Dynamic drought transmission threshold identification method and system

Abstract

The invention discloses a dynamic drought transmission threshold identification method and system, and relates to the technical field of drought transmission dynamic monitoring. Time sequence basic data is acquired, the time sequence basic data is segmented through a sliding window with a preset window length, and segmentation data of each window is acquired; calculating each meteorological drought index in a preset continuous time scale and a soil humidity index in a month scale for the segmented data; calculating a correlation coefficient between each meteorological drought index in a preset continuous time scale and a soil humidity index in a month scale based on a Pearson's correlation coefficient, and obtaining a preset time scale corresponding to the maximum correlation coefficient as drought transmission time; and calculating a drought transmission threshold value based on the drought transmission time, the soil humidity index of one month scale, combined with the joint probability distribution and the first-order Bayesian network. According to the invention, dynamic identification of different levels of drought transmission thresholds is realized.

Description

Technical Field

[0001] The present invention relates to the technical field of dynamic monitoring of drought transmission, and more specifically, to a method and system for identifying dynamic drought transmission thresholds. Background Art

[0002] The existing research on drought transmission mainly focuses on the identification of transmission time, while the research on transmission thresholds is less. And there are the following problems in the identification of existing transmission thresholds: First, most of the existing identification of transmission thresholds is based on regression models and uses fixed thresholds to determine whether drought occurs (such as SRI < -1 is drought), which cannot automatically identify the transmission thresholds of meteorological drought to different levels (mild, moderate, severe, and extreme) of agricultural drought or hydrological drought, and cannot describe the seasonal changes of transmission thresholds, resulting in insufficient early warning accuracy; Second, drought transmission thresholds are affected by climate change and human activities and are not constant over a long period. The existing research methods cannot automatically identify the long-term dynamic change process of different-level drought transmission thresholds and are difficult to meet the needs of long-term drought early warning.

[0003] Therefore, how to dynamically identify different-level drought transmission thresholds is an urgent problem to be solved by those skilled in the art. Summary of the Invention

[0004] In view of this, the present invention provides a method and system for identifying dynamic drought transmission thresholds to solve the problems existing in the above background art.

[0005] To achieve the above object, the present invention adopts the following technical solutions: A method for identifying dynamic drought transmission thresholds includes: Obtaining time-series basic data, and segmenting the time-series basic data through a sliding window with a preset window length to obtain segmented data for each window; Calculating sequences of various meteorological drought indices within a preset continuous time scale and a soil moisture index sequence on a monthly scale for the segmented data; Calculating the correlation coefficients between sequences of various meteorological drought indices within a preset continuous time scale and a soil moisture index sequence on a monthly scale based on the Pearson correlation coefficient, and obtaining the preset time scale corresponding to the maximum correlation coefficient as the drought transmission time; Calculating drought transmission thresholds based on the drought transmission time, a soil moisture index sequence on a monthly scale, combined with a joint probability distribution and a first-order Bayesian network.

[0006] Preferably, the joint probability distribution specifically includes: Using the Copula function to construct the joint probability distribution function of SPI- And SSMI-1 and calculating the least squares Euclidean distance, is the drought transmission time, SPI- is the meteorological drought index with a time scale of ; SSMI-1 is a sequence of soil moisture index on a monthly scale; the Copula function corresponding to the least squares Euclidean distance is selected as the best fitting function for SPI- and SSMI-1, and the joint distribution function is as follows: ; where and respectively represent the marginal distribution functions of SPI- and SSMI-1, C is the optimal Copula function, and are arbitrary real numbers.

[0007] Preferably, the calculation of the drought transmission threshold specifically includes: Presetting the upper limit value y of agricultural drought at different levels and the initial meteorological drought interval , the upper limit of the interval is set to -0.5, and the lower limit is set to -0.6. Iterate this interval continuously according to the following formula at an interval of -0.1. Each iteration obtains a probability value P. When P is first equal to or greater than the determination probability threshold, stop the iteration; obtain the mean value of and at this time as the drought transmission threshold from meteorological drought to agricultural drought. If the lower limit is updated to -10 and P still does not exceed the determination probability threshold, immediately stop the iteration and regard it as no transmission behavior, and the transmission threshold does not exist; ; In the formula, X is SPI- , Y is SSMI-1, is the joint probability distribution of SPI- and SSMI-1, and respectively represent the marginal distribution functions of SPI- and SSMI-1, and C is the optimal Copula function.

[0008] A dynamic drought transmission threshold identification system includes: A data segmentation module, which obtains the time-series basic data and segments the time-series basic data through a sliding window with a preset window length to obtain the segmented data of each window; An index calculation module, which calculates each meteorological drought index sequence within a preset continuous time scale and a soil moisture index sequence on a monthly scale for the segmented data; Drought transfer time acquisition module, which calculates the correlation coefficient between each meteorological drought index sequence within a preset continuous time scale and the soil moisture index sequence at a one-month scale based on the Pearson correlation coefficient, and obtains the preset time scale corresponding to the maximum correlation coefficient as the drought transfer time; Transfer threshold calculation module, which calculates the drought transfer threshold based on the drought transfer time, the soil moisture index at a one-month scale, the joint probability distribution, and the first-order Bayesian network.

[0009] Preferably, the joint probability distribution in the transfer threshold calculation module specifically includes: Using the Copula function to construct the joint probability distribution function of SPI- And SSMI-1 and calculating the least squares Euclidean distance, Is the drought transfer time, SPI- Is the time scale of Of the meteorological drought index, SSMI-1 is the soil moisture index sequence at a one-month scale; the Copula function corresponding to the least squares Euclidean distance is selected as the best fit function of SPI- And SSMI-1, and the joint distribution function is as follows: ; Among them, And Respectively represent the marginal distribution functions of SPI- And SSMI-1, C is the optimal Copula function, And Are arbitrary real numbers.

[0010] Preferably, the calculation of the drought transfer threshold in the transfer threshold calculation module specifically includes: Preset the upper limit value y of agricultural drought at different levels and the initial meteorological drought interval , the upper limit of the interval Is set to -0.5, and the lower limit Is set to -0.6, and the interval is continuously iterated according to the following formula at an interval of -0.1. Each iteration obtains a probability value P. When P is first equal to or greater than the decision probability threshold, the iteration stops; obtain the mean value of And At this time is the drought transfer threshold from meteorological drought to agricultural drought. If the lower limit Is updated to -10 and P still does not exceed the decision probability threshold, the iteration is immediately stopped, regarded as no transfer behavior occurring, and the transfer threshold does not exist; ; In the formula, X is SPI- , Y is SSMI-1, Is SPI- The joint probability distribution with SSMI-1, and respectively represent the marginal distribution functions of SPI- and SSMI-1, and C is the optimal Copula function.

[0011] According to the above technical solution, compared with the prior art, the present invention discloses a method and system for identifying dynamic drought transfer thresholds, which can automatically identify the thresholds for the transfer of meteorological drought to other types of droughts at different levels based on easily obtainable basic data; through the sliding window technique, it can describe the seasonal variations of the transfer thresholds and automatically capture the long-term variation characteristics of the drought transfer thresholds in each month over a long period, improving the comprehensiveness of the transfer threshold identification. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the drawings in the following description are only the embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on the provided drawings.

[0013] Figure 1 It is a flowchart of the method steps provided by the present invention; Figure 2 It is a diagram of the dynamic change process provided by the present invention; Figure 3 It is a technical flowchart provided by the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0014] The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of the present invention.

[0015] The embodiments of the present invention disclose a method for identifying dynamic drought transfer thresholds, as Figure 1 shown, including: Obtain time-series basic data, and divide the time-series basic data through a sliding window with a preset window length to obtain the divided data of each window; Calculate the sequences of various meteorological drought indices within a preset continuous time scale and the soil moisture index sequence on a monthly scale for the divided data; Calculate the correlation coefficient between each meteorological drought index sequence within a preset continuous time scale and the soil moisture index sequence at the one - month scale based on the Pearson correlation coefficient, and obtain the preset time scale corresponding to the maximum correlation coefficient as the drought transfer time; Calculate the drought transfer threshold based on the drought transfer time, the soil moisture index at the one - month scale, combined with the joint probability distribution and the first - order Bayesian network.

[0016] In a specific embodiment, the joint probability distribution specifically includes: Use the Copula function to construct the joint probability distribution function of SPI - and SSMI - 1 and calculate the least - squares Euclidean distance. is the drought transfer time, SPI - is the standardized precipitation index sequence (meteorological drought index) with a time scale of , and SSMI - 1 is the soil moisture index sequence (agricultural drought index) at the one - month scale; the Copula function corresponding to the least - squares Euclidean distance is selected as the best - fitting function of SPI - and SSMI - 1, and the joint distribution function is as follows: ; where, and represent the marginal distribution functions of SPI - and SSMI - 1 respectively, C is the optimal Copula function, and are arbitrary real numbers.

[0017] In a specific embodiment, calculating the drought transfer threshold specifically includes: Preset the upper limit value y of agricultural drought at different levels and the initial meteorological drought interval , the upper limit of the interval is set to - 0.5, and the lower limit is set to - 0.6. Iterate this interval continuously according to the following formula at an interval of - 0.1. Each iteration can obtain a probability value P (the probability that meteorological drought transfers to agricultural drought at different levels within this interval. The meaning of all P in the following formula is the same). When P is first equal to or greater than the decision probability threshold, stop the iteration. Obtain the mean value of and at this time as the drought transfer threshold from meteorological drought to agricultural drought. If the lower limit is updated to - 10 and P still does not exceed the decision probability threshold, immediately stop the iteration. This indicates that no transfer behavior occurs and the transfer threshold does not exist.

[0018] ; In the formula, X is SPI - , Y is SSMI-1, is SPI- the joint probability distribution of SPI- and respectively represent the marginal distribution functions of SPI- and SSMI-1, and C is the optimal Copula function.

[0019] A dynamic drought transmission threshold identification system, comprising: A data segmentation module, which obtains time-series basic data and segments the time-series basic data through a sliding window with a preset window length to obtain the segmented data of each window; An index calculation module, which calculates various meteorological drought index sequences within a preset continuous time scale and a soil moisture index sequence at a one-month scale for the segmented data; A drought transmission time acquisition module, which calculates the correlation coefficients between various meteorological drought index sequences within a preset continuous time scale and a soil moisture index sequence at a one-month scale based on the Pearson correlation coefficient, and obtains the preset time scale corresponding to the maximum correlation coefficient as the drought transmission time; A transmission threshold calculation module, which calculates the drought transmission threshold based on the drought transmission time, the soil moisture index at a one-month scale, combined with the joint probability distribution and the first-order Bayesian network.

[0020] In a specific embodiment, the joint probability distribution in the transmission threshold calculation module specifically includes: Using the Copula function to construct the joint probability distribution function of SPI- and SSMI-1 and calculating the least squares Euclidean distance, is the drought transmission time, SPI- is the meteorological drought index with a time scale of , and SSMI-1 is the soil moisture index sequence at a one-month scale; the Copula function corresponding to the least squares Euclidean distance is selected as the best fitting function of SPI- and SSMI-1, and the joint distribution function is as follows: ; wherein, and respectively represent the marginal distribution functions of SPI- and SSMI-1, C is the optimal Copula function, and are arbitrary real numbers.

[0021] In a specific embodiment, calculating the drought transmission threshold in the transmission threshold calculation module specifically includes: The upper limit values y of the preset agricultural drought at different levels and the initial meteorological drought interval , the upper limit of the interval is set to -0.5, and the lower limit is set to -0.6. Iterate this interval continuously according to the following formula at an interval of -0.1. Each iteration obtains a probability value P. When P is first equal to or greater than the determination probability threshold, stop the iteration; obtain the mean value of and at this time as the drought transmission threshold from meteorological drought to agricultural drought. If the lower limit is updated to -10 and P still does not exceed the determination probability threshold, immediately stop the iteration, which is regarded as no transmission behavior and the transmission threshold does not exist; ; In the formula, X is SPI- , Y is SSMI-1, is the joint probability distribution of SPI- and SSMI-1, and respectively represent the marginal distribution functions of SPI- and SSMI-1, and C is the optimal Copula function.

[0022] In a specific embodiment, the present invention includes the following steps: (1) Window selection and data segmentation The long-term dynamic monitoring of the transmission threshold is calculated through a sliding window, and the selection of the window length can be adjusted according to actual needs. First, the long-term basic data (precipitation, soil moisture) obtained is segmented using a sliding window with a set window length, and the sliding step length is determined to be 1 year. Assuming that 60-year monthly data can be obtained and the window length is set to n years, then the original data can be segmented into (60 - n + 1) different windows. Then, the calculation in part (2) is performed in each window.

[0023] (2) Calculation of standardized drought index Calculate SPI (SPI-1 to SPI-12) and SSMI-1 on 12 consecutive time scales respectively. Meteorological drought is represented by the standardized meteorological drought index (SPI), and agricultural drought is represented by the standardized soil moisture index (SSMI).

[0024] (3) Determination of transmission time Use the Pearson correlation coefficient to calculate the correlation coefficients between SPI-1 to SPI-12 in each window and the corresponding month's SSMI-1, and find the SPI time scale corresponding to the maximum correlation coefficient value for each month . is the monthly drought transmission time, indicating that the agricultural drought in the current month is transmitted from the meteorological drought in the previous months. Then, using the transmission times of each month in different windows, calculate according to part (4) to determine the corresponding transmission threshold.

[0025] (4) Calculation of transmission threshold ① Construction of the joint distribution of SPI- and SSMI-1.

[0026] First, take the 5 commonly used two-dimensional Copula functions in hydro-meteorology (Frank, Clayton, Gumbel, Gaussian, and Student's t) as candidates. Use each Copula function to construct the joint probability distribution function of SPI- and SSMI-1 and calculate the least squares Euclidean distance (SED). The Copula function corresponding to the minimum SED is selected as the best fitting function of SPI- and SSMI-1. The joint distribution function is as follows: where, and represent the marginal distribution functions of SPI- and SSMI-1 respectively, C is the optimal Copula function, and are arbitrary real numbers.

[0027] ② Calculation of transmission threshold The transmission threshold is solved using the joint probability distribution and the first-order Bayesian network. First, preset the upper limit value y of agricultural drought at different levels and the initial meteorological drought interval , the upper limit of the interval is set to -0.5, the lower limit is set to -0.6, and iterate this interval continuously according to the following formula at an interval of -0.1. Each iteration obtains a probability value P. When P is first equal to or greater than the decision probability threshold, stop the iteration; obtain the mean value of and at this time as the drought transmission threshold from meteorological drought to agricultural drought. If the lower limit is updated to -10 and P still does not exceed the decision probability threshold, immediately stop the iteration, regarded as no transmission behavior occurring, and the transmission threshold does not exist; ; In the formula, y is the upper limit value at different levels, and its values are -0.5, -1, -1.5, and -2 respectively, representing slight, moderate, severe, and extreme; X is SPI- , Y is SSMI-1, is SPI- the joint probability distribution with SSMI-1, and respectively represent the marginal distribution functions of SPI- and SSMI-1, and C is the optimal Copula function.

[0028] It should be noted that the upper limit of the initial meteorological drought interval, -0.5, cannot be changed because SPI < -0.5 is generally considered the critical value for the occurrence of meteorological drought. The iteration interval of -0.1 and the lower limit of the interval can be changed according to the needs of the user, but too large an interval range will lead to inaccurate identification of the transfer threshold, and too small an interval will result in a large memory occupation for the operation. The probability discrimination criterion of 0.8 can also be changed according to the needs, and a multi-probability discrimination criterion can be set according to the actual situation, so as to obtain the dynamic transfer threshold under multiple scenarios.

[0029] The core innovation points of the present invention are as follows: Hierarchical threshold and Copula optimization: Joint distributions of SPI and SSMI / SRI are constructed for four scenarios of mild, moderate, severe, and extreme drought respectively to improve the transfer thresholds at different levels; the optimal function is selected from 5 Copula functions (Gaussian, Gumbel, Frank, Clayton, Student’s t) for construction based on the minimum Euclidean square distance (SED) criterion to avoid certain errors caused by using a single Copula function.

[0030] Bayesian network iterative optimization: Calculate the conditional probability through the first-order Bayesian network and dynamically adjust the meteorological drought interval ; when P≥0.8, take the interval mean as the transfer threshold, and support the customization of the probability threshold P (such as 0.6~0.9).

[0031] Dynamic sliding window mechanism: Use a sliding window with a length of 20 - 30 years to segment the data, which can balance data stability and sensitivity at the same time, and then use a step size of 1 year to achieve annual update, which can dynamically capture the dynamic changes of the transfer threshold under the influence of climate change and human activities.

[0032] Use of multi-source data: The calculation framework can input soil moisture and runoff data respectively according to the needs, and obtain the transfer thresholds of meteorological - agricultural and meteorological - hydrological drought respectively, not limited to the identification of one type of transfer threshold. At the same time, more abundant transfer thresholds of different types of drought can also be obtained by calculating other types of drought indices.

[0033] Monthly precipitation and soil moisture data are used to calculate the transfer thresholds of meteorological-agricultural drought, and monthly precipitation and runoff (or runoff depth) data are used to calculate the transfer thresholds of meteorological-hydrological drought.

[0034] Suppose the monthly precipitation and soil moisture data of a certain area from 1965 to 2024 (a total of 60 years) are obtained. Using the method proposed by the present invention, the dynamic change process of the transfer thresholds from meteorological drought to different levels of agricultural drought in each month during different periods can be obtained. Figure 2 It shows the dynamic change process of the transfer thresholds of meteorological-agricultural drought at different levels calculated by this method. In use, specific months or all months can be selected for plotting according to one's own needs. Taking May as an example, the dynamic change process of the transfer thresholds of different levels of drought can be clearly observed, and accurate transfer thresholds for each period can be provided. It solves the problems that the existing methods cannot automatically identify the transfer thresholds of different levels of drought and cannot dynamically identify the changes in transfer thresholds.

[0035] The implementation steps are as Figure 3 shown, which are: data input Sliding window segmentation Calculation of multi-scale standardized drought index Construction of Copula joint distribution Iteration of Bayesian network Dynamic threshold output. In actual use, the final result can be obtained by inputting data.

[0036] In this specification, each embodiment is described in a progressive manner. The key point of each embodiment is to illustrate the differences from other embodiments. The same or similar parts among the embodiments can be referred to each other. For the devices disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the description is relatively simple, and the relevant parts can be referred to the description of the method part.

[0037] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present invention. Various modifications to these embodiments will be obvious to those skilled in the art. The general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to these embodiments shown herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A dynamic drought transmission threshold identification method, characterized in that: include: Acquire time series basic data, segment the time series basic data through a sliding window with a preset window length, and obtain segmented data of each window; Calculating each meteorological drought index sequence within a preset continuous time scale and a soil moisture index sequence on a monthly scale for the segmented data; Based on the Pearson correlation coefficient, the correlation coefficients of each meteorological drought index series within a preset continuous time scale and the soil moisture index series on a monthly scale are calculated, and the preset time scale corresponding to the maximum correlation coefficient is obtained as the drought transmission time; The drought transmission threshold is calculated based on the drought transmission time, a monthly soil moisture index combined with joint probability distribution and a first-order Bayesian network.

2. A dynamic drought transmission threshold identification method according to claim 1, characterized in that: The joint probability distribution specifically includes: Using Copula functions to build SPI The joint probability distribution function with SSMI-1 and the least squares Euclidean distance are calculated. Passing time for drought, SPI- The time scale is The meteorological drought index is SSMI-1, which is a monthly soil moisture index sequence. The Copula function corresponding to the least squares Euclidean distance is selected as SPI- The best fitting function with SSMI-1, the joint distribution function is as follows: ； in, and Respectively represent SPI- and the marginal distribution function of SSMI-1, C is the optimal Copula function, and is any real number.

3. A dynamic drought transmission threshold identification method according to claim 2, characterized in that: The calculation of the drought transmission threshold specifically includes: Preset upper limit y of agricultural drought at different levels and initial meteorological drought interval , the upper limit of the interval Set to -0.5, the lower limit Set it to -0.6, and iterate the interval at intervals of -0.1 according to the following formula. Each iteration obtains a probability value P. When P is equal to or greater than the judgment probability threshold for the first time, the iteration stops. and The mean of is the drought transfer threshold from meteorological drought to agricultural drought. If the lower limit When P is updated to -10, it still does not exceed the judgment probability threshold and the iteration stops immediately. It is considered that no transfer behavior occurs and the transfer threshold does not exist. ； Where X is SPI- , Y is SSMI-1, For SPI- The joint probability distribution with SSMI-1, and Respectively represent SPI- and the marginal distribution function of SSMI-1, and C is the optimal Copula function.

4. A dynamic drought transfer threshold identification system, using a dynamic drought transfer threshold identification method according to any one of claims 1 to 3, characterized in that: include: A data segmentation module is used to obtain basic time series data, segment the basic time series data using a sliding window of a preset window length, and obtain segmented data for each window; An index calculation module calculates each meteorological drought index sequence within a preset continuous time scale and a soil moisture index sequence on a monthly scale for the segmented data; The drought transfer time acquisition module calculates the correlation coefficient between each meteorological drought index sequence in a preset continuous time scale and the soil moisture index sequence on a monthly scale based on the Pearson correlation coefficient, and obtains the preset time scale corresponding to the maximum correlation coefficient as the drought transfer time; The transmission threshold calculation module calculates the drought transmission threshold based on the drought transmission time, a soil moisture index on a monthly scale, a joint probability distribution and a first-order Bayesian network.

5. A dynamic drought transmission threshold identification system according to claim 4, characterized in that: The joint probability distribution in the transfer threshold calculation module specifically includes: Using Copula functions to build SPI The joint probability distribution function with SSMI-1 and the least squares Euclidean distance are calculated. Passing time for drought, SPI- The time scale is The meteorological drought index is SSMI-1, which is a monthly soil moisture index sequence. The Copula function corresponding to the least squares Euclidean distance is selected as SPI- The best fitting function with SSMI-1, the joint distribution function is as follows: ； in, and Respectively represent SPI- and the marginal distribution function of SSMI-1, C is the optimal Copula function, and is any real number.

6. A dynamic drought transmission threshold identification system according to claim 5, characterized in that: Calculating the drought transfer threshold in the transfer threshold calculation module specifically includes: Preset upper limit value y of agricultural drought at different levels and initial meteorological drought interval , the upper limit of the interval Set to -0.5, the lower limit Set it to -0.6, and iterate the interval at intervals of -0.1 according to the following formula. Each iteration obtains a probability value P. When P is equal to or greater than the judgment probability threshold for the first time, the iteration stops. and The mean of is the drought transfer threshold from meteorological drought to agricultural drought. If the lower limit When P is updated to -10, it still does not exceed the judgment probability threshold and the iteration stops immediately. It is considered that no transfer behavior occurs and the transfer threshold does not exist. ； Where X is SPI- , Y is SSMI-1, For SPI- The joint probability distribution with SSMI-1, and Respectively represent SPI- and the marginal distribution function of SSMI-1, and C is the optimal Copula function.

Images