series of the extreme precipitation index obtained in step 3 to analyze the correlation between circulation factors, global warming, and the extreme precipitation index;

[0070] Step 5.1: Calculate the annual-scale mean extreme precipitation index for the Yangtze River Basin region and convert the atmospheric circulation index and global mean near-surface temperature anomaly to the annual scale;

[0071] Step 5.2: Use the Spearman rank correlation analysis method to calculate the correlation coefficient between the annual-scale Yangtze River Basin regional average extreme precipitation index and the annual-scale atmospheric circulation index and the global average near-surface temperature anomaly, and perform a significance test to preliminarily identify the statistical correlation between the extreme precipitation index, the circulation index, and global warming:

[0072]

[0073] Where: Represents the difference in rank between the i-th observation value in the two variables n represents the sample size; represents the rank of the i-th observation in variables X and Y; ρ sThe correlation coefficient ∈ [-1, 1] indicates that the result is positive for positive correlation and negative for negative correlation, and the absolute value is closer to 1, the correlation is stronger.

[0074] For a large sample (usually n > 30), if there is a significant rank correlation between two variables, the t distribution can be approximated:

[0075]

[0076] In this embodiment, the correlation coefficient matrix is drawn based on step 5.2, as shown in Figure 3 The AMO, GTEM, summer monsoon index (EASM and SCSM) are significantly correlated with the frequency, intensity and duration of extreme precipitation in the Yangtze River Basin. ENSO is also an important factor, while AO, NAO, PDO and SASM have relatively small effects on extreme precipitation in the Yangtze River Basin.

[0077] Step 6, future extreme precipitation scenario estimation: combining the daily scale station and grid precipitation observation data obtained in step 1 and the global climate model output data, using the non-uniform multi-dimensional bias correction method, the bias correction of the large-scale climate model output data is carried out, and the future extreme precipitation scenario is constructed;

[0078] Step 6.1, extract CMIP6 multi-model ensemble precipitation data, unify the time range (baseline period, future scenario period), and use spatial interpolation method to spatially interpolate the precipitation data output by the climate model, so that the spatial resolution is consistent with the observation value;

[0079] Step 6.2, using multi-dimensional bias correction method, considering the "non-uniformity" of correlation structure, bias correction is carried out on large-scale climate model, and future precipitation data set with spatial and temporal correlation under different greenhouse gas emission scenarios is obtained;

[0080] Step 6.2.1, based on the cumulative distribution function F (t) m,p , the cumulative probability τ m,p (t) of the model simulation value x m,p (t) at time t in the future is calculated, and the quantile position of the model simulation value in the CDF is determined:

[0081] τ m,p (t) = F (t) m,p [x m,p (t)], τ m,p (t) ∈ {0, 1}

[0082] Step 6.2.2, calculate the change ratio Δ m,p (t) of the model simulation value x m (t) in the future relative to the historical period:

[0083]

[0084] Where: is the inverse function of the CDF of the original model data in the historical period;

[0085] Step 6.2.3, the cumulative probability τ m,p (t) Substitute the inverse function of the historical CDF of the observed data into To correct for system deviations:

[0086]

[0087] Step 6.2.4, correct the historical bias value x o:m,h:p (t) multiplied by the relative change Δ m (t) Get the corrected pattern data x m,c (t) Ensure that the relative trends of the models for future climate change are preserved while correcting for biases:

[0088] x m,c (t) = x o:m,h:p (t)Δ m (t)

[0089] In step 6.2.5, using the observed data rank matrix as a reference, the model data corrected in step 624 are reordered using the R2D2 rank reordering method to reconstruct the correlation structure of precipitation data between model grids and obtain future precipitation datasets with different scenarios that exhibit spatiotemporal correlations. The R2D2 rank reordering method is based on the R function package "MBC."

[0090] In this embodiment, Figure 4 As shown, the calibration of the multi-model ensemble shows a general improvement in simulation performance across the basin, particularly in reducing overall overestimations. Before calibration, the multi-model ensemble overestimated annual mean precipitation in the Yangtze River Basin across most regions: significantly in the upper reaches (absolute error 0–5 mm / day), underestimated in the central and eastern parts of the lower reaches (absolute error -0.5–0 mm / day), and less pronounced in other regions, with absolute errors remaining between 0 and 1 mm / day. After calibration, the absolute errors in the climatological state are now within 0–1 mm / day, indicating that the bias-corrected model has effectively improved the applicability of the CMIP6 multi-model ensemble for the Yangtze River Basin.

[0091] like Figure 5As shown in the calibrated climate change signals for 24 different CMIP6 models under the SSP126 scenario, compared to the historical baseline period (1961-2014), multi-year mean precipitation increases across most of the Yangtze River Basin under this scenario, with the largest changes concentrated in the central-western, southwestern, and southeastern regions of the basin. Thirteen models, including ESM1, MRI-ESM2-0, MIROC6, INM-CM5-0, EC-Earth3-Veg-LR, EC-Earth3-Veg, and EC-Earth3, show increasing precipitation across most of the Yangtze River Basin, with the largest increases projected by ACCESS-ESM1-5, ACCESS-CM2, and ESM1. Eleven models, including NorESM2-MM, show decreasing precipitation in some regions, primarily in the southwestern and southeastern regions of the Yangtze River Basin.

Claims

1. A method for diagnosing, attributing, and estimating extreme precipitation changes, characterized by: The steps include: Step 1: Collect daily-scale station and grid precipitation observation data within the basin, atmospheric circulation index, global mean near-surface temperature data, and global climate model data; Step 2: Data preprocessing: interpolation of missing data and data space interpolation of the collected data; Step 3: Calculation of extreme precipitation index: Based on the daily precipitation observation data measured at the meteorological station, the annual time series of extreme precipitation index is calculated; Step 4: Analyze the spatiotemporal variation trend of the extreme precipitation index. Combined with the annual time series of the extreme precipitation index obtained in step 3, the climatological characteristics of the precipitation index in different periods are compared to test the spatiotemporal variation trend of the extreme precipitation index. Step 5: Attribution analysis of extreme precipitation trend changes. Combine the atmospheric circulation index and global mean near-surface temperature data obtained in step 1 with the annual time series of extreme precipitation index obtained in step 3 to analyze the correlation between circulation factors, global warming, and extreme precipitation index. Step 6: Forecast future extreme precipitation scenarios. Combining the daily-scale station and grid precipitation observation data obtained in step 1 with the global climate model output data, the non-uniform multidimensional bias correction method is used to perform bias correction on the large-scale climate model output data to construct future extreme precipitation scenarios.

2. The method for diagnosing, attributing, and estimating extreme precipitation changes according to claim 1, wherein: In step 1, the daily-scale station and gridded precipitation observation data include daily-scale precipitation observation data covering multiple stations over many years and the CN05.1 gridded daily observation dataset; the atmospheric circulation index includes the atmospheric / ocean circulation index and the summer monsoon index; the global mean near-surface temperature data are the global mean near-surface temperature anomalies; and the global climate model data are the latest CMIP6 multi-model ensemble historical simulation and future scenario data.

3. The method for diagnosing, attributing, and estimating extreme precipitation changes according to claim 1, wherein: In step 2, daily climate averages are used to fill in missing records of daily-scale station precipitation observation data; based on the first-order conservative interpolation algorithm in the CDO toolbox, the precipitation data of the climate model are uniformly interpolated to the same 0.25°×0.25° spatial resolution grid as the observation data to ensure the consistency of the spatial resolution of the two.

4. The method for diagnosing, attributing, and estimating extreme precipitation changes according to claim 1, wherein: In step 3, the extreme precipitation index series was calculated using the R package "climdex.pcic".

5. The method for diagnosing, attributing, and estimating extreme precipitation changes according to claim 1, wherein: Based on the kernel density estimation (KDE), the probability density curves of each extreme precipitation index were drawn to show the probability distribution of each extreme precipitation index. The Kolmogorov-Smirnov test was applied to determine whether the probability distributions of the extreme precipitation index series in two different periods were significantly different. The iterative Mann-Kendall trend test method was used to test the long-term change trend of each extreme precipitation index.

6. The method for diagnosing, attributing, and estimating extreme precipitation changes according to claim 1, wherein: In step 5, the attribution analysis of extreme precipitation trend changes specifically includes the following steps: Step 51, calculate the regional average extreme precipitation index on an annual scale; convert the atmospheric circulation index and the global average near-surface temperature anomaly to an annual scale; Step 52: Use the Spearman rank correlation analysis method to calculate the correlation coefficient between the annual regional average extreme precipitation index, the annual atmospheric circulation index, and the global average near-surface temperature anomaly, and perform a significance test to preliminarily identify the statistical correlation between the extreme precipitation index, the circulation index, and global warming: Where: represents the difference in rank between the i-th observation value in the two variables, n represents the sample size; represents the rank of the i-th observation in variables X and Y; ρ s ∈[-1,1] represents the correlation coefficient. A positive result indicates positive correlation, and a negative result indicates negative correlation. The closer the absolute value is to 1, the stronger the correlation. For large samples, if there is a significant rank correlation between two variables, the t distribution is used for approximation:

7. The method for diagnosing, attributing, and estimating extreme precipitation changes according to claim 1, wherein: In step 6, the prediction of future extreme precipitation scenarios specifically includes the following steps: Step 61: extract the CMIP6 multi-model ensemble precipitation data, unify the time range, and use the spatial interpolation method to spatially interpolate the precipitation data output by the climate model so that its spatial resolution is consistent with the observation value; Step 62: Using the multidimensional bias correction method, bias correction is performed on the large-scale climate model based on the consideration of the "inconsistency" problem of the correlation structure, thereby obtaining a gridded dataset of future precipitation under different greenhouse gas emission scenarios with spatiotemporal correlation.

8. The method for diagnosing, attributing, and estimating extreme precipitation changes according to claim 7, wherein: In step 62, the quantile increment mapping (QDM) bias correction method is used to perform a separate one-dimensional bias correction on the climate variables in each model grid, so that the statistical characteristics are as consistent as possible with the observed data and the variation trend of the climate variables is preserved. The correlation structure of the precipitation data between the model grids is reconstructed using the R2D2 method based on rank reordering, which specifically includes the following steps: Step 621: Based on the cumulative distribution function F (t) m,p , calculate the model simulation value x at a certain time t in the future m,p The cumulative probability τ of (t) m,p (t), determine the quantile position of the model simulation value in the CDF: t m,p (t)=F (t) m,p [x] m,p (t)],τ m,p (t)∈{0,1} Step 622: Calculate the model simulation value x m,p (t) The change ratio in the future relative to the historical period Δ m (t): Where: is the inverse function of the CDF of the original model data in the historical period; Step 623: The cumulative probability τ m,p (t) Substitute the inverse function of the historical CDF of the observed data into To correct for system deviations: Step 624: Correct the historical deviation value x o:m,h:p (t) multiplied by the relative change Δ m (t) Get the corrected pattern data x m,c (t) Ensure that the relative trends of the models for future climate change are preserved while correcting for biases: x m,c (t)=x o:m,h:p (t)Δ m (t) Step 625: Using the rank matrix of the observed data as a reference, the R2D2 rank reordering method is used to reorder the model data that has been corrected in step 624 to reconstruct the correlation structure of precipitation data between model grids, thereby obtaining future precipitation datasets for different scenarios with spatiotemporal correlation. The R2D2 rank reordering method is based on the R function package "MBC".

9. A device for diagnosing, attributing and estimating extreme precipitation changes, characterized in that: include: one or more processors; a storage device for storing one or more programs and user data; When the one or more programs are executed by one or more processors, the one or more processors implement the extreme precipitation change diagnosis, attribution and prediction method according to any one of claims 1 to 8.

10. A storage medium for diagnosing, attributing and estimating extreme precipitation changes, characterized in that: A computer program is stored thereon, and when the program is executed by a processor, the method for diagnosing, attributing and predicting extreme precipitation changes as described in any one of claims 1 to 8 is implemented.

Images