Multi-source data fusion-based drought and flood sudden change event identification and intensity quantification method

Abstract

The invention discloses a drought and flood sudden change event identification and intensity quantification method based on multi-source data fusion, and the method comprises the steps: comprehensively employing a Mann-Kendall trend analysis method, an R / S analysis method, an M-K sudden change detection method, a sliding T detection method, a Morlet wavelet analysis method, an inverse distance interpolation method and other data analysis methods and models through the multi-source data fusion; the spatial-temporal change characteristics of drought and flood sudden turning are systematically and accurately analyzed. The method can comprehensively and accurately analyze the spatial and temporal change characteristics of drought and flood sudden change, reliably predicts the future trend, provides an important scientific basis for regional water resource management and agricultural disaster prevention and reduction, and provides technical support for drought and flood disaster response in related fields.

Description

[0001] This application is a divisional application. The original application was entitled "A Method for Characteristic Analysis and Trend Prediction of Drought and Flood Transition Based on Multi-Source Data Fusion", with application number 202510819784.9 and application date of June 19, 2025. Technical Field

[0002] This invention belongs to the field of hydrological and meteorological analysis technology, and relates to a method for analyzing the characteristics and predicting trends of rapid drought-flood transitions based on multi-source data fusion. Background Technology

[0003] Rapid drought-flood transitions in a watershed refer to the phenomenon of a rapid shift from a drought state to a flood state, or vice versa, within a short period of time. Against the backdrop of global warming, the seasonal variations of these transitions are becoming increasingly frequent. Existing research indicates that "rapid drought-flood transitions" and "coexistence of drought and flood" represent the annual and intra-annual seasonal variations in watershed precipitation. Frequent occurrences of these transitions can lead to severe disasters and significant economic losses. Due to the suddenness and abruptness of these transitions, and the time-lag effect of drought relief and drainage efforts on these events, clarifying the spatiotemporal characteristics of rapid drought-flood transitions in a watershed is crucial for disaster prevention and control.

[0004] Against the backdrop of global climate change, extreme weather events are occurring frequently, and the phenomenon of rapid shifts between drought and flood has a serious impact on agriculture, water resource management, the ecological environment, and the socio-economic aspects.

[0005] Currently, research on the phenomenon of rapid drought-flood transitions mainly focuses on the analysis of single data sources (such as meteorological or hydrological data), lacking the fusion analysis of multi-source data. Traditional methods struggle to comprehensively and accurately capture the spatiotemporal characteristics of rapid drought-flood transitions, and their predictive capabilities are limited, failing to meet the needs of refined management and forward-looking decision-making for drought and flood disasters in practical applications. Therefore, developing a method for analyzing the characteristics and predicting trends of rapid drought-flood transitions based on multi-source data fusion has significant practical implications. Summary of the Invention

[0006]

[0005] In view of this, the purpose of the present invention is to provide a method for analyzing and predicting the characteristics and trends of rapid drought-flood transitions based on multi-source data fusion, so as to solve the problems of single characteristics and insufficient predictive ability of existing methods for analyzing and predicting the characteristics and trends of rapid drought-flood transitions.

[0007] To achieve the above objectives, the present invention provides the following technical solution:

[0008] A method for analyzing the characteristics and predicting trends of rapid shifts between drought and flood based on multi-source data fusion includes the following steps: Step 1: Obtain multi-source data of a certain time series within the assessment area;

[0009] Step 2: Based on multi-source data for this region, calculate the drought and flood indices for different areas;

[0010] Step 3: Use the entropy weight method to fuse the drought and flood indices calculated in Step 2 to construct the comprehensive drought and flood index (SCDI);

[0011] Step 4: Calculate the Z-score of Mann-Kendall trend analysis and the Hurst exponent of R / S analysis. The Z-score is used to test the significance of the time series, and the Hurst exponent is used to predict the persistence of future changes in the rapid shift between drought and flood in the region. The Z-score and Hurst exponent are used together to assess the temporal variation characteristics and future trends of the rapid shift between drought and flood in the region.

[0012] Step 5: Calculate the MK mutation test and plot the UF and UB statistics curves. Combine the sliding T test to identify possible mutation years. Use Morlet wavelet analysis to identify the cycle of change of the drought-flood transition index.

[0013] Step 6: Use inverse distance interpolation to interpolate and obtain the frequency of sudden shifts between drought and flood in the region.

[0014] As one of the preferred technical solutions, in step 1, the multi-source data includes: meteorological data, hydrological data, remote sensing data, and socio-economic data.

[0015] As one of the preferred technical solutions, in step 1, if there is missing data, linear interpolation is used to supplement the missing data to ensure the continuity and integrity of the data, and the difference in dimensions is eliminated through data standardization.

[0016] As one of the preferred technical solutions, in step 2, the drought and flood indices for different fields include: meteorological drought and flood index (SPEI, SPI), hydrological drought and flood index (SRI), agricultural drought and flood index (VSWI, VCI) and socio-economic drought and flood index (SEDI).

[0017] As one of the further preferred technical solutions, in step 2, the meteorological drought and flood index is calculated by the standardized difference between precipitation and potential evapotranspiration or average precipitation; the hydrological drought and flood index is calculated in the same way as the meteorological drought and flood index, but is based on runoff data; the agricultural drought and flood index is calculated by the ratio of normalized vegetation index to surface temperature or the historical relative proportion of normalized vegetation index; and the socioeconomic drought and flood index is based on the distribution fitting and standardized transformation of the difference between water supply and demand data.

[0018] As a further preferred technical solution, the meteorological drought and flood index includes the Standardized Precipitation Evapotranspiration Index (SPEI) and the Standardized Precipitation Index (SPI), and the calculation method is as follows:

[0019] Potential evaporation of PET i:

[0020]

[0021] In the formula: i represents the months of the year, with values ​​ranging from 1 to 12; K is a correction factor calculated based on latitude and longitude; T is the monthly average temperature; I is the annual total heating index; and M is a coefficient determined by I.

[0022] The difference D between precipitation and potential evapotranspiration i :

[0023] D i =P i -PET i

[0024] In the formula: P i For the precipitation in month i, PET i This represents the potential evapotranspiration in month i.

[0025] For the difference sequence D i Normalize the data to obtain SPEI time series data;

[0026] Use ArcGIS to perform inverse distance spatial interpolation to obtain SPEI raster data, and classify drought and flood levels according to SPEI size;

[0027]

[0028] In the formula: t is the probability weighted interval; c0 is the first constant, with a value of 2.515517; c1 is the second constant, with a value of 0.802853; c2 is the third constant, with a value of 0.010328; d1 is the fourth constant, with a value of 10432788; d2 is the fifth constant, with a value of 0.189269; d3 is the sixth constant, with a value of 0.001308; ln(·) is the natural logarithm function; F(x) is the cumulative probability of precipitation distribution calculated based on the Γ distribution; x is the precipitation amount in a certain period; β and γ are the shape and scale parameters of the Γ distribution function.

[0029] The SPI time series within the evaluation area was calculated using the above steps, and inverse distance interpolation was performed using ArcGIS to convert it into raster data.

[0030] The SPI drought and flood classification is the same as the SPEI.

[0031] As a further preferred technical solution, the hydrological drought and flood index is the Standardized Runoff Index (SRI), which is calculated in the same way as the SPI and is used to describe hydrological drought and flood; the SRI drought and flood level classification is the same as the SPEI.

[0032] As a further preferred technical solution, the Agricultural Flood and Drought Index (VSWI) is an important indicator characterizing the water supply status of vegetation. It is used to assess the water conditions for vegetation growth and regional droughts and floods, and the calculation formula is as follows:

[0033] VSWI i =NDVI i / LST i

[0034] In the formula: i represents the month of the year, ranging from 1 to 12; NDVI is the monthly normalized vegetation index; LST is the average land surface temperature in month i; drought and flood levels are classified according to VSWI size.

[0035] The Vegetation Status Index (VCI) effectively reflects the vegetation growth status, and thus the agricultural drought and flood situation. The VCI calculation formula is as follows:

[0036]

[0037] In the formula: i represents the month within the year, with a value ranging from 1 to 12; NDVI i The NDVI value for the i-th month of a given year; NDVI max and NDVI min These represent the maximum and minimum NDVI values ​​for the i-th month over many years; drought and flood levels are classified according to the size of the VCI index.

[0038] As one of the further preferred technical solutions, the difference between water supply data and water demand data is fitted to a distribution using a nonparametric kernel density estimation method, and the cumulative frequency distribution of the difference is transformed into a standard normal distribution using an equal probability transformation to obtain the socioeconomic drought and flood index (SEDI), calculated as follows:

[0039] The probability density function f(e) for nonparametric kernel density estimation is:

[0040]

[0041] Where: n is the total number of samples; h is the window width; e is the difference between water supply data and water demand data; e i Let be the i-th sample value of the difference; K(·) is the kernel function.

[0042] Kernel density estimation is performed using the Gaussian kernel function, with the default window width h in MATLAB. The expression for the Gaussian kernel function K(x) is as follows:

[0043]

[0044] In the formula: x represents the degree of difference between different sample values ​​and the current value in a relative sense.

[0045] Drought and flood levels are classified according to the size of the SEDI index.

[0046] As one of the preferred technical solutions, in step 3, the weight of the Comprehensive Drought and Flood Index (SCDI) is determined using the entropy weight method, specifically including:

[0047] Standardize each indicator;

[0048] Weights are assigned based on the degree of dispersion of each indicator;

[0049] A linear weighted model was constructed to integrate various drought and flood indices.

[0050] As one of the further preferred technical solutions, the specific process of step 3 is as follows:

[0051] The entropy weight method is used to fuse the drought and flood indices calculated in step 2 to construct the comprehensive drought and flood index (SCDI). The specific calculation formula is as follows:

[0052] SCDI i =w1×SPEI i +w2×SPI i +w3×SRI i +w4×VSWI i +w5×VCI i +w6×SEDI i

[0053] In the formula: i represents each month of the year, with a value range of 1 to 12; w1 to w6 are the weight values ​​of the six indicators, respectively.

[0054] As one of the preferred technical solutions, in step 4, when calculating the Z value using the Mann-Kendall trend test method, a weighted moving average method is used to preprocess the data, giving higher weight to recent data to enhance the sensitivity to recent drought and flood trends. In the optimized mutation test algorithm, the sliding window technique adaptively adjusts the window size according to the fluctuation characteristics of the data, using a smaller window in areas of drastic data fluctuation and a larger window in relatively stable areas to improve the accuracy of mutation detection.

[0055] As one of the further preferred technical solutions, the specific process is as follows:

[0056] Calculate the Z value of the drought-flood rapid transition index to test the significance of the time series. Z>0 indicates that the series shows an upward trend, meaning that the drought-flood rapid transition situation has an increasing trend in a certain direction (such as "drought to flood" or "flood to drought"); Z<0 indicates that the series shows a downward trend, that is, the characteristics of drought-flood rapid transition are weakening; when |Z| is greater than or equal to 1.64, 1.96, and 2.58, the significance tests with confidence levels of 90%, 95%, and 99% are passed respectively, indicating that the trend has a high credibility;

[0057] Select the time series data of the drought-flood rapid transition index after standardized processing as the basis to calculate the Hurst index (H) of the R / S method; if 0<H<0.5, it means that the time series has anti-persistence and the future trend is opposite to the past; if H>0.5, it means that the time series has persistence and the future trend is the same as the past; if H = 0.5, it indicates that the time series is random and has no obvious trend characteristics; judge the future continuous or reverse trend of the drought-flood rapid transition phenomenon according to the calculated Hurst index as a reference basis for prediction.

[0058] As one of the preferred technical solutions, in step 5, when there is uncertainty in the sliding test result (such as the test statistics of multiple years are close to the mutation condition), reduce the sliding window step size and re-conduct the test, and combine the M-K statistic curves UF and UB to jointly enhance the accuracy of mutation identification.

[0059] As one of the further preferred technical solutions, use the M-K mutation test method to draw the statistic curves UF and UB, and identify possible mutation years by observing the intersection of the two curves and whether the statistic at the intersection passes a specific significance test (such as the 95% significance level); at the same time, combine the sliding T test (the sliding test step size n = 5) to further verify and accurately identify the mutation point; if UF and UB intersect in a certain year and the test statistic of that year meets the mutation condition, and the sliding T test also supports the mutation judgment, then that year is the mutation year of the drought-flood rapid transition feature.

[0060] As one of the preferred technical solutions, in step 5, conduct an analysis of the change cycle of the drought-flood rapid transition index. If the Morlet wavelet analysis method identifies multiple periodic signals with similar intensities, comprehensively consider the regional climate historical data and the recent climate change trend to determine the most representative main cycle; the specific process of wavelet analysis includes:

[0061] Select wavelet scale and frequency parameters;

[0062] Analyze the periodic characteristics of the time series and extract the main periodic signal;

[0063] Determine the main cycle and secondary cycle of the drought-flood rapid transition phenomenon through variance.

[0064] As one of the further preferred technical solutions, wavelet transform is performed on the exponential sequence to analyze the real part and variance of the wavelet transform. The fluctuation of the real part reflects the periodic change characteristics of the rapid shift between drought and flood at different time scales, while the variance can be used to measure the intensity of each period. Vibration signals at different time scales are extracted, their coverage and intensity are determined, and the period with obvious peaks is identified as the main period. The periodic law of the rapid shift between drought and flood is derived for long-term prediction and law summarization.

[0065] As one of the preferred technical solutions, step 6 specifically involves the following process: Inverse distance interpolation. Based on the geographical location of each rain gauge station and the frequency of abrupt shifts between drought and flood, and using the inverse distance weighting principle, stations that are closer in distance have a greater influence on the interpolation point and thus a higher weight. This method generates a spatial distribution map, which visually displays the spatial distribution differences of abrupt shifts between drought and flood in the region.

[0066] The formula for inverse distance interpolation is as follows:

[0067]

[0068] In the formula: Z j It is the predicted value of the frequency of sudden shifts between drought and flood at the spatial interpolation point j; d i,j Z is the distance between a known point i and an interpolation point j in space; i is the frequency value of sudden changes between drought and flood at the i-th known point; n is the number of known sample points involved in the interpolation calculation.

[0069] As one of the preferred technical solutions, in step 6, if the spatial distribution result obtained by the inverse distance interpolation method has local anomalies (such as isolated high or low value points), the form of the distance weight function is adjusted (such as changing from a simple inverse proportional function to an exponential decay function) or the weight of local data is increased, and the interpolation is recalculated to make the spatial distribution result more reasonable.

[0070] The beneficial effects of this invention are as follows:

[0071] This invention discloses a method for analyzing the characteristics and predicting the trend of abrupt shifts in drought and flood based on multi-source data fusion, comprising the following steps: Step 1: Obtain multi-source data of a certain time series within the assessment area; Step 2: Calculate drought and flood indices for different fields based on the multi-source data of the area; Step 3: Use the entropy weight method to fuse the drought and flood indices calculated in Step 2 to construct a comprehensive drought and flood index (SCDI); Step 4: Calculate the Z-value of Mann-Kendall trend analysis and the Hurst index of R / S analysis. The Z-value is used to test the significance of the time series, and the Hurst index is used to predict the persistence of future changes in abrupt shifts in drought and flood in the region. The Z-value and Hurst index are used together to assess the temporal variation characteristics and future trends of abrupt shifts in drought and flood in the region; Step 5: Use the MK mutation test to plot the UF and UB curves, and combine them with the sliding T test to identify possible mutation years. Use the Morlet wavelet analysis method to identify the cycle of change of the drought and flood shift index; Step 6: Use the inverse distance interpolation method to interpolate and obtain the frequency of abrupt shifts in drought and flood in the region. This invention can comprehensively and accurately analyze the spatiotemporal variation characteristics of rapid shifts between drought and flood, and reliably predict their future trends, providing important scientific basis for regional water resource management and agricultural disaster prevention and mitigation, and providing technical support for related fields to cope with drought and flood disasters.

[0072] Compared with known existing technologies, the present invention has the following advantages:

[0073] 1. By fusing multi-source data, the spatiotemporal variation characteristics of rapid drought-flood transitions can be comprehensively and accurately identified, overcoming the limitations of analysis based on a single data source.

[0074] 2. By comprehensively utilizing various data analysis methods and models such as Mann-Kendall trend analysis, R / S analysis, MK mutation test, sliding T test, Morlet wavelet analysis, and inverse distance interpolation, the spatiotemporal variation characteristics of rapid drought-flood transitions are systematically and accurately analyzed. Compared with traditional single methods, this improves the reliability and comprehensiveness of the analysis results.

[0075] 3. By constructing predictive models, we can predict the future trends of rapid shifts between drought and flood in the region, providing relevant departments with sufficient time to formulate response strategies in advance, effectively reducing losses caused by drought and flood disasters, and ensuring the stable development of the social economy and the safety of the ecological environment. Attached Figure Description

[0076] To make the objectives, technical solutions, and beneficial effects of this invention clearer, the following figures are provided for illustration:

[0077] Figure 1 This is a schematic diagram of the process of the present invention.

[0078] Figure 2This represents the frequency of SPEI drought phenomena.

[0079] Figure 3 The frequency of flooding phenomena is represented by the SPI.

[0080] Figure 4 This represents the frequency of drought phenomena in the SRI (Solar Response Index).

[0081] Figure 5 This represents the frequency of VSWI flood phenomena.

[0082] Figure 6 This represents the frequency of drought phenomena in the VCI (Various Indicators of Influence).

[0083] Figure 7 This represents the frequency of SEDI flooding phenomena.

[0084] Figure 8 This indicates the future trend of the overall drought-to-flood phenomenon in Shandong Province.

[0085] Figure 9 This indicates the future trend of the overall flood-to-drought phenomenon in Shandong Province.

[0086] Figure 10 This refers to the frequency of drought-to-flood transitions in Shandong Province.

[0087] Figure 11 This refers to the overall frequency of flooding turning into drought.

[0088] In the attached diagram, N represents a compass. Detailed Implementation

[0089] The preferred embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0090] This invention analyzes the characteristics and future trends of rapid drought-flood transitions in Shandong Province using multi-source data. The study area is Shandong Province. Located on the eastern coast of my country in the lower reaches of the Yellow River, Shandong Province has a mild climate with four distinct seasons, significant inter-annual precipitation variations, and uneven intra-annual precipitation distribution. In recent years, rapid drought-flood transitions have frequently occurred in Shandong Province, severely impacting agricultural production, water resource management, and the socio-economic situation. Therefore, developing a scientific and effective method for analyzing the characteristics and predicting trends of rapid drought-flood transitions is of great significance for disaster prevention and mitigation efforts in Shandong Province.

[0091] Example

[0092] Please refer to Figure 1 The method for analyzing the characteristics and predicting the trend of rapid shifts between drought and flood in Shandong Province provided in this embodiment includes the following steps:

[0093] Step 1: Obtain multi-source data of a certain time series within the assessment area, including meteorological data, hydrological data, remote sensing data, and socio-economic data.

[0094] The meteorological data time series covers the period from 1991 to 2020, sourced from 83 surface meteorological stations in Shandong Province. Where data was missing, linear interpolation was used to supplement the missing data to ensure continuity and completeness. Hydrological data uses monthly runoff data from Shandong Province from 1991 to 2018; remote sensing data utilizes MODIS imagery data from 2001 to 2020 (such as NDVI vegetation cover and LST land surface temperature), while Landsat 5 data was used for NDVI and LST data from 1991 to 2000; socioeconomic data includes gridded water supply, population density, and GDP data; other data include land use, transportation network, and atmospheric circulation data.

[0095] Step 2: Calculate the meteorological drought and flood index (SPEI, SPI), hydrological drought and flood index (SRI), agricultural drought and flood index (VSWI, VCI), and socioeconomic drought and flood index (SEDI) respectively.

[0096] By integrating multi-source data from meteorology, remote sensing, and historical disaster data, and coupling various algorithms and technologies, this study establishes a comprehensive drought and flood index (SCDI) for Shandong Province based on multi-year precipitation, soil moisture, meteorological, and remote sensing data. The SCDI is derived from standardized precipitation evapotranspiration index (SPEI), standardized precipitation index (SPI), standardized runoff index (SRI), vegetation water supply index (VSWI), vegetation status index (VCI), and socioeconomic drought and flood index (SEDI). This SCDI identifies and quantifies the phenomenon of rapid drought and flood transitions in Shandong Province. Specifically, the SPEI and SPI indices are calculated using meteorological station data, the SRI is calculated from runoff gridded data, the VSWI and VCI indices are calculated from NDVI and LST remote sensing data, and the SEDI data is derived from gridded water supply data and the Shandong Statistical Yearbook. Finally, the entropy weight method is applied to construct a comprehensive drought and flood index (SCDI) suitable for Shandong Province.

[0097] Step 3: Use the entropy weight method to fuse the drought and flood indices calculated in step (2) to construct the comprehensive drought and flood index (SCDI). The specific calculation formula is as follows:

[0098] SCDI i =w1×SPEI i +w2×SPI i +w3×SRI i +w4×VSWI i +w5×VCI i +w6×SEDI i

[0099] In the formula: W1 to W6 are the weight values ​​of the six indicators respectively. In this embodiment, the weight values ​​calculated by the entropy weight method are 0.3015, 0.2076, 0.2187, 0.1067, 0.0224 and 0.1431 respectively.

[0100] This embodiment defines the abrupt shift between drought and flood from a temporal composite perspective, namely, the sudden change between drought and flood in adjacent months. The magnitude index (S) of the abrupt shift is used to measure the intensity of this phenomenon, and the specific calculation formula is as follows:

[0101]

[0102] In the formula: S a The sum of the maximum and minimum values ​​in the comprehensive drought and flood index sequence is j = i + 1.

[0103] Based on the calculation results, the intensity levels of the rapid transition from drought to flood are divided into the following categories according to the magnitude of the rapid transition index: S≤-1.8 indicates extreme drought to flood, -1.8<S≤-1.2 indicates severe drought to flood, -1.2<S≤-0.8 indicates moderate drought to flood, -0.8<S≤-0.4 indicates mild drought to flood, -0.4<S≤0.4 indicates normal conditions, S>1.8 indicates extreme flood to drought, 1.2<S≤1.8 indicates severe flood to drought, 0.8<S≤1.2 indicates moderate flood to drought, and 0.4<S≤0.8 indicates mild flood to drought.

[0104] Step 4: Calculate the Z-score for Mann-Kendall trend analysis and the Hurst index for R / S analysis: The Z-score is used to test the significance of the time series, and the Hurst index is used to predict the persistence of future changes in the rapid shift between drought and flood in the region; the Z-score and Hurst index are used together to assess the temporal variation characteristics and future trends of the rapid shift between drought and flood in the region.

[0105] The Z-score of the drought-flood abrupt change index is calculated to test the significance of the time series. Z > 0 indicates an upward trend in the series, meaning that the abrupt change in drought-flood conditions is strengthening in a certain direction (e.g., "drought to flood" or "flood to drought"); Z < 0 indicates a downward trend in the series, meaning that the abrupt change characteristic is weakening. When |Z| is greater than or equal to 1.64, 1.96, and 2.58, the significance tests at confidence levels of 90%, 95%, and 99% are passed, respectively, indicating that the trend has high reliability.

[0106] Select the time series data of the drought-flood abrupt alternation index after standardization as the basis to calculate the Hurst index (H) of the R / S method. If 0 < H < 0.5, it indicates that the time series has anti-persistence and the future trend is opposite to the past; if H > 0.5, it indicates that the time series has persistence and the future trend is the same as the past; if H = 0.5, it indicates that the time series is random and has no obvious trend characteristics. Judge the future continuous or reverse trend of the drought-flood abrupt alternation phenomenon according to the calculated Hurst index, which serves as a reference basis for prediction.

[0107] Conduct a spatial overlay analysis on the Z values and H values of the statistical quantities of the drought-flood abrupt alternation phenomena at each level in Shandong Province. It can be seen that the overall development trend of the drought-flood abrupt alternation phenomena is towards a benign direction. Only in a small part of the areas in Qingdao City and Liaocheng City, the harmfulness of the overall drought-to-flood phenomenon has increased. When the Z value shows an insignificant trend change, the Hurst value can provide information about the potential long-term persistence or anti-persistence of the sequence. The combination of the two can more comprehensively grasp the trend characteristics of the drought-flood abrupt alternation and avoid misjudgment or omission that may exist in a single method.

[0108] Step 5: Calculate the Mann-Kendall mutation test method to draw the statistical quantities UF and UB curves, and jointly identify the possible mutation years in combination with the moving T test. Use the Morlet wavelet analysis method to identify the change period of the drought-flood abrupt alternation index.

[0109] Use the combined test algorithm based on the Mann-Kendall mutation test method and the moving T test, and integrate it into the data analysis software to draw the statistical quantities UF and UB curves. Set the moving test step size n = 5, and comprehensively scan and judge the possible mutation years according to the algorithm. For each year, calculate the test statistic in detail and compare it with the preset significance level. If the Mann-Kendall mutation test UF and UB curves of the drought-flood abrupt alternation intensity intersect in a certain year and the test statistic of the moving T test in that year meets the mutation condition (less than the specific significance threshold of 0.05), then determine that year as the mutation year and clarify the reliability of its mutation. The combined use of the Mann-Kendall mutation test and the moving T test effectively refines the mutation points and avoids misjudgment that may be caused by a single method.

[0110] Morlet wavelet analysis was used to select appropriate wavelet transform parameters, such as wavelet scale and center frequency, based on the time series characteristics of the drought-flood abrupt change index. Through multiple experiments and comparative analyses, the optimal parameter combination was determined to improve the accuracy and resolution of the periodic analysis. Wavelet transform was applied to the frequency of drought-flood abrupt changes at different levels in Shandong Province. The real part and variance of the wavelet transform were analyzed in depth. By extracting vibration signals at different time scales, the coverage and intensity of each period were accurately determined. In the wavelet real part plot, white areas represent positive real parts, indicating a high frequency of drought-flood abrupt changes; black areas represent negative real parts, indicating a low frequency. The wavelet variance plot shows the periodic changes of drought-flood abrupt changes at different time scales. During the analysis, the wavelet transform real part plot and variance plot were drawn to visually display the periodic analysis results, facilitating researchers' understanding and analysis.

[0111] The frequency of sudden shifts between drought and flood in this region was obtained by interpolating the spatial data using the inverse distance interpolation method.

[0112] On the ArcGIS platform, inverse distance interpolation was used to spatially interpolate the frequency of abrupt drought-flood transition events at various rain gauge stations to identify such events in Shandong Province. During the interpolation process, parameters such as the distance attenuation coefficient were adjusted appropriately based on the geographical characteristics and rain gauge distribution of Shandong Province. In areas with complex terrain or sparse rain gauge distribution, the distance attenuation coefficient was appropriately reduced to improve interpolation accuracy; in areas with relatively flat terrain and evenly distributed rain gauges, the distance attenuation coefficient was appropriately increased to speed up calculations. Through interpolation calculations, a spatial distribution map of abrupt drought-flood transitions was generated, visually presenting the spatial differences in the abrupt drought-flood transition phenomenon in Shandong Province.

[0113] from Figure 2 , Figure 3 It can be seen that there is no significant regional difference in meteorological drought in Shandong Province. The frequency of meteorological drought is relatively high in the central and southern parts of Shandong Province, while the frequency of meteorological drought is relatively low in the northwest and northeast regions. The frequency of meteorological floods gradually increases from the northwest to the southeast. Figure 4 This indicates that hydrological drought in Shandong Province exhibits a clear northwest-southeast increasing distribution pattern, with Linyi and Rizhao experiencing higher frequencies of drought. Figure 5 , Figure 6 The data shows that agricultural flooding in Shandong Province generally exhibits a northwest-southeast increasing distribution pattern, while areas with high frequency of agricultural drought are mainly concentrated in the northwest. Figure 7 The distribution pattern of flooding in Shandong Province shows a gradual decrease from northwest to southeast.

[0114] Figure 8 The data shows that the overall trend of drought turning into flood in Shandong Province is positive, with only a small number of areas facing an increased risk of this transition in the future. Figure 9This indicates that the overall trend of flooding turning into drought in Shandong Province is not significantly positive, with only a few areas showing a significant increase.

[0115] from Figure 10 , Figure 11 It can be seen that the phenomenon of rapid drought-flood transition in Shandong Province has regional and clustered characteristics. Both drought-to-flood and flood-to-drought events are spatially characterized by being more frequent in the southwest and less frequent in the northeast, and there are significant differences in the frequency of rapid drought-flood transitions among different regions.

[0116] Finally, it should be noted that the above preferred embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail through the above preferred embodiments, those skilled in the art should understand that various changes can be made to it in form and detail without departing from the scope defined by the claims of the present invention.

Claims

1. A method for identifying and quantifying the intensity of sudden drought-flood transition events based on multi-source data fusion, characterized in that, include: Step 1: Obtain multi-source data of a preset time series within the evaluation area; The multi-source data includes meteorological data, hydrological data, remote sensing data, and socio-economic data; Step 2: Based on the multi-source data, calculate the drought and flood index; the drought and flood index includes the Standardized Precipitation Evapotranspiration Index (SPEI), Standardized Precipitation Index (SPI), Standardized Runoff Index (SRI), Vegetation Water Supply Index (VSWI), Vegetation Status Index (VCI), and Socioeconomic Drought and Flood Index (SEDI). Step 3: Based on the entropy weight method, perform data fusion on the various drought and flood indices to obtain the comprehensive drought and flood index SCDI: SCDI i = w1×SPEI i +w2×SPI i +w3×SRI i +w4×VSWI i +w5×VCI i +w6×SEDI i In the formula, W1 to W6 are the weight values ​​of the six indicators, respectively; Step 4: Calculate the magnitude index S of the rapid shift from drought to flood: In the formula, S a The sum of the maximum and minimum values ​​in the comprehensive drought and flood index sequence is denoted as i, where i represents each month of the year and takes values ​​from 1 to 12, and j represents adjacent months. Based on the value of the drought-flood transition magnitude index S, and by comparing it with a preset intensity level threshold, the intensity quantification level of the drought-flood transition event is obtained; wherein, the preset intensity level threshold is specifically: S≤-1.8 indicates extreme drought turning into flood; -1.8<S≤-1.2 indicates severe drought turning into flood; -1.2<S≤-0.8 indicates moderate drought turning into flood; -0.8<S≤-0.4 indicates mild drought turning into flood; -0.4<S≤0.4 indicates normal conditions; S>1.8 indicates extreme flood turning into drought; 1.2<S≤1.8 indicates severe flood turning into drought; 0.8<S≤1.2 indicates moderate flood turning into drought; 0.4<S≤0.8 indicates mild flood turning into drought.

2. The method according to claim 1, characterized in that, The calculation process for each drought and flood index includes: The standardized precipitation evapotranspiration index (SPEI) and the standardized precipitation index (SPI) are calculated as follows: Potential evaporation of PET i : In the formula, i represents the months of the year, with values ​​ranging from 1 to 12; K is a correction factor calculated based on latitude and longitude; T is the monthly average temperature; I is the annual total heating index; and M is a coefficient determined by I. The difference D between precipitation and potential evapotranspiration i : D i =P i -PET i In the formula, P i For the precipitation in month i, PET i This represents the potential evapotranspiration in month i. For the difference sequence D i Normalize the data to obtain SPEI time series data; Use ArcGIS to perform inverse distance spatial interpolation to obtain SPEI raster data, and classify drought and flood levels according to SPEI size; In the formula: t is the probability weighted interval; c0 is the first constant, with a value of 2.515517; c1 is the second constant, with a value of 0.802853; c2 is the third constant, with a value of 0.010328; d1 is the fourth constant, with a value of 10432788; d2 is the fifth constant, with a value of 0.189269; d3 is the sixth constant, with a value of 0.001308; ln(·) is the natural logarithm function; F(x) is the cumulative probability of precipitation distribution calculated based on the Γ distribution; x is the precipitation amount in a certain period; β and γ are the shape and scale parameters of the Γ distribution function. The SPI time series within the evaluation area was calculated using the above steps, and inverse distance interpolation was performed using ArcGIS to convert it into raster data. The SPI drought and flood classification is the same as the SPEI. The hydrological drought and flood index is a standardized runoff index calculated using the same method as the SPI, used to describe hydrological drought and flood; the SRI drought and flood classification is the same as the SPEI. The agricultural drought and flood index is an important indicator of vegetation water supply status. It is used to assess the water conditions for vegetation growth and regional climate disasters such as drought and flood. The calculation formula is as follows: VSWI i =NDVI i / LST i In the formula, i represents the month of the year, ranging from 1 to 12; NDVI is the monthly normalized vegetation index; LST is the average land surface temperature in month i; drought and flood levels are classified according to VSWI size. The formula for calculating the Vegetation State Index (VCI) is as follows: In the formula, i represents the month within the year, with a value ranging from 1 to 12; NDVI i The NDVI value for the i-th month of a given year; NDVI max and NDVI min These are the maximum and minimum NDVI values ​​for the i-th month over many years; drought and flood levels are classified according to the size of the VCI index. The difference between water supply and demand data is fitted to a distribution using a nonparametric kernel density estimation method, and the cumulative frequency distribution of the difference is transformed into a standard normal distribution using an equal probability transformation to obtain the socioeconomic drought and flood index (SEDI). The calculation formula is as follows: The probability density function f(e) for nonparametric kernel density estimation is: Where: n is the total number of samples; h is the window width; e is the difference between water supply data and water demand data; e i Let be the i-th sample value of the difference; K(·) is the kernel function; Kernel density estimation is performed using the Gaussian kernel function, with the default window width h in MATLAB. The expression for the Gaussian kernel function K(x) is as follows: In the formula, x represents the degree of difference between different sample values ​​and the current value in a relative sense; The weights of the comprehensive drought and flood index are determined using the entropy weight method, specifically including: Standardize each indicator; Weights are assigned based on the degree of dispersion of each indicator; A linear weighted model was constructed to integrate various drought and flood indices.

3. The method according to claim 1, characterized in that, Also includes: Step 5: Calculate the Z-score of Mann-Kendall trend analysis and the Hurst exponent of R / S analysis. The Z-score is used to test the significance of the time series, and the Hurst exponent is used to predict the persistence of future changes in the rapid shift between drought and flood in the region. The Z-score and Hurst exponent are used together to assess the temporal variation characteristics and future trends of the rapid shift between drought and flood in the region. Step 6: Calculate the MK mutation test and plot the UF and UB statistics curves. Combine the sliding T test to identify possible mutation years. Use Morlet wavelet analysis to identify the cycle of change of the drought-flood transition index. Step 7: Use inverse distance interpolation to interpolate and obtain the frequency of sudden shifts between drought and flood in the region.

4. The method according to claim 3, characterized in that, Step 5 specifically includes: when calculating the Z-value using the Mann-Kendall trend test, the weighted moving average method is used to preprocess the data, giving higher weight to recent data to enhance the sensitivity to recent drought and flood trends; the optimized mutation detection algorithm, in the sliding window technique, adaptively adjusts the window size according to the fluctuation characteristics of the data, using a smaller window in areas of drastic data fluctuation and a larger window in relatively stable areas to improve the accuracy of mutation detection; The significance of the time series was tested by calculating the Z-score of the drought-flood transition index. Z>0 indicates an upward trend in the series, meaning that the trend of drought-flood transition from drought to flood or vice versa is increasing. Z<0 indicates a downward trend in the series, meaning that the characteristics of drought-flood transition are weakening. When Z is greater than or equal to 1.64, 1.96, and 2.58, the significance tests at confidence levels of 90%, 95%, and 99% are passed, respectively, indicating that the trend has high reliability. Select the time series data of the drought-flood rapid alternation index after standardization as the basis to calculate the Hurst index H of the R / S method; if 0 < H < 0.5, it indicates that the time series has anti-persistence and the future trend is opposite to the past; if H > 0.5, it indicates that the time series has persistence and the future trend is the same as the past; if H = 0.5, it indicates that the time series is random and has no obvious trend characteristics; judge the future continuous or reversal trend of the drought-flood rapid alternation phenomenon based on the calculated Hurst index, which is used as a reference basis for prediction.

5. The method according to claim 3, characterized in that, The specific content of step 6 includes: when there is uncertainty in the sliding test result, reduce the sliding window step size and re-conduct the test, and combine the M-K statistic curves UF and UB to jointly enhance the accuracy of mutation identification; Conduct an analysis of the change cycle of the drought-flood rapid alternation index. If the Morlet wavelet analysis method identifies multiple periodic signals with similar intensities, comprehensively consider the regional climate historical data and the recent climate change trend to determine the most representative main cycle; the specific process of wavelet analysis includes: Select wavelet scale and frequency parameters; Analyze the periodic characteristics of the time series and extract the main periodic signal; Determine the main cycle and secondary cycle of the drought-flood rapid alternation phenomenon through variance.

6. The method according to claim 3, characterized in that, The specific content of step 7 includes: based on the geographical locations and drought-flood rapid alternation frequency data of each rain gauge station, according to the inverse distance weighting principle, the closer the station is, the greater the influence on the interpolation point and the higher the weight; generate a spatial distribution map through this method to visually display the spatial distribution differences of the drought-flood rapid alternation phenomenon in the region; Among them, the calculation formula of the inverse distance interpolation method is as follows: In the formula: Z j It is the predicted value of the frequency of sudden shifts between drought and flood at the spatial interpolation point j; d i,j Z is the distance between a known point i and an interpolation point j in space; i is the frequency value of sudden changes between drought and flood at the i-th known point; n is the number of known sample points involved in the interpolation calculation; If there are local anomalies in the spatial distribution result obtained by the inverse distance interpolation method, adjust the form of the distance weight function or increase the weight of local data, and re-conduct the interpolation calculation to make the spatial distribution result more reasonable.

Images