A Multi-Scale SPEI Prediction and Drought Monitoring Method Integrating Site and Grid Data

Abstract

This invention discloses a multi-scale SPEI prediction and drought monitoring method that integrates station and grid data, belonging to the field of meteorological disaster monitoring technology, and aims to integrate monthly gridded SPEI data. Grid Data and SPEI weather station data for the same period Station The data is fused. During the fusion process, in order to fully preserve the SPEI data of the meteorological stations... Station The characteristics of the meteorological stations are analyzed, and the number of stations during Anusplin interpolation is controlled. Grid points surrounding each meteorological station are selected, with those beyond 20km retained as additional points for the meteorological stations. Then, the selected grid points are fused with the meteorological stations, and an Anusplin meteorological interpolation model is used, with terrain features as covariates, to generate batches of 1km resolution SPEI at 1 / 3 / 6 / 12 / 24-month scales. GS Dataset. Based on this dataset, the abrupt changes, trends, and sustainability of meteorological drought in a specified region are analyzed.

Description

Technical Field

[0001] This invention relates to the field of meteorological disaster monitoring technology, and in particular to a multi-scale SPEI prediction and drought monitoring method that integrates station and grid data. Background Technology

[0002] Drought is a phenomenon characterized by an imbalance between water supply and demand caused by prolonged periods of abnormally low or no precipitation. It is one of the most significant meteorological disasters affecting the socio-economic and ecological environment. Against the backdrop of global warming, my country's karst regions, mainly located in the southwestern provinces, are experiencing a significant increase in the frequency and intensity of droughts. Therefore, monitoring the evolution of drought in the southwestern karst regions is of great value in guiding government disaster prevention and mitigation efforts.

[0003] Currently, the standardized precipitation evapotranspiration index (SPEI) is widely used in meteorological drought monitoring. This index not only considers the impacts of precipitation and evapotranspiration on meteorological drought but also has multi-scale characteristics. Commonly used SPEI datasets for meteorological drought monitoring include: ① SPEIbasev2.9 monthly dataset, with a spatial resolution of 0.5 degrees, a time span of 1901-2022, spatially covering the globe, and a time scale of 1-48 months; ② Based on the SPEIbasev2.7 monthly dataset, a 6-month SPEI dataset with a 1km resolution from 1901-2021 was generated by downscaling it to comprehensively consider environmental parameters; ③ SPEI-RF dataset, with a spatial resolution of 1km, a time span of 2001-2020, spatially covering China, and time scales of 1 / 3 / 6 / 9 / 12 / 24 months; ④ SPEI datasets generated by calculating the SPEI index based on meteorological stations and then using spatial interpolation. In summary, the 0.5° gridded SPEI dataset can reflect the long-term characteristics of drought, but it has certain limitations in characterizing the detailed information of drought changes. The downscaled SPEI dataset can reflect the local details of drought to a certain extent, but it has certain uncertainties due to the environmental changes and downscaling methods selected during downscaling. The 1km resolution SPEI dataset obtained by SPEI-RF can effectively reflect the details of drought and has high accuracy, but its time series is only from 2001 to 2020, which is relatively short and cannot fully reflect the evolution characteristics of meteorological drought before and after the ecological restoration project in the Southwest Karst region. The direct interpolation method of meteorological station SPEI has a certain impact on the accuracy of the interpolation results due to the location and density of the stations. Summary of the Invention

[0004] To address the shortcomings of existing SPEI datasets, this invention provides a multi-scale SPEI prediction and drought monitoring method that integrates station and grid data. The following technical solution is adopted:

[0005] A multi-scale SPEI prediction and drought monitoring method that integrates station and grid data includes the following steps:

[0006] Step 1: Meteorological station data processing. Obtain daily-scale meteorological datasets from meteorological stations within a specified time period, and synthesize the daily-scale data into monthly-scale meteorological data.

[0007] Step 2: Calculate the SPEI at 1 / 3 / 6 / 12 / 24-month scales for meteorological stations. Station value;

[0008] Step 3: Filter grid points that meet the set conditions and treat each grid point as a station; with the meteorological station in the study area as the center, filter grid points located outside the set distance of the meteorological station as the densification points of the station.

[0009] Step 4: Extract SPEI at 1 / 3 / 6 / 12 / 24-month scales. Grid Based on the results of step 3, extract SPEI values ​​at 1 / 3 / 6 / 12 / 24-month scales from the SPEIbase dataset that meet the conditional grid points. Grid value;

[0010] Step 5, merge grid SPEI Grid With site SPEI Station Based on the results of steps 2 and 4, several stations were first randomly selected from the meteorological stations as a validation dataset; then the remaining stations were fused with the grid points to obtain the SPEI at 1 / 3 / 6 / 12 / 24-month scales. GS Merge the point data and convert it to xls format;

[0011] Step 6: Anusplin interpolation generates 1km resolution SPEI at 1 / 3 / 6 / 12 / 24 month scales. GS Dataset.

[0012] By adopting the above technical solution, it is proposed to transform the SPEI baseV2.9 monthly scale grid with a resolution of 0.5° into a single SPEI base. Grid Data and SPEI weather station data for the same period Station The process involves merging the meteorological stations. During the merging process, it is crucial to fully preserve the SPEI meteorological station data. StationThe characteristics of the meteorological stations were analyzed, and the number of stations during Anusplin interpolation was controlled. Grid points surrounding each meteorological station were selected, with those beyond 20km retained as additional points for the meteorological stations. Then, the selected grid points were fused with the meteorological stations, and an Anusplin meteorological interpolation model was used, with terrain features as covariates, to generate batches of 1km resolution SPEI values ​​at 1 / 3 / 6 / 12 / 24-month scales. GS Dataset.

[0013] This technical solution not only makes full use of the observation data from meteorological stations, but also densifies the meteorological stations by selecting grid points, which can effectively improve SPEI. GS Data set accuracy; predicted SPEI at 1km resolution and 1 / 3 / 6 / 12 / 24-month scales. GS The dataset can effectively capture the occurrence, development, and end times of meteorological drought, and monitor its sudden changes, trends, and persistence. At the same time, this dataset has a long time series, which can effectively characterize the spatiotemporal variation features of meteorological drought in some specific regions, such as the karst region of Southwest China.

[0014] Optionally, in step 1, the daily-scale meteorological dataset includes variables such as precipitation, maximum temperature, minimum temperature, average temperature, average wind speed, and sunshine duration.

[0015] Optionally, in step 2, based on monthly meteorological data, firstly, the monthly potential evapotranspiration is calculated using the Penman-Monteith formula recommended by FAO-56; secondly, the difference between monthly precipitation and potential evapotranspiration is calculated; then, the cumulative water deficit over 1 / 3 / 6 / 12 / 24 months is calculated; finally, the SPEI at the 1 / 3 / 6 / 12 / 24-month scale is calculated. Station value.

[0016] Optionally, in step 3, the SPEIbase grid data of the study area at 0.5° is converted into point data.

[0017] Optionally, in step 6, based on the results of step 5, the SPEI values ​​at 1 / 3 / 6 / 12 / 24-month scales are respectively... GS The fusion point data was imported into SPSS, and batch files in .dat format were exported. Based on the Anusplin meteorological interpolation model, SPEI at 1 / 3 / 6 / 12 / 24-month scales was generated in batches, using the 1km resolution elevation of the study area as a covariate. GS Dataset.

[0018] Optionally, step 7, SPEI, may also be included. GS Interpolation results verification of fusion point data: The accuracy of the interpolation results is verified by selecting three indicators: correlation coefficient (r), bias (BIAS), and root mean square error (RMSE).

[0019] Optionally, step 8 may also be included, based on SPEI. GS Abrupt changes in meteorological drought were tested based on the annual SPEI of the specified region. GS -12 data were used to detect whether there were abrupt changes in the meteorological drought time series within a specified time period using the Mann-Kendall mutation test.

[0020] Optionally, step 9, based on SPEI, is also included. GS Analysis of meteorological drought variation trends and persistence, based on the annual SPEI of the specified region. GS Using data from -12, we analyzed the temporal variation characteristics of meteorological drought on an annual scale, and employed a combination of Theil-Sen Median and Mann-Kendall significance tests and the Hurst index to analyze the significance and persistence of the long-term trend of meteorological drought.

[0021] By employing the aforementioned technical solutions and utilizing methods such as the Mann-Kendall mutation test, Theil-Sen Median trend analysis, Mann-Kendall significance test, and Hurst index, it is possible to analyze the SPEI (Self-Performance Index) of specific regions, such as the Southwest Karst region. GS The spatiotemporal evolution characteristics are monitored.

[0022] In summary, the present invention has at least one of the following beneficial technical effects:

[0023] This invention provides a multi-scale SPEI prediction and drought monitoring method that integrates station and gridded data. It aims to integrate 0.5° resolution SPEIbaseV2.9 monthly-scale gridded SPEI data. Grid Data and SPEI weather station data for the same period Station The process involves merging the meteorological stations. During the merging process, it is crucial to fully preserve the SPEI meteorological station data. Station The characteristics of the meteorological stations were analyzed, and the number of stations during Anusplin interpolation was controlled. Grid points surrounding each meteorological station were selected, with those beyond 20km retained as additional points for the meteorological stations. Then, the selected grid points were fused with the meteorological stations, and an Anusplin meteorological interpolation model was used, with terrain features as covariates, to generate batches of 1km resolution SPEI values ​​at 1 / 3 / 6 / 12 / 24-month scales. GS Dataset. Finally, the Mann-Kendall mutation test, Theil-Sen Median trend analysis, Mann-Kendall significance test, and Hurst index were used to analyze the SPEI of the Southwest Karst region. GS The spatiotemporal evolution characteristics are monitored. Attached Figure Description

[0024] Figure 1 This is a flowchart illustrating a multi-scale SPEI prediction and drought monitoring method that integrates station and grid data according to the present invention.

[0025] Figure 2 SPEI is a specific embodiment of the present invention. GS Results verification charts (1 / 3 / 6 / 12 / 24-month scales);

[0026] Figure 3a and Figure 3b SPEI is a specific embodiment of the present invention. GS Comparison chart of monthly data between -03 and SPEIbase-03 in the Southwest Karst region;

[0027] Figures 4a-4d SPEI is a specific embodiment of the present invention. GS -12 mutation test diagram in the Southwest Karst region; Detailed Implementation

[0028] The present invention will be further described in detail below with reference to the accompanying drawings.

[0029] This invention discloses a multi-scale SPEI prediction and drought monitoring method that integrates site and grid data.

[0030] Reference Figure 1 A multi-scale SPEI prediction and drought monitoring method that integrates station and grid data includes the following steps:

[0031] Step 1: Meteorological station data processing. Obtain daily-scale meteorological datasets from meteorological stations within a specified time period, and synthesize the daily-scale data into monthly-scale meteorological data.

[0032] Step 2: Calculate the SPEI at 1 / 3 / 6 / 12 / 24-month scales for meteorological stations. Station value;

[0033] Step 3: Filter grid points that meet the set conditions and treat each grid point as a station; with the meteorological station in the study area as the center, filter grid points located outside the set distance of the meteorological station as the densification points of the station.

[0034] Step 4: Extract SPEI at 1 / 3 / 6 / 12 / 24-month scales. Grid Based on the results of step 3, extract SPEI values ​​at 1 / 3 / 6 / 12 / 24-month scales from the SPEIbase dataset that meet the conditional grid points. Grid value;

[0035] Step 5, merge grid SPEI Grid With site SPEIStation Based on the results of steps 2 and 4, several stations were first randomly selected from the meteorological stations as a validation dataset; then the remaining stations were fused with the grid points to obtain the SPEI at 1 / 3 / 6 / 12 / 24-month scales. GS Merge the point data and convert it to xls format;

[0036] Step 6: Anusplin interpolation generates 1km resolution SPEI at 1 / 3 / 6 / 12 / 24 month scales. GS Dataset.

[0037] It is proposed to use SPEIbaseV2 with a 0.5° resolution and a 9-month scale grid for SPEI. Grid Data and SPEI weather station data for the same period Station The integration process was carried out in order to fully preserve the SPEI meteorological stations. Station The characteristics of the meteorological stations were analyzed, and the number of stations during Anusplin interpolation was controlled. Grid points surrounding each meteorological station were selected, with those beyond 20km retained as additional points for the meteorological stations. Then, the selected grid points were fused with the meteorological stations, and an Anusplin meteorological interpolation model was used, with terrain features as covariates, to generate batches of 1km resolution SPEI values ​​at 1 / 3 / 6 / 12 / 24-month scales. GS Dataset.

[0038] This technical solution not only makes full use of the observation data from meteorological stations, but also densifies the meteorological stations by selecting grid points, which can effectively improve SPEI. GS Data set accuracy; predicted SPEI at 1km resolution and 1 / 3 / 6 / 12 / 24-month scales. GS The dataset can effectively capture the occurrence, development, and end times of meteorological drought, and monitor its sudden changes, trends, and persistence. At the same time, this dataset has a long time series, which can effectively characterize the spatiotemporal variation features of meteorological drought in some specific regions, such as the karst region of Southwest China.

[0039] In step 1, the daily meteorological dataset includes variables such as precipitation, maximum temperature, minimum temperature, average temperature, average wind speed, and sunshine duration.

[0040] In step 2, based on monthly meteorological data, the monthly potential evapotranspiration is first calculated using the Penman-Monteith formula recommended by FAO-56; then, the difference between monthly precipitation and potential evapotranspiration is calculated; next, the cumulative water deficit over 1 / 3 / 6 / 12 / 24 months is calculated; finally, the SPEI at the 1 / 3 / 6 / 12 / 24-month scale is calculated. Station value.

[0041] In step 3, the SPEIbase grid data of the study area at 0.5° is converted into point data.

[0042] In step 6, based on the results of step 5, the SPEI values ​​at the 1 / 3 / 6 / 12 / 24 month scales are respectively... GS The fusion point data was imported into SPSS, and batch files in .dat format were exported. Based on the Anusplin meteorological interpolation model, SPEI at 1 / 3 / 6 / 12 / 24-month scales was generated in batches, using the 1km resolution elevation of the study area as a covariate. GS Dataset.

[0043] It also includes step 7, SPEI GS Interpolation results verification of fusion point data: The accuracy of the interpolation results is verified by selecting three indicators: correlation coefficient (r), bias (BIAS), and root mean square error (RMSE).

[0044] It also includes step 8, based on SPEI GS Abrupt changes in meteorological drought were tested based on the annual SPEI of the specified region. GS -12 data were used to detect whether there were abrupt changes in the meteorological drought time series within a specified time period using the Mann-Kendall mutation test.

[0045] It also includes step 9, based on SPEI GS Analysis of meteorological drought variation trends and persistence, based on the annual SPEI of the specified region. GS Using data from -12, we analyzed the temporal variation characteristics of meteorological drought on an annual scale, and employed a combination of Theil-Sen Median and Mann-Kendall significance tests and the Hurst index to analyze the significance and persistence of the long-term trend of meteorological drought.

[0046] By employing methods such as the Mann-Kendall mutation test, Theil-Sen Median trend analysis, Mann-Kendall significance test, and Hurst index, the spatiotemporal evolution characteristics of SPEIGS in specific regions, such as the Southwest Karst region, can be monitored.

[0047] The following specific embodiments illustrate the implementation principle of a multi-scale SPEI prediction and drought monitoring method that integrates station and grid data according to the present invention:

[0048] Step 1: Meteorological data processing;

[0049] Obtain daily meteorological data from stations in the Southwest Karst region from 1982 to 2019, including variables such as precipitation, maximum temperature, minimum temperature, average temperature, average wind speed, and sunshine duration. Use Python programming to synthesize monthly meteorological data from the daily data.

[0050] Step 2: Calculate the SPEI at 1 / 3 / 6 / 12 / 24-month scales for the site Station value;

[0051] Based on the results of step 1, the monthly potential evapotranspiration was first calculated using the Penman-Monteith formula recommended by FAO-56; then, the difference between monthly precipitation and potential evapotranspiration was calculated; next, the cumulative water deficit over 1 / 3 / 6 / 12 / 24 months was calculated; finally, the SPEI at the 1 / 3 / 6 / 12 / 24 month scale was calculated. Station Value. The specific calculation formulas are as follows: 1.1-1.6

[0052] ①Calculate the monthly potential evapotranspiration (PET) of a site based on Penman-Monteith:

[0053]

[0054] In the formula, PET is the potential evapotranspiration, Δ is the slope of the saturated vapor pressure curve, Rn is the net surface radiation, G is the soil heat flux, γ is the wet-dry constant, T is the temperature, U2 is the average wind speed, and e is the mean wind speed. s e is the saturated vapor pressure. a This is the actual water vapor pressure.

[0055] ②Calculate the difference (D) between monthly precipitation and potential evapotranspiration at each station:

[0056] D i =P i -PET i (1.2)

[0057] In the formula, Pi is the precipitation in the i-th month, and PETi is the potential evapotranspiration in the i-th month.

[0058] ③ Construct a cumulative water surplus / deficit series at multiple time scales and calculate its probability distribution.

[0059]

[0060] In the formula, k is the monthly time scale, 1≤k≤24, and n represents a certain month.

[0061] ④ Use the Log-logistic distribution to... The water balance series is fitted with three parameters and normalized to calculate the cumulative probability at a given time scale.

[0062] The three-parameter Log-logistic probability density function is:

[0063]

[0064] In the formula, α, β, and γ represent the scale, shape, and position parameters, respectively, which can be obtained using the linear moments method. Based on the Log-logistic probability distribution function, the cumulative probability at a given time scale can be calculated as follows:

[0065]

[0066] ⑤ Standardize the probability distribution F(x) to obtain the SPEI value of the site.

[0067]

[0068] When P ≤ 0.5, P = 1 - F(x); when P > 0.5, P = 1 - P, and the sign of SPEI is reversed. In the formula, C0 = 2.515517, C1 = 0.802853, C2 = 0.010328, d1 = 1.432788, d2 = 0.189269, and d3 = 0.001308.

[0069] Step 3: Filter grid points that meet the criteria

[0070] The SPEIbase data at a 0.5° grid in the study area were converted into point data. Each grid point was considered as a station; grid points located more than 20 km away from the meteorological stations in the study area were selected as densification points for the meteorological stations.

[0071] Step 4: Extract SPEI at 1 / 3 / 6 / 12 / 24-month scales. Grid value

[0072] Based on the results of step 3, extract SPEI at 1 / 3 / 6 / 12 / 24-month scales from the SPEIbase data that meet the conditional grid points. Grid Values, spanning from 1982 to 2019;

[0073] Step 5: Merge grid SPEI Grid With site SPEI Station

[0074] Based on the results of steps 2 and 4, 100 meteorological stations were first randomly selected as the validation dataset; then the remaining stations were fused with the grid to obtain the SPEI at 1 / 3 / 6 / 12 / 24-month scales. Grid Merge the point data and convert it to xls format;

[0075] Step 6: Anusplin interpolation generates 1km resolution SPEI at 1 / 3 / 6 / 12 / 24 month scales. Grid Dataset

[0076] Based on the results of step 5, the SPEI values ​​at 1 / 3 / 6 / 12 / 24-month scales were respectively... Grid The fusion point data was imported into SPSS, and batch files in .dat format were exported. Based on the Anusplin meteorological interpolation model, SPEI at 1 / 3 / 6 / 12 / 24-month scales was generated in batches, using the 1km resolution elevation of the study area as a covariate. Grid Dataset;

[0077] Step 7: SPEI GS Dataset Result Validation

[0078] The correlation coefficient (r), mean absolute error (MAE), and root mean square error (RMSE) are used to verify the accuracy of the interpolation results. Specific formulas 1.7-1.9 are as follows:

[0079]

[0080] In the formula, xi represents the station SPEI. Station , yi is SPEI GS Interpolation result, and is the average value, and n is the number of samples.

[0081] Step 8: Based on SPEI GS meteorological drought abrupt change test

[0082] Based on annual SPEI of the Southwest Karst Region GS -12 data points were used to detect the presence of abrupt change points in the meteorological drought time series from 1982 to 2019 using the Mann-Kendall mutation test; specific formulas 1.10 to 1.14 are as follows:

[0083] ① Constructing the order column:

[0084]

[0085] In the formula

[0086] ② Define the statistic:

[0087]

[0088] In the formula

[0089]

[0090] UFk is a statistical sequence calculated in chronological order of the time series. Given a significance level α, if UFk > Ua, it indicates a significant trend in the sequence. The time series is then reversed and UBk is calculated, with UBk set to Ua. k =UF k (k = n, n-1, ..., 1). Given a significance level α = 0.05, then U0.05 = ±1.96. If UFk and UBk exceed the critical line, it indicates a significant upward or downward trend. If the two curves UFk and UBk intersect between the critical lines, then the intersection point is the abrupt change point of the time series.

[0091] Step 9: Based on SPEI GS Analysis of the changing trends and persistence of meteorological drought

[0092] Based on annual SPEI of the Southwest Karst Region GS Using data from -12, we analyzed the temporal variation characteristics of meteorological drought from 1982 to 2019, and before the ecological restoration project (1982-1999) and after (2000-2019). We employed a combination of Theil-Sen Median trend analysis, Mann-Kendall significance test, and Hurst index to analyze the significance and persistence of the long-term trend of meteorological drought. The specific calculation process is as follows:

[0093] ①The Theil-Sen Median trend degree β is calculated using formula 1.15 as follows:

[0094]

[0095] In the formula, β is the per-pixel SPEI. GS The trend of -12, where i and j are time series; SPEI i and SPEI j SPEI of pixels at times i and j respectively GS -12 value; when β>0, it indicates that the time series is on an upward trend, and vice versa.

[0096] ② The Mann-Kendall method for determining the significance of a trend is shown in formulas 1.16 to 1.19 below:

[0097] Define the test statistic Z

[0098]

[0099] In the formula,

[0100]

[0101] Where: Z is the normal distribution statistic, Var(S) is the variance, and sgn is the sign function. At a given confidence level α, if |Z| ≥ Z 1-α / 2 , it indicates that there is a significant change in the time series at the α level. To judge the significance of the change trend of the SPEI GS -12 time series at the confidence level of α = 0.05.

[0102] ③ The Hurst exponent is used to judge the persistence of the change trend. The formulas 1.20 to 1.24 are as follows:

[0103] Define the mean sequence:

[0104] Cumulative deviation:

[0105] Range:

[0106] Standard deviation:

[0107] Calculation of the Hurst exponent: For the ratio If there is the following relationship R / S ∝ τ H , then the value of H can be obtained by fitting with the formula 1.24 in the double logarithmic coordinate system (ln(τ), ln(R / S)).

[0108] ln(R / S) = ln(c) + Hln(τ) (1.24)

[0109] In the formula, c is a constant and H is the Hurst exponent. When H = 0.5, it indicates that the SPEI GS -12 sequence is a random sequence, without persistence and no dependence on past trends; when 0 < H < 0.5, it indicates that the future change situation of the SPEI GS -12 is opposite to the past, and the smaller H is, the stronger the anti-persistence; when 0.5 < H < <0.75, it indicates that the future change situation of the SPEI GS -12 is consistent with the past, with weak persistence; when H ≥ 0.75, it indicates that the future change situation of the SPEI GS -12 is consistent with the past, with strong persistence, and the larger H is, the stronger the persistence.

[0110] To verify the results of this invention, a 1km resolution SPEI dataset at 1 / 3 / 6 / 12 / 24-month scales in the southwestern karst region was generated, and the results were evaluated using the verification dataset. The results are as GS shown. From Figure 2 the Figure 2It can be seen that the SPEI at the 1 / 3 / 6 / 12 / 24-month scale predicted using the method of this invention... GS The dataset performs well in terms of accuracy, R 2 The PMI ranged between 0.595 and 0.760, RMSE between 0.485 and 0.624, and MAE between 0.363 and 0.494. (Based on SPEI...) GS Taking -03 as an example, it is compared with the monthly data of SPEIbase-03 in eight southwestern provinces, and the results are shown in Figure 3. As can be seen from Figure 3, the SPEI predicted using this invention... GS The -03 data shows a high degree of consistency with the monthly variation trend of SPEIbase-03 data in the eight southwestern provinces (r>0.95, RMSE<0.3, MAE<0.3), effectively characterizing the aridity-humidity variation. Taking the drought event that occurred in the karst region of southwestern China in 2009-2010 as an example, SPEI... GS -03 is used to reflect the monthly evolution trend of the drought and is compared with SPEIbase-03 data. This allows us to understand the approximate evolution of the drought event in the Southwest Karst region from 2009 to 2010: it occurred in Guizhou Province in July 2009; the drought area gradually expanded and intensified from August to September; subsequently, from October to March 2010, the drought area further expanded, reaching its most severe stage; starting in April 2010, the drought area gradually decreased and the intensity eased; it essentially ended by June. The invention predicts the SPEI... GS The drought evolution and spatial distribution patterns reflected by the -03 data are in good agreement with the SPEIbase-03 data, indicating that the SPEI data predicted by the method of this invention are accurate. GS The dataset has high accuracy and reliability. From Figures 4a-4d It can be seen that the alternation of dry and wet seasons was quite pronounced in the Southwest Karst region from 1982 to 2019, with intersections of UF and UB curves, showing some differences in their trends; after abrupt changes occurred in most areas, the UF and UB curves did not exceed ±1.96, indicating insignificant trends. Figure 4aTaking Guangxi Zhuang Autonomous Region as an example, the period before 1995 was mainly arid. Between 1988 and 1992, the UF curve was less than -1.96 for four years, indicating a significant trend towards drought during this period. From 1996 to 2006, the UF curve was greater than 0 and intersected with the UB curve multiple times, but the trend was not significant. From 2007 to 2014, the UF curve was mainly less than 0, intersecting with the UB curve in 2011, and then showed an upward trend until 2019, but the trend did not pass the 0.05 significance test. The changing trends and persistence of meteorological drought in the Southwest Karst region at different times show significant spatial differences. From 1982 to 2019, meteorological drought in the study area was mainly characterized by weak, persistent, and insignificant decrease (43.34%), weak, persistent, and insignificant increase (28.61%), and strong, persistent, and insignificant decrease (17.96%). Among them, the areas with weak, persistent, and insignificant decrease were mainly distributed in western Guizhou, eastern and southern Yunnan, areas other than western Sichuan, and parts of Hubei, Hunan, and Chongqing; the areas with weak, persistent, and insignificant increase were concentrated in Guangxi, Guizhou, Hunan, and Guangdong; and the areas with strong, persistent, and insignificant decrease were mainly distributed in Yunnan and western Sichuan. Before the ecological restoration project from 1982 to 1999, meteorological drought in the western part of the study area mainly showed a weak, persistent, and insignificant increase (30.88%), while in the eastern part it mainly showed a strong, persistent, and insignificant increase (30.49%). After the ecological restoration project from 2000 to 2019, meteorological drought in the southern and northern parts of the study area mainly showed a weak, persistent, and insignificant increase (44.86%), while in the western part it mainly showed a strong, persistent, and insignificant decrease (15.91%). In the border area of ​​Guizhou, Guangxi, and Hunan, it mainly showed a strong, persistent, and insignificant increase (13.09%), while in the northeastern and southwestern parts it mainly showed a weak, persistent, and insignificant decrease (12.51%).

[0111] The above are all preferred embodiments of the present invention and are not intended to limit the scope of protection of the present invention. Therefore, all equivalent changes made in accordance with the structure, shape and principle of the present invention should be covered within the scope of protection of the present invention.

Claims

1. A multi-scale SPEI prediction and drought monitoring method that integrates station and grid data, characterized in that, Includes the following steps: Step 1: Meteorological station data processing. Obtain daily-scale meteorological datasets from meteorological stations within a specified time period, and synthesize the daily-scale data into monthly-scale meteorological data. Step 2: Calculate the SPEI at 1 / 3 / 6 / 12 / 24-month scales for meteorological stations. Station value; Step 3: Filter grid points that meet the set conditions and treat each grid point as a station; with the meteorological station in the study area as the center, filter grid points located outside the set distance of the meteorological station as the densification points of the station. Step 4: Extract SPEI at 1 / 3 / 6 / 12 / 24-month scales. Grid Based on the results of step 3, extract SPEI values ​​at 1 / 3 / 6 / 12 / 24-month scales from the SPEIbase dataset that meet the conditional grid points. Grid value; Step 5, merge grid SPEI Grid With site SPEI Station Based on the results of steps 2 and 4, several stations were first randomly selected from the meteorological stations as a validation dataset; then the remaining stations were fused with the grid points to obtain the SPEI at 1 / 3 / 6 / 12 / 24-month scales. GS Merge the point data and convert it to xls format; Step 6: Anusplin interpolation generates 1km resolution SPEI at 1 / 3 / 6 / 12 / 24 month scales. GS Dataset.

2. The multi-scale SPEI prediction and drought monitoring method according to claim 1, which integrates station and grid data, is characterized in that: In step 1, the daily meteorological dataset includes variables such as precipitation, maximum temperature, minimum temperature, average temperature, average wind speed, and sunshine duration.

3. The multi-scale SPEI prediction and drought monitoring method according to claim 2, characterized in that: In step 2, based on the monthly meteorological data, the monthly potential evapotranspiration is first calculated using the Penman-Monteith formula recommended by FAO-56; then, the difference between monthly precipitation and potential evapotranspiration is calculated; next, the cumulative water deficit for 1 / 3 / 6 / 12 / 24 months is calculated; finally, the SPEI at the 1 / 3 / 6 / 12 / 24-month scale is calculated. Station value.

4. The multi-scale SPEI prediction and drought monitoring method according to claim 1, characterized in that: In step 3, the SPEIbase grid data of the study area at 0.5° is converted into point data.

5. The multi-scale SPEI prediction and drought monitoring method according to claim 1, characterized in that: In step 6, based on the results of step 5, the SPEI values ​​at the 1 / 3 / 6 / 12 / 24 month scales are respectively... GS The fusion point data was imported into SPSS, and batch files in .dat format were exported. Based on the Anusplin meteorological interpolation model, SPEI at 1 / 3 / 6 / 12 / 24-month scales was generated in batches, using the 1km resolution elevation of the study area as a covariate. GS Dataset.

6. The multi-scale SPEI prediction and drought monitoring method according to claim 1, characterized in that, It also includes step 7, SPEI GS Interpolation results verification of fusion point data: The accuracy of the interpolation results is verified by selecting three indicators: correlation coefficient (r), bias (BIAS), and root mean square error (RMSE).

7. The multi-scale SPEI prediction and drought monitoring method according to claim 1, characterized in that, It also includes step 8, based on SPEI GS Abrupt changes in meteorological drought were tested based on the annual SPEI of the specified region. GS -12 data were used to detect whether there were abrupt changes in the meteorological drought time series within a specified time period using the Mann-Kendall mutation test.

8. The multi-scale SPEI prediction and drought monitoring method according to claim 1, characterized in that, It also includes step 9, based on SPEI GS Analysis of meteorological drought variation trends and persistence, based on the annual SPEI of the specified region. GS Using data from -12, we analyzed the temporal variation characteristics of meteorological drought on an annual scale, and employed a combination of Theil-SenMedian and Mann-Kendall significance tests and the Hurst index to analyze the significance and persistence of the long-term trend of meteorological drought.

Images