Method and system for identifying key human factors influencing hydrological drought

Abstract

The invention provides a method and system for identifying key human factors affecting hydrological drought, and belongs to the field of water resource management.The method comprises the steps that natural runoff data of a research area is obtained, and a hydrological drought judgment threshold value is calculated; according to the hydrological drought determination threshold value, obtaining a natural hydrological drought occurrence frequency in the research time period and an actually measured hydrological drought occurrence frequency in the research time period; calculating the difference value between the actually measured hydrological drought occurrence frequency in the research time period and the natural hydrological drought occurrence frequency in the research time period, and dividing the influence degree of human factors on hydrological drought in the research area; establishing a classification model of influence degrees of human factors on hydrological drought on different time scales through a random forest model, and calculating relative contribution rates of different human factors on different time scales; and key human factors influencing hydrological drought are identified. Fine division and quantitative attribution of the human factors are realized, and the accuracy of identifying the human factors influencing hydrological drought is improved.

Description

Technical Field

[0001] The present invention relates to the field of water resources management, and in particular to a method and system for identifying key human factors that affect hydrological drought. Background Art

[0002] In the context of the Anthropocene geological era, the coupling effect between the artificial water cycle system and the natural water cycle is becoming increasingly significant. Large-scale water resource management projects reconstruct terrestrial hydrological processes through three mechanisms: water withdrawal directly reduces regional water reserves, cross-basin water transfer realizes the spatial redistribution of water resources, and reservoir group scheduling forms water volume regulation in the temporal dimension. This three-dimensional intervention system of space, time and storage not only changes the natural runoff situation, but also significantly alleviates the negative impact of drought climate by optimizing the spatiotemporal allocation of water resources. It is worth noting that the systematic transformation of water-energy flux by human activities has exceeded the traditional water cycle framework. The combined factors of land use changes such as afforestation or deforestation, rapid urbanization, atmospheric pollutant emissions and the layout of high-water-consuming industries have reconstructed the energy-water exchange paradigm between land and atmosphere, ultimately leading to a systematic shift in the characteristic values ​​of hydrological drought and affecting runoff generation.

[0003] Currently, there are two main traditional methods for assessing the impact of human activities on hydrological drought. One is a comparative analysis method based on historical runoff periods. By dividing the period into a natural baseline and a period of human interference, non-parametric tests and mutation detection are used to analyze the statistical significance of characteristic parameters such as drought intensity, duration, and frequency. When observed runoff data are limited, an undisturbed reference basin is introduced to establish a comparison baseline for natural hydrological processes. The other is a model-driven approach that focuses on the parametric representation of human activities. By constructing natural-anthropogenic dual scenarios in distributed hydrological models or watershed water cycle models, scenario comparison methods are used to decouple the drought impacts of specific human activities. However, existing technologies have significant limitations, mainly manifested in the difficulty of single-factor analysis paradigms in analyzing the synergistic or antagonistic effects of multiple factors, the attribution credibility is weakened due to model structural uncertainty, and the coupling between human behavioral heterogeneity and the nonlinear response of the hydrological system has not been effectively characterized. Therefore, existing technologies are unable to quantify the combined or single impacts of complex human activities on hydrological drought, resulting in inaccurate identification of the human factors affecting hydrological drought. Summary of the Invention

[0004] In order to solve the above problems, the present invention proposes a method and system for identifying key human factors affecting hydrological drought, which can achieve fine division and quantitative attribution of human factors and improve the accuracy of identifying human factors affecting hydrological drought.

[0005] To achieve the above-mentioned purpose, an embodiment of the present invention provides a method for identifying key human factors affecting hydrological drought, comprising: obtaining natural runoff data in a study area, calculating standardized runoff indexes at different time scales, and using the runoff value corresponding to the standardized runoff index being equal to a preset threshold as a hydrological drought determination threshold; wherein the different time scales include: month, season, and year; based on a preset study period, comparing the hydrological drought determination threshold with the natural runoff data of the study area at different time scales to obtain the frequency of natural hydrological drought within the study period; obtaining measured runoff data in the study area, comparing the hydrological drought determination threshold with the measured runoff data of the study area at different time scales to obtain the research drought frequency. The measured frequency of hydrological drought during the study period was calculated; the difference between the measured frequency of hydrological drought during the study period and the frequency of natural hydrological drought during the study period was calculated, and the degree of influence of human factors on hydrological drought in the study area was divided into aggravating effect, mitigating effect and neutral effect; based on the degree of influence of human factors on hydrological drought in the study area and the pre-acquired basic data of the study area, a classification model of the degree of influence of human factors on hydrological drought at different time scales was established through the random forest model, and the relative contribution rate of different human factors at different time scales was calculated; based on the relative contribution rate of different human factors at different time scales, the key human factors affecting hydrological drought were identified.

[0006] An embodiment of the present invention proposes a method for identifying key human factors affecting hydrological drought. The method uses dynamic comparative analysis to couple historical natural runoff with measured runoff, constructs a hydrological drought determination threshold assessment system, calculates the measured frequency of hydrological drought and the frequency of natural hydrological drought within a study period, and divides the degree of influence of human factors on hydrological drought in the study area into aggravating effect, mitigating effect, and neutral effect. A classification model is constructed based on the degree of influence of human factors on hydrological drought, and the classification model quantifies the contribution rate of the characteristics of the human factors to be identified. The key human factors affecting hydrological drought are then identified based on the contribution rate of the characteristics of the human factors to be identified. Through multi-source data fusion and dynamic coupling modeling, the synergistic impact of complex human activities on hydrological drought is effectively quantified, the key human factors affecting hydrological drought are quantitatively attributed, and the accuracy of identifying human factors affecting hydrological drought is improved.

[0007] Furthermore, natural runoff data of the study area are obtained, standardized runoff indexes of different time scales are calculated, and the runoff value corresponding to the standardized runoff index being equal to a preset threshold is used as the threshold for determining hydrological drought; wherein, different time scales include: month, season and year, including: calculating the cumulative natural runoff series of different time scales based on the natural runoff data of the study area; fitting the cumulative natural runoff series of different time scales through several initial frequency distribution functions to obtain the fitted distribution frequency graph corresponding to each initial frequency distribution function; based on the fitted distribution frequency graph corresponding to each initial frequency distribution function, screening the frequency distribution function that meets the preset test requirements to obtain the target frequency distribution function; calculating the standardized runoff index series based on the target frequency distribution function and the cumulative natural runoff series of different time scales, and using the runoff value corresponding to the index in the standardized runoff index series being equal to the preset threshold as the threshold for determining hydrological drought, wherein, different time scales include: month, season and year.

[0008] In the above scheme, historical natural runoff data are analyzed, multi-time scale cumulative runoff series are constructed, and multiple frequency distribution function goodness-of-fit tests are used to screen the optimal target frequency distribution function, and the hydrological drought judgment thresholds under different time dimensions are accurately quantified. Thus, the reliability of the hydrological drought judgment thresholds is enhanced by strictly screening the frequency distribution functions, providing a reliable data standard for subsequent multi-time scale dynamic analysis to capture the differentiated interference patterns of human activities on hydrological droughts, thereby improving the attribution accuracy of human factors on hydrological droughts, and further improving the accuracy of identifying human factors affecting hydrological droughts.

[0009] Furthermore, a standardized runoff index sequence is calculated based on the target frequency distribution function and the cumulative natural runoff sequence at different time scales, and the runoff value corresponding to the index in the standardized runoff index sequence being equal to a preset threshold is used as the hydrological drought judgment threshold, including: fitting the cumulative natural runoff sequence at different time scales based on the target frequency distribution function to obtain a target fitting result; converting the target fitting result into a standardized runoff index sequence based on a preset standardization processing algorithm; and selecting the runoff value corresponding to the index equal to -1 in the standardized runoff index sequence as the hydrological drought judgment threshold.

[0010] In the above scheme, the target frequency distribution function is used to fit multi-time scale runoff data to generate a standardized runoff index sequence. The runoff value corresponding to the index equal to -1 in the standardized runoff index sequence is selected as the threshold for hydrological drought judgment, and a quantitative benchmark for the natural hydrological state is established. The statistical distribution characteristics of the standardized runoff index sequence are used to analyze the degree of deviation between the measured runoff and the natural benchmark, and the natural fluctuations and human interference signals are effectively separated through multi-time scale comparison. This method enhances the characterization ability of the hydrological drought judgment threshold through standardized processing at a dynamic scale, and combines distribution differences to quantitatively identify the disturbance intensity of human activities on the runoff process, thereby improving the objectivity and accuracy of the human attribution of drought, and further improving the accuracy of identifying human factors affecting hydrological drought.

[0011] Furthermore, based on the preset research period, the hydrological drought judgment threshold and the natural runoff data of the study area at different time scales are compared to obtain the frequency of natural hydrological drought within the study period, including: obtaining the number of time periods in which the natural runoff data in the study area is less than the hydrological drought judgment threshold based on the preset research period and the cumulative natural runoff series at different time scales; obtaining the frequency of natural hydrological drought within the study period based on the total number of time periods in the preset research period and the number of time periods in which the natural runoff data in the study area is less than the hydrological drought judgment threshold.

[0012] In the above scheme, the number of time periods in which the natural runoff data in the study area was lower than the hydrological drought judgment threshold was counted, and its ratio to the total number of time periods was calculated to accurately obtain the frequency of natural hydrological drought in the study period. This frequency value was used as a quantitative benchmark and compared with the measured drought frequency to effectively separate natural fluctuations and human disturbance signals, providing an objective reference for the subsequent identification of the degree of influence of human factors, quantitatively calculating the drought frequency at multiple time scales, significantly improving the measurability of the intensity of human interference, and thereby improving the accuracy of identifying human factors affecting hydrological drought.

[0013] Furthermore, the measured runoff data of the study area are obtained, and the hydrological drought judgment threshold and the measured runoff data of the study area at different time scales are compared to obtain the measured frequency of hydrological drought in the study period, including: calculating the cumulative measured runoff volume series at different time scales based on the measured runoff data of the study area at different time scales; obtaining the number of time periods in which the measured runoff data of the study area is less than the hydrological drought judgment threshold based on the preset study period and the cumulative measured runoff volume series at different time scales; obtaining the measured frequency of hydrological drought in the study period based on the total number of time periods in the preset study period and the number of time periods in which the measured runoff data of the study area is less than the hydrological drought judgment threshold.

[0014] In the above scheme, the number of time periods in which the measured runoff data in the study area is less than the hydrological drought judgment threshold is counted, and its ratio to the total number of time periods is calculated to accurately quantify the actual probability of drought under the interference of human activities. The measured frequency is compared with the natural background frequency, and the difference between the two is used to objectively separate the degree of influence of human factors on hydrological drought, providing a quantitative basis for the analysis of the synergistic effect of multiple factors, effectively capturing the dynamic interference of human activities on drought characteristics, and thus improving the accuracy of identifying human factors affecting hydrological drought.

[0015] Furthermore, the difference between the measured frequency of hydrological drought during the study period and the frequency of natural hydrological drought during the study period is calculated, and the degree of influence of human factors on hydrological drought in the study area is divided into aggravating effect, mitigating effect and neutral effect, including: calculating the difference between the measured frequency of hydrological drought during the study period and the frequency of natural hydrological drought during the study period to obtain the degree of influence of human factors on hydrological drought in the study area; if the degree of influence of human factors on hydrological drought in the study area is greater than the preset influence threshold, the degree of influence of human factors on hydrological drought in the study area is divided into aggravating effect; if the degree of influence of human factors on hydrological drought in the study area is less than the preset influence threshold, the degree of influence of human factors on hydrological drought in the study area is divided into mitigating effect; if the degree of influence of human factors on hydrological drought in the study area is equal to the preset influence threshold, the degree of influence of human factors on hydrological drought in the study area is divided into neutral effect.

[0016] In the above scheme, the difference between the frequency of natural hydrological drought at different time scales and the frequency of measured hydrological drought at different time scales is calculated, and the type of the degree of influence of human factors on hydrological drought is divided according to the difference threshold. The three types of aggravation effect, mitigation effect and neutral effect are proposed to break through the limitations of traditional single factor analysis, deconstruct the differentiated impact of human factors on hydrological drought, refine the classification of human factors affecting hydrological drought, and improve the accuracy of identifying human factors affecting hydrological drought.

[0017] Furthermore, based on the degree of influence of human factors on hydrological drought in the study area and the pre-acquired basic data of the study area, a classification model of the degree of influence of human factors on hydrological drought at different time scales was established through a random forest model, and the relative contribution rate of different human factors at different time scales was calculated, including: obtaining the basic data of human factors in the study area and calculating the average annual change rate of each human factor; using the average annual change rate of each human factor as a prediction factor and the degree of influence of human factors on hydrological drought at different time scales as the classification result, the prediction factor was input into a preset decision tree ensemble learning model for classification learning to obtain an initial classification model; based on the initial classification model and the prediction factor, the predicted degree of influence of human factors on hydrological drought was obtained; based on the hydrological drought judgment threshold and the prediction factor, the measured degree of influence of human factors on hydrological drought was obtained; based on the predicted degree of influence of human factors on hydrological drought and the measured degree of influence of human factors on hydrological drought, the model training accuracy was obtained; the initial classification model was optimized until the model training accuracy met the preset training requirements to obtain a target classification model; based on the target classification model, the relative contribution rate of different human factors at different time scales was calculated.

[0018] In the above scheme, the basic data of human factors in the study area are combined to quantify the average annual change rate of each human factor as a prediction factor, which is input into the decision tree ensemble learning model. The difference in frequency between measured and natural hydrological droughts is mapped to the degree of influence of the corresponding human factors on hydrological droughts. The model parameters are iteratively optimized by calculating the model training accuracy so that the matching degree between the predicted classification results and the actually observed drought types reaches the preset accuracy threshold. The nonlinear classification ability of machine learning is used to analyze the synergistic effect of multiple factors, and the dynamic annual change rate characteristics are used to capture the cumulative effects of human activities, so as to achieve accurate classification of heterogeneous human interference, thereby improving the accuracy of identifying human factors affecting hydrological droughts.

[0019] Furthermore, based on the target classification model, the relative contribution rates of different human factors at different time scales are calculated, including: obtaining the characteristic data of human factors in the area to be studied based on the average annual change rate of each human factor; inputting the characteristic data of human factors in the area to be studied into the target classification model to obtain the predicted contribution evaluation results of each decision tree to the characteristic data of human factors in the area to be studied; and performing weighted averaging on the predicted contribution evaluation results to obtain the characteristic contribution rate of the human factors to be identified.

[0020] In the above scheme, the target classification model integrates the tree structure characteristics of the decision tree model, performs a multi-dimensional split node importance evaluation on the annual change rate characteristics of the input human factors, and then performs a weighted average on the contribution results output by all decision trees. Finally, the global characteristic contribution rate of each factor to the drought type classification is output. By utilizing the decision tree group's ability to analyze nonlinear relationships, the complex mechanism of the synergistic impact of multiple factors on drought is effectively captured, the contribution quantification error of the human factor characteristics to be identified is reduced, the attribution accuracy under the composite impact scenario is improved, and the accuracy of identifying human factors affecting hydrological drought is thereby improved.

[0021] Furthermore, based on the contribution rate of the characteristics of the human factors to be identified, the key human factors affecting hydrological drought are identified, including: calculating the contribution rates of different human factor characteristics at different time scales to obtain the contribution rates of several human factor characteristics to be identified; sorting the contribution rates of several human factor characteristics to be identified, and screening the characteristics of the human factors to be identified that meet the preset contribution rate requirements as the key human factors affecting hydrological drought.

[0022] In the above scheme, the contribution rates of human factor characteristics at multiple time scales are calculated and ranked, a contribution gradient screening mechanism is established, key human factors are screened according to preset contribution rate requirements, and the spatiotemporal distribution characteristics of contribution rates are used to analyze the persistence and cumulative effects of human activities, accurately distinguish the key human factors affecting hydrological drought, and improve the accuracy of identifying human factors affecting hydrological drought.

[0023] An embodiment of the present invention also provides a key human factor identification system affecting hydrological drought, comprising: a hydrological drought determination threshold calculation module, a natural hydrological drought occurrence frequency calculation module, a measured hydrological drought occurrence frequency calculation module, a human factor influence degree classification module, a contribution rate calculation module and a key human factor identification module; the hydrological drought determination threshold calculation module is used to obtain natural runoff data in a study area, calculate standardized runoff indexes at different time scales, and use the runoff value corresponding to the standardized runoff index being equal to a preset threshold as the hydrological drought determination threshold; wherein the different time scales include: month, season and year; the natural hydrological drought occurrence frequency calculation module is used to compare the hydrological drought determination threshold with the natural runoff data of the study area at different time scales based on a preset study period to obtain the natural hydrological drought occurrence frequency within the study period; the measured hydrological drought occurrence frequency calculation module is used to obtain the natural runoff data of the study area The measured runoff data are compared with the hydrological drought judgment threshold and the measured runoff data of the study area at different time scales to obtain the measured frequency of hydrological drought in the study period; the human factor impact degree classification module is used to calculate the difference between the measured frequency of hydrological drought in the study period and the frequency of natural hydrological drought in the study period, and divide the degree of influence of human factors on hydrological drought in the study area into aggravating effect, mitigation effect and neutral effect; the contribution rate calculation module is used to establish a classification model of the degree of influence of human factors on hydrological drought at different time scales through the random forest model based on the degree of influence of human factors on hydrological drought in the study area and the pre-acquired basic data of the study area, and calculate the relative contribution rate of different human factors at different time scales; the key human factor identification module is used to identify the key human factors affecting hydrological drought based on the relative contribution rate of different human factors at different time scales.

[0024] An embodiment of the present invention provides a system for identifying key human factors affecting hydrological drought. The hydrological drought threshold calculation module combines a natural hydrological drought occurrence frequency calculation module and a measured hydrological drought occurrence frequency calculation module, using dynamic comparative analysis to couple historical natural runoff with measured runoff. This system constructs a hydrological drought threshold assessment system to calculate the measured hydrological drought occurrence frequency and the natural hydrological drought occurrence frequency within a study period. The human factor impact degree classification module classifies the degree of human factor influence on hydrological drought within the study area into exacerbation effect, mitigation effect, and neutral effect. The contribution rate calculation module constructs a classification model based on the degree of human factor influence on hydrological drought, and uses the classification model to quantify the contribution rate of the human factor characteristics to be identified. The key human factor identification module then identifies the key human factors affecting hydrological drought based on the contribution rate of the human factor characteristics to be identified. Through multi-source data fusion and dynamic coupling modeling, the synergistic impact of complex human activities on hydrological drought is effectively quantified, the key human factors affecting hydrological drought are quantitatively attributed, and the accuracy of identifying human factors affecting hydrological drought is improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Figure 1 A schematic flow chart of the steps of a method for identifying key human factors affecting hydrological drought provided by one embodiment of the present invention;

[0026] Figure 2 A schematic diagram of the fitting distribution frequency of monthly time-scale flow data of a watershed in a region coded as 1010001, according to a method for identifying key human factors affecting hydrological drought provided by one embodiment of the present invention;

[0027] Figure 3 A schematic diagram of monthly time-scale flow data thresholds for a watershed in a region coded as 1010001, according to a method for identifying key human factors affecting hydrological drought provided by an embodiment of the present invention;

[0028] Figure 4 A schematic diagram of the change in the frequency of hydrological drought caused by human factors on a monthly time scale in a basin coded as 1010001 in a certain region, according to a method for identifying key human factors affecting hydrological drought provided by an embodiment of the present invention;

[0029] Figure 5 A partially enlarged schematic diagram of the change in the frequency of hydrological drought caused by human factors on a monthly time scale in a basin coded as 1010001 in a certain region, according to a method for identifying key human factors affecting hydrological drought provided by one embodiment of the present invention;

[0030] Figure 6 A schematic diagram showing the relative contribution rates of different human factors affecting hydrological drought on a monthly time scale over a certain land area according to a method for identifying key human factors affecting hydrological drought provided by one embodiment of the present invention;

[0031] Figure 7 A schematic diagram of the module structure of a system for identifying key human factors affecting hydrological drought provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0032] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0033] Example 1

[0034] See also Figure 1 , Figure 1 This is a method provided by an embodiment of the present invention. Figure 1As shown, the embodiment of the present invention provides a method for identifying key human factors affecting hydrological drought, including steps 101 to 105, each of which is specifically as follows:

[0035] Step 101: Obtain natural runoff data for the study area, calculate standardized runoff indexes at different time scales, and use the runoff value corresponding to the standardized runoff index being equal to a preset threshold as a hydrological drought determination threshold; wherein the different time scales include: month, season, and year;

[0036] Step 102: Based on a preset study period, the hydrological drought determination threshold is compared with the natural runoff data of the study area at different time scales to obtain the frequency of natural hydrological drought in the study period;

[0037] Step 103: Obtain measured runoff data for the study area, compare the hydrological drought determination threshold with the measured runoff data for the study area at different time scales, and obtain the measured frequency of hydrological drought within the study period;

[0038] Step 104, calculating the difference between the measured hydrological drought occurrence frequency during the study period and the natural hydrological drought occurrence frequency during the study period, and classifying the degree of influence of human factors on hydrological drought in the study area into aggravating effect, mitigating effect, and neutral effect;

[0039] Step 105: Based on the degree of impact of human factors on hydrological drought in the study area and pre-acquired basic data of the study area, a classification model of the degree of impact of human factors on hydrological drought at different time scales is established using a random forest model, and the relative contribution rate of different human factors at different time scales is calculated;

[0040] Step 106 , based on the relative contribution rates of different human factors at different time scales, identify the key human factors that affect hydrological drought.

[0041] As a preferred embodiment, in the process of analyzing the runoff of the study area, it usually refers to analyzing the runoff of each basin in the study area. In this embodiment, 2099 basins on land in a certain area are taken, and the study period is set from October 2002 to January 2004. The monthly time scale is used as an example for specific explanation, which will not be repeated below; historical natural runoff data of the study area are collected, and standardized runoff data series of time scales such as month, season and year are calculated. The value of standardized runoff data equal to -1 is selected from the standardized runoff data series of different scales as the hydrological drought judgment threshold. The hydrological drought judgment threshold is used to quantitatively evaluate the hydrological drought situation of each basin in the study area. Then, according to the set study period, the historical measured runoff data of the study area is obtained within the study period. Combined with the historical natural runoff data of the study area and the hydrological drought judgment threshold, the historical natural hydrological drought events and the historical measured hydrological drought events within the study period are obtained respectively, and the frequency of natural hydrological drought and the frequency of measured hydrological drought are correspondingly counted. By calculating the difference between the frequency of natural hydrological drought and the frequency of measured hydrological drought, the impact of human factors on hydrological drought is classified into three types: "aggravation effect," "mitigation effect," and "neutral effect." "Aggravation effect" refers to human activities causing hydrological drought to become more severe or frequent, "mitigation effect" refers to human activities helping to alleviate the impact of hydrological drought, and "neutral effect" refers to human activities having no significant impact on hydrological drought. Based on the type of human factor impact on hydrological drought and the basic human factor data of the study area, an ensemble learning model is combined to construct a target classification model. In this embodiment, the ensemble learning model can use a random forest model. The target classification model is then used to calculate the relative contribution rate of each feature of the basic human factor data of the study area at different time scales. The basic human factor data of the study area include population density, water withdrawal, reservoir density, and land use type. Finally, based on the relative contribution rate of each feature at different time scales, the key human factors affecting hydrological drought are identified at multiple time scales.

[0042] An embodiment of the present invention proposes a method for identifying key human factors affecting hydrological drought. The method uses dynamic comparative analysis to couple historical natural runoff with measured runoff, constructs a hydrological drought determination threshold assessment system, calculates the measured frequency of hydrological drought and the frequency of natural hydrological drought within a study period, and divides the degree of influence of human factors on hydrological drought in the study area into aggravating effect, mitigating effect, and neutral effect. A classification model is constructed based on the degree of influence of human factors on hydrological drought, and the classification model quantifies the contribution rate of the characteristics of the human factors to be identified. The key human factors affecting hydrological drought are then identified based on the contribution rate of the characteristics of the human factors to be identified. Through multi-source data fusion and dynamic coupling modeling, the synergistic impact of complex human activities on hydrological drought is effectively quantified, the key human factors affecting hydrological drought are quantitatively attributed, and the accuracy of identifying human factors affecting hydrological drought is improved.

[0043] As a preferred solution, natural runoff data of the study area is obtained, standardized runoff indexes of different time scales are calculated, and the runoff value corresponding to the standardized runoff index being equal to a preset threshold is used as the hydrological drought judgment threshold; wherein, different time scales include: month, season and year, including: calculating the cumulative natural runoff series of different time scales based on the natural runoff data of the study area; fitting the cumulative natural runoff series of different time scales through several initial frequency distribution functions to obtain the fitted distribution frequency graph corresponding to each initial frequency distribution function; based on the fitted distribution frequency graph corresponding to each initial frequency distribution function, screening the frequency distribution function that meets the preset test requirements to obtain the target frequency distribution function; calculating the standardized runoff index series based on the target frequency distribution function and the cumulative natural runoff series of different time scales, and using the runoff value corresponding to the index in the standardized runoff index series being equal to the preset threshold as the hydrological drought judgment threshold, wherein, different time scales include: month, season and year.

[0044] As a preferred implementation method, historical natural runoff data of the area to be studied is collected from the historical database of each watershed in the area to be studied. In this embodiment, the natural runoff data from January 1961 to December 2010 provided by the historical database of a regional forestry bureau are used for interpretation. The cumulative natural runoff sequences for 1, 3, and 12 consecutive months (equivalent to the preset consecutive months) are calculated. The cumulative natural runoff sequence for 1 consecutive month reflects the flow change on the monthly time scale, the cumulative natural runoff sequence for 3 consecutive months reflects the flow change on the seasonal scale, and the cumulative natural runoff sequence for 12 consecutive months reflects the flow change on the annual scale. Runoff data at different time scales are thus constructed. The specific calculation formula is as follows:

[0045]

[0046] Where, is the cumulative natural runoff in the jth month of the i-th year on the k-th time scale, with k values ​​equal to 1, 3, and 12, corresponding to the monthly, seasonal, and annual time scales, respectively; Q i,j is the natural runoff in the jth month of the i-th year;

[0047] Several applicable frequency distribution functions (equivalent to the initial frequency distribution function) are selected to fit the runoff data on three time scales: monthly, seasonal, and annual. Each frequency distribution function corresponds to a data fitting distribution frequency diagram. Among them, several applicable frequency distribution functions include normal distribution norm, lognormal distribution logn, generalized extreme value distribution gev, and exponential distribution exp. The KS test is used to select the most applicable frequency distribution function for each watershed at a significance level of 5% (equivalent to the preset test requirement). In this embodiment, the processing of a watershed with a region code of 1010001 on a monthly time scale is taken as an example. See Figure 2 , Figure 2 A schematic diagram of the fitting distribution frequency of monthly time-scale flow data of a basin coded as 1010001 in a certain area according to a method for identifying key human factors affecting hydrological drought provided by an embodiment of the present invention is shown in FIG. Figure 2 As shown in the figure, from the fitting results, the exponential distribution fitting effect of the basin coded as 1010001 in a certain area is the best. Therefore, the exponential distribution is used as the target frequency distribution function for subsequent processing of this basin. After selecting the optimal target frequency distribution function for each basin, the hydrological drought judgment threshold is set. It is worth mentioning that the processing process for the seasonal time scale and the annual time scale is the same.

[0048] In the above scheme, historical natural runoff data are analyzed, multi-time scale cumulative runoff series are constructed, and multiple frequency distribution function goodness-of-fit tests are used to screen the optimal target frequency distribution function, and the hydrological drought judgment thresholds under different time dimensions are accurately quantified. Thus, the reliability of the hydrological drought judgment thresholds is enhanced by strictly screening the frequency distribution functions, providing a reliable data standard for subsequent multi-time scale dynamic analysis to capture the differentiated interference patterns of human activities on hydrological droughts, thereby improving the attribution accuracy of human factors on hydrological droughts, and further improving the accuracy of identifying human factors affecting hydrological droughts.

[0049] As a preferred solution, a standardized runoff index sequence is calculated based on a target frequency distribution function and cumulative natural runoff sequences at different time scales, and the runoff value corresponding to an index in the standardized runoff index sequence equal to a preset threshold is used as a hydrological drought determination threshold, including: fitting the cumulative natural runoff sequences at different time scales based on the target frequency distribution function to obtain a target fitting result; converting the target fitting result into a standardized runoff index sequence based on a preset standardization processing algorithm; and selecting a runoff value corresponding to an index equal to -1 in the standardized runoff index sequence as a hydrological drought determination threshold.

[0050] As a preferred implementation method, the optimal target frequency distribution function of each basin is used to fit the runoff data on three time scales and the target fitting results are standardized to convert the runoff data on different time scales into a standardized runoff series with a mean of 0 and a variance of 1. Figure 3 , Figure 3 A schematic diagram of monthly time scale flow data thresholds for a river basin coded as 1010001 in a certain region according to a method for identifying key human factors affecting hydrological drought provided by an embodiment of the present invention; Figure 3 As shown in the figure, taking the basin with the target frequency distribution function as the exponential distribution as an example, the corresponding runoff value with the standardized runoff sequence index equal to -1 (equivalent to the preset data threshold) is used as the hydrological drought judgment threshold. Figure 3 In the example, on a monthly time scale, a monthly flow of 117 cubic meters / month is used as the hydrological drought determination threshold for a watershed in a region coded as 1010001. In this embodiment, the fitting process and the standardization process can be implemented using existing technologies, and the processing processes for the seasonal and annual time scales are similar.

[0051] In the above scheme, the target frequency distribution function is used to fit multi-time scale runoff data to generate a standardized runoff index sequence. The runoff value corresponding to the index equal to -1 in the standardized runoff index sequence is selected as the threshold for hydrological drought judgment, and a quantitative benchmark for the natural hydrological state is established. The statistical distribution characteristics of the standardized runoff index sequence are used to analyze the degree of deviation between the measured runoff and the natural benchmark, and the natural fluctuations and human interference signals are effectively separated through multi-time scale comparison. This method enhances the characterization ability of the hydrological drought judgment threshold through standardized processing at a dynamic scale, and combines distribution differences to quantitatively identify the disturbance intensity of human activities on the runoff process, thereby improving the objectivity and accuracy of the human attribution of drought, and further improving the accuracy of identifying human factors affecting hydrological drought.

[0052] As a preferred solution, based on a preset study period, the hydrological drought judgment threshold and the natural runoff data of the study area at different time scales are compared to obtain the frequency of natural hydrological drought within the study period, including: based on the preset study period and the cumulative natural runoff series at different time scales, obtaining the number of time periods in which the natural runoff data in the study area is less than the hydrological drought judgment threshold; based on the total number of time periods in the preset study period and the number of time periods in which the natural runoff data in the study area is less than the hydrological drought judgment threshold, obtaining the frequency of natural hydrological drought within the study period.

[0053] As a preferred implementation method, the historical natural runoff data of the study area in each basin is compared with the hydrological drought determination threshold, and the frequency of natural hydrological drought at different time scales within the study period is statistically analyzed. The specific calculation formula for the frequency of natural hydrological drought at different time scales is as follows:

[0054]

[0055] Where, P n is the frequency of natural hydrological drought during the study period; n n is the number of periods when the natural runoff is less than the drought threshold; N is the total number of periods.

[0056] In the above scheme, the number of time periods in which the natural runoff data in the study area was lower than the hydrological drought judgment threshold was counted, and its ratio to the total number of time periods was calculated to accurately obtain the frequency of natural hydrological drought in the study period. This frequency value was used as a quantitative benchmark and compared with the measured drought frequency to effectively separate natural fluctuations and human disturbance signals, providing an objective reference for the subsequent identification of the degree of influence of human factors, quantitatively calculating the drought frequency at multiple time scales, significantly improving the measurability of the intensity of human interference, and thereby improving the accuracy of identifying human factors affecting hydrological drought.

[0057] As a preferred solution, the measured runoff data of the study area is obtained, and the hydrological drought judgment threshold and the measured runoff data of the study area at different time scales are compared to obtain the measured frequency of hydrological drought in the study period, including: calculating the cumulative measured runoff volume series at different time scales based on the measured runoff data of the study area at different time scales; obtaining the number of time periods in which the measured runoff data of the study area is less than the hydrological drought judgment threshold based on the preset study period and the cumulative measured runoff volume series at different time scales; obtaining the measured frequency of hydrological drought in the study period based on the total number of time periods in the preset study period and the number of time periods in which the measured runoff data of the study area is less than the hydrological drought judgment threshold.

[0058] As a preferred embodiment, the historical measured runoff data of the study area of ​​each river basin is compared with the hydrological drought determination threshold, and the measured hydrological drought occurrence frequency at different time scales during the study period is statistically analyzed. In this embodiment, the measured data from January 1961 to December 2010 provided by a regional geological survey bureau is used as the historical measured runoff data of the study area of ​​each river basin for explanation; the cumulative measured runoff data for 1, 3, and 12 consecutive months are calculated to reflect the flow changes on three different time scales: monthly, seasonal, and annual, respectively, to obtain the historical measured runoff data of the study area; the specific calculation formula for the measured hydrological drought occurrence frequency at different time scales is as follows:

[0059]

[0060] Where, P o is the observed frequency of hydrological drought during the study period; n o is the number of periods when the measured runoff is less than the drought threshold; N is the total number of periods;

[0061] Then, the difference between the frequency of natural hydrological drought at different time scales and the frequency of measured hydrological drought at different time scales was calculated, and the types of the degree of influence of human factors on hydrological drought were divided. The types of the degree of influence of human factors on hydrological drought can be divided into three types: "aggravation effect", "mitigation effect" and "neutral effect". Among them, "aggravation effect" refers to human activities causing hydrological drought to become more severe or frequent, "mitigation effect" refers to human activities helping to reduce the impact of hydrological drought, and "neutral effect" refers to human activities having no significant impact on hydrological drought.

[0062] In the above scheme, the number of time periods in which the measured runoff data in the study area is less than the hydrological drought judgment threshold is counted, and its ratio to the total number of time periods is calculated to accurately quantify the actual probability of drought under the interference of human activities. The measured frequency is compared with the natural background frequency, and the difference between the two is used to objectively separate the degree of influence of human factors on hydrological drought, providing a quantitative basis for the analysis of the synergistic effect of multiple factors, effectively capturing the dynamic interference of human activities on drought characteristics, and thus improving the accuracy of identifying human factors affecting hydrological drought.

[0063] As a preferred solution, the difference between the measured frequency of hydrological drought during the study period and the frequency of natural hydrological drought during the study period is calculated, and the degree of influence of human factors on hydrological drought in the study area is divided into aggravating effect, mitigating effect and neutral effect, including: calculating the difference between the measured frequency of hydrological drought during the study period and the frequency of natural hydrological drought during the study period to obtain the degree of influence of human factors on hydrological drought in the study area; if the degree of influence of human factors on hydrological drought in the study area is greater than a preset influence threshold, the degree of influence of human factors on hydrological drought in the study area is divided into aggravating effect; if the degree of influence of human factors on hydrological drought in the study area is less than the preset influence threshold, the degree of influence of human factors on hydrological drought in the study area is divided into mitigating effect; if the degree of influence of human factors on hydrological drought in the study area is equal to the preset influence threshold, the degree of influence of human factors on hydrological drought in the study area is divided into neutral effect.

[0064] As a preferred implementation method, the difference between the frequency of natural hydrological drought at different time scales and the frequency of measured hydrological drought at different time scales is calculated. The specific quantitative formula is as follows:

[0065] I=P o -P n ;

[0066] Where, I is the impact of human factors on the frequency of hydrological drought; P o is the observed frequency of hydrological drought during the study period; P n is the frequency of natural hydrological drought during the study period; see Figure 4 and Figure 5 , Figure 4 A schematic diagram of the change in the frequency of hydrological drought caused by human factors on a monthly time scale in a basin coded as 1010001 in a certain region, according to a method for identifying key human factors affecting hydrological drought provided by an embodiment of the present invention; Figure 5 A partially enlarged schematic diagram of the change in the frequency of hydrological drought caused by human factors on a monthly time scale in a basin coded as 1010001 in a certain area, according to a method for identifying key human factors affecting hydrological drought provided by an embodiment of the present invention; Figure 4 As shown in Figure 2, between 1961 and 2010, I = -0.13 was calculated, indicating that human factors caused a 13% decrease in the frequency of hydrological droughts on a monthly time scale. Figure 5 As shown, Figure 5 for Figure 4 The partial enlarged view of the solid line box area can reflect the identification principle of the human factors affecting the frequency of hydrological drought. Specifically, in the study period of the study area from October 2002 to January 2004, the frequency of the historical natural runoff data of the study area being lower than the hydrological drought determination threshold is 5, and the frequency of the historical measured runoff data of the study area being lower than the hydrological drought determination threshold is 1, that is, the human factors caused the frequency of hydrological drought on the monthly time scale in this period to decrease by 4; the operation on the seasonal time scale and the annual time scale is the same.

[0067] The impact of human factors on hydrological drought is then divided into three types: "aggravating effect," "mitigating effect," and "neutral effect." "Aggravating effect" refers to when human activities cause hydrological drought to become more severe or frequent; "mitigating effect" refers to when human activities help alleviate the impact of hydrological drought; and "neutral effect" refers to when human activities have no significant impact on hydrological drought. The specific classification is as follows:

[0068]

[0069] Where I is the degree of influence of human factors on the frequency of hydrological drought; E = 1 means that human factors have an aggravating effect on hydrological drought; E = 2 means that human factors have a neutral effect on hydrological drought; E = 3 means that human factors have a mitigating effect on hydrological drought.

[0070] In the above scheme, the difference between the frequency of natural hydrological drought at different time scales and the frequency of measured hydrological drought at different time scales is calculated, and the type of the degree of influence of human factors on hydrological drought is divided according to the difference threshold. The three types of aggravation effect, mitigation effect and neutral effect are proposed to break through the limitations of traditional single factor analysis, deconstruct the differentiated impact of human factors on hydrological drought, refine the classification of human factors affecting hydrological drought, and improve the accuracy of identifying human factors affecting hydrological drought.

[0071] As a preferred solution, based on the degree of influence of human factors on hydrological drought in the study area and the pre-acquired basic data of the study area, a classification model of the degree of influence of human factors on hydrological drought at different time scales is established through a random forest model, and the relative contribution rate of different human factors at different time scales is calculated, including: obtaining the basic data of human factors in the study area and calculating the average annual change rate of each human factor; using the average annual change rate of each human factor as a prediction factor and the degree of influence of human factors on hydrological drought at different time scales as the classification result, the prediction factor is input into a preset decision tree ensemble learning model for classification learning to obtain an initial classification model; based on the initial classification model and the prediction factor, the predicted degree of influence of human factors on hydrological drought is obtained; based on the hydrological drought judgment threshold and the prediction factor, the measured degree of influence of human factors on hydrological drought is obtained; based on the predicted degree of influence of human factors on hydrological drought and the measured degree of influence of human factors on hydrological drought, the model training accuracy is obtained; the initial classification model is optimized until the model training accuracy meets the preset training requirements to obtain a target classification model; based on the target classification model, the relative contribution rate of different human factors at different time scales is calculated.

[0072] As a preferred implementation method, basic data on human factors in the area to be studied are obtained and the average annual change rate of each type of data is calculated to simulate the impact of different human factors on hydrological drought. The basic data on human factors in the area to be studied can be obtained from a geological survey bureau in a certain area. The basic data on human factors in the area to be studied include: population density, water withdrawal, reservoir density, and land use type in the study area. A random forest classification model (equivalent to the initial classification model) of the impact of human factors on hydrological drought is established. The random forest is an ensemble learning model composed of multiple decision trees. The random forest is assumed to consist of M decision trees, denoted as T1, T2,…, T M , the final classification result is obtained by the majority voting method of all decision trees, which is expressed as:

[0073]

[0074] T i (x)=class i (x;θ i );

[0075] x=(x p ,x d ,x u ,x c ,x im ,x ir ,x f );

[0076] Where, represents the classification results of the impact of human factors on hydrological drought obtained using the random forest model; T i Represents the i-th decision tree, class i (x;θ i ) is the i-th tree according to the input feature x and split rule θ i The classification result obtained, θ i is the parameter of the i-th tree (tree structure and splitting rule); x is the input feature vector; x p 、x d 、x u 、x c 、x im 、x ir and x f They represent the average annual rate of change of population density, dam density, water use, proportion of cultivated land, proportion of impervious land, proportion of irrigated land, and proportion of forest area, respectively;

[0077] In this embodiment, the average annual change rate of the basic data of human factors in the study area is used as the prediction factor input, and the degree of impact of human factors on hydrological drought is used as the classification result. The random forest model is used to establish initial classification models at different time scales, such as x p 、x d 、x u 、x c 、x im 、x ir and x f They represent the average annual change rates of population density, dam density, water use, proportion of cultivated land, proportion of impervious area, proportion of irrigated area and proportion of forest area respectively, and are used as predictors of the random forest model to construct the initial classification model. Then, by inputting the predictors into the initial classification model, the degree of influence of human factors on hydrological drought is predicted, which can be recorded as Then, the hydrological drought threshold and prediction factor are compared to obtain the impact of measured human factors on hydrological drought, which can be recorded as E. The accuracy of the initial classification model at different time scales (equivalent to the model training accuracy) is calculated using E. The accuracy of the initial classification model at different time scales is evaluated. The accuracy of the initial classification model is continuously optimized until the accuracy of the initial classification model meets the training requirements. The initial classification model that meets the training requirements is selected as the target classification model. In this embodiment, the accuracy rates at the monthly, seasonal, and annual time scales reach 0.70, 0.70, and 0.77, respectively.

[0078] In the above scheme, the basic data of human factors in the study area are combined to quantify the average annual change rate of each human factor as a prediction factor, which is input into the decision tree ensemble learning model. The difference in frequency between measured and natural hydrological droughts is mapped to the degree of influence of the corresponding human factors on hydrological droughts. The model parameters are iteratively optimized by calculating the model training accuracy so that the matching degree between the predicted classification results and the actually observed drought types reaches the preset accuracy threshold. The nonlinear classification ability of machine learning is used to analyze the synergistic effect of multiple factors, and the dynamic annual change rate characteristics are used to capture the cumulative effects of human activities, so as to achieve accurate classification of heterogeneous human interference, thereby improving the accuracy of identifying human factors affecting hydrological droughts.

[0079] As a preferred solution, based on the target classification model, the relative contribution rates of different human factors at different time scales are calculated, including: obtaining the characteristic data of human factors in the area to be studied based on the average annual change rate of each human factor; inputting the characteristic data of human factors in the area to be studied into the target classification model to obtain the predicted contribution evaluation results of each decision tree to the characteristic data of human factors in the area to be studied; and performing weighted averaging on the predicted contribution evaluation results to obtain the characteristic contribution rate of the human factors to be identified.

[0080] As a preferred implementation method, in the trained random forest model (equivalent to the target classification model), each decision tree will evaluate the contribution value of different features (equivalent to the characteristic data of human factors in the area to be studied) in the prediction process to obtain the prediction contribution evaluation result. The weighted average of the contribution value of the characteristic data of human factors in each area to be studied is used to evaluate the relative importance of different human factors to obtain the contribution rate of the human factors to be identified. The random forest model measures the importance of features through the Gini coefficient. The core idea is that the more a feature reduces the Gini impurity when the decision tree node is split, the more important the feature is. Among them, the feature x j The importance of is the average reduction in Gini impurity among M decision trees, expressed as:

[0081]

[0082] Where, is feature x jThe reduction of Gini impurity of a node in a tree; G(t) is the Gini impurity of node t; t L and t R Use feature x for node t respectively j The two child nodes obtained after division, |t L | and |t R | are the number of samples of the left child node and the right child node respectively; p(c|t) is the proportion of samples of category c in node t, and C is the number of categories (in this embodiment, C=3).

[0083] In the above scheme, the target classification model integrates the tree structure characteristics of the decision tree model, performs a multi-dimensional split node importance evaluation on the annual change rate characteristics of the input human factors, and then performs a weighted average on the contribution results output by all decision trees. Finally, the global characteristic contribution rate of each factor to the drought type classification is output. By utilizing the decision tree group's ability to analyze nonlinear relationships, the complex mechanism of the synergistic impact of multiple factors on drought is effectively captured, the contribution quantification error of the human factor characteristics to be identified is reduced, the attribution accuracy under the composite impact scenario is improved, and the accuracy of identifying human factors affecting hydrological drought is thereby improved.

[0084] As a preferred solution, the key human factors affecting hydrological drought are identified based on the contribution rate of the characteristics of the human factors to be identified, including: calculating the contribution rates of different human factor characteristics at different time scales to obtain the contribution rates of several human factor characteristics to be identified; sorting the contribution rates of several human factor characteristics to be identified, and screening the characteristics of the human factors to be identified that meet the preset contribution rate requirements as the key human factors affecting hydrological drought.

[0085] As a preferred implementation method, the contribution rates of the human factor features to be identified are sorted, and the human factor features to be identified whose contribution rates meet the preset contribution rate requirements are selected as the key human factors affecting hydrological drought. The preset contribution rate requirements can be set according to actual conditions, see Figure 6 , Figure 6 A schematic diagram of the relative contribution rate ranking of different human factors affecting hydrological drought on a monthly time scale in a certain region of land according to a method for identifying key human factors affecting hydrological drought provided by an embodiment of the present invention; Figure 6 As shown in the data, on a monthly time scale, the contribution rates of dam density, cultivated land area, and irrigated area reached 17.56%, 15.40%, and 14.86%, respectively. This identified dam construction and operation, as well as agricultural irrigation, as the anthropogenic factors with the greatest impact on hydrological drought on a monthly time scale. Therefore, dam construction and operation, as well as agricultural irrigation, are the key anthropogenic factors influencing hydrological drought on a monthly time scale. Similarly, the same method was used to identify the key anthropogenic factors influencing hydrological drought on a seasonal and annual time scale.

[0086] In the above scheme, the contribution rates of human factor characteristics at multiple time scales are calculated and ranked, a contribution gradient screening mechanism is established, key human factors are screened according to preset contribution rate requirements, and the spatiotemporal distribution characteristics of contribution rates are used to analyze the persistence and cumulative effects of human activities, accurately distinguish the key human factors affecting hydrological drought, and improve the accuracy of identifying human factors affecting hydrological drought.

[0087] Example 2

[0088] See also Figure 7 , Figure 7 This is a schematic diagram of the module structure of a key human factor identification system that affects hydrological drought provided by an embodiment of the present invention. Figure 7 As shown, the embodiment of the present invention proposes a key human factor identification system affecting hydrological drought, including: a hydrological drought judgment threshold calculation module 201, a natural hydrological drought occurrence frequency calculation module 202, a measured hydrological drought occurrence frequency calculation module 203, a human factor influence degree classification module 204, a contribution rate calculation module 205 and a key human factor identification module 206; the hydrological drought judgment threshold calculation module 201 is used to obtain natural runoff data in the study area, calculate the standardized runoff index of different time scales, and use the runoff value corresponding to the standardized runoff index equal to the preset threshold as the hydrological drought judgment threshold; wherein the different time scales include: month, season and year; the natural hydrological drought occurrence frequency calculation module 202 is used to compare the hydrological drought judgment threshold and the natural runoff data of the study area at different time scales based on a preset study period to obtain the natural hydrological drought occurrence frequency within the study period; the measured hydrological drought occurrence frequency calculation module 203 03 is used to obtain the measured runoff data of the study area, compare the hydrological drought judgment threshold and the measured runoff data of the study area at different time scales, and obtain the measured frequency of hydrological drought in the study period; the human factor influence degree classification module 204 is used to calculate the difference between the measured frequency of hydrological drought in the study period and the frequency of natural hydrological drought in the study period, and divide the degree of influence of human factors on hydrological drought in the study area into aggravating effect, mitigating effect and neutral effect; the contribution rate calculation module 205 is used to establish a classification model of the degree of influence of human factors on hydrological drought at different time scales through a random forest model based on the degree of influence of human factors on hydrological drought in the study area and the pre-acquired basic data of the study area, and calculate the relative contribution rate of different human factors at different time scales; the key human factor identification module 206 is used to identify the key human factors affecting hydrological drought based on the relative contribution rate of different human factors at different time scales.

[0089] As a preferred embodiment, in the process of analyzing the runoff of the study area, it usually refers to analyzing the runoff of each basin in the study area. In this embodiment, 2099 basins on land in a certain area are taken, and the study period is set from October 2002 to January 2004. The monthly time scale is used as an example for specific explanation, which will not be repeated below; the hydrological drought judgment threshold calculation module 201 collects the historical natural runoff data of the study area, calculates the standardized runoff data series of the time scales of month, season and year, and selects the standardized runoff from the standardized runoff data series of different scales. The value of -1 is used as the hydrological drought determination threshold, and the hydrological drought determination threshold is used to quantitatively evaluate the hydrological drought situation of each basin in the study area. The natural hydrological drought occurrence frequency calculation module 202 and the measured hydrological drought occurrence frequency calculation module 203 then obtain the historical measured runoff data of the study area within the study period according to the set study period, and combine the historical natural runoff data of the study area and the hydrological drought determination threshold to obtain the historical natural hydrological drought events and the historical measured hydrological drought events within the study period, and correspondingly calculate the frequency of natural hydrological drought and the measured hydrological drought. The human factors impact degree classification module 204 divides the types of human factors impact on hydrological drought by calculating the difference between the frequency of natural hydrological drought and the frequency of measured hydrological drought. The types of human factors impact on hydrological drought can be divided into three types: "aggravation effect", "mitigation effect" and "neutral effect". Among them, "aggravation effect" means that human activities cause hydrological drought to become more severe or frequent, "mitigation effect" means that human activities help to reduce the impact of hydrological drought, and "neutral effect" means that human activities have no significant impact on hydrological drought. Then, according to the impact of human factors on hydrological drought, the human factors impact on hydrological drought can be classified into three types: "aggravation effect", "mitigation effect" and "neutral effect". The impact type and the basic data of human factors in the area to be studied are combined with an ensemble learning model to construct a target classification model. In this embodiment, the ensemble learning model can adopt a random forest model. The contribution rate calculation module 205 then calculates the relative contribution rate of each feature of the basic data of human factors in the area to be studied at different time scales through the target classification model. The basic data of human factors in the area to be studied include: population density, water withdrawal, reservoir density, and land use type. Finally, the key human factor identification module 206 identifies the key human factors affecting hydrological drought at multiple time scales according to the relative contribution rate of each feature at different time scales.

[0090] An embodiment of the present invention provides a system for identifying key human factors affecting hydrological drought. The hydrological drought threshold calculation module combines a natural hydrological drought occurrence frequency calculation module and a measured hydrological drought occurrence frequency calculation module, using dynamic comparative analysis to couple historical natural runoff with measured runoff. This system constructs a hydrological drought threshold assessment system to calculate the measured hydrological drought occurrence frequency and the natural hydrological drought occurrence frequency within a study period. The human factor impact degree classification module classifies the degree of human factor influence on hydrological drought within the study area into exacerbation effect, mitigation effect, and neutral effect. The contribution rate calculation module constructs a classification model based on the degree of human factor influence on hydrological drought, and uses the classification model to quantify the contribution rate of the human factor characteristics to be identified. The key human factor identification module then identifies the key human factors affecting hydrological drought based on the contribution rate of the human factor characteristics to be identified. Through multi-source data fusion and dynamic coupling modeling, the synergistic impact of complex human activities on hydrological drought is effectively quantified, the key human factors affecting hydrological drought are quantitatively attributed, and the accuracy of identifying human factors affecting hydrological drought is improved.

[0091] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the technical principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

[0092] In the description of this specification, the reference terms "one embodiment," "some embodiments," "examples," "specific examples," or "some examples" mean that the specific features, structures, materials, or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present application. Moreover, the specific features, structures, materials, or characteristics described may be combined in any appropriate manner in any one or more embodiments or examples. In addition, those skilled in the art may combine and integrate different embodiments or examples described in this specification, as well as features of different embodiments or examples, unless they are mutually inconsistent.

[0093] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be understood to indicate or imply relative importance or implicitly specify the number of technical features indicated. Thus, features specified as "first" or "second" may explicitly or implicitly include at least one of such features. In the description of this application, "plurality" means two or more, unless otherwise specifically defined.

Claims

1. A method for identifying key human factors affecting hydrological drought, characterized by: include: Obtain natural runoff data for the study area, calculate standardized runoff indexes at different time scales, and use the runoff value corresponding to the standardized runoff index being equal to a preset threshold as the hydrological drought determination threshold; wherein the different time scales include: month, season, and year; Based on a preset research period, the hydrological drought determination threshold is compared with the natural runoff data of the research area at different time scales to obtain the frequency of natural hydrological drought in the research period; Obtaining measured runoff data of the study area, comparing the hydrological drought determination threshold with the measured runoff data of the study area at different time scales, and obtaining the measured frequency of hydrological drought within the study period; Calculate the difference between the observed frequency of hydrological drought during the study period and the frequency of natural hydrological drought during the study period, and classify the degree of human influence on hydrological drought in the study area into aggravating effect, mitigating effect, and neutral effect; Based on the impact of human factors on hydrological drought in the study area and pre-acquired basic data of the study area, a classification model of the impact of human factors on hydrological drought at different time scales was established using a random forest model, and the relative contribution rate of different human factors at different time scales was calculated; Based on the relative contribution rates of different human factors at different time scales, the key human factors affecting hydrological drought are identified.

2. The method for identifying key human factors affecting hydrological drought according to claim 1, characterized in that: The natural runoff data of the study area is obtained, the standardized runoff index of different time scales is calculated, and the runoff value corresponding to the standardized runoff index being equal to the preset threshold is used as the hydrological drought determination threshold; wherein the different time scales include: month, season and year, including: Calculate the cumulative natural runoff series at different time scales based on the natural runoff data of the study area; fitting the cumulative natural runoff series of different time scales by using a plurality of initial frequency distribution functions to obtain a fitting distribution frequency graph corresponding to each initial frequency distribution function; Based on the fitted distribution frequency graph corresponding to each initial frequency distribution function, the frequency distribution function that meets the preset test requirements is screened to obtain the target frequency distribution function; A standardized runoff index sequence is calculated based on the target frequency distribution function and the cumulative natural runoff sequence at different time scales, and the runoff value corresponding to the index in the standardized runoff index sequence being equal to a preset threshold is used as the hydrological drought judgment threshold, wherein the different time scales include: month, season and year.

3. The method for identifying key human factors affecting hydrological drought according to claim 2, characterized in that: Calculating a standardized runoff index sequence based on the target frequency distribution function and the cumulative natural runoff sequences at different time scales, and using the runoff value corresponding to the index in the standardized runoff index sequence being equal to a preset threshold as a hydrological drought determination threshold, including: Fitting the cumulative natural runoff series at different time scales based on the target frequency distribution function to obtain a target fitting result; Based on a preset standardization processing algorithm, the target fitting result is converted into a standardized runoff index sequence; The runoff value corresponding to the index equal to -1 in the standardized runoff index sequence is selected as the hydrological drought determination threshold.

4. The method for identifying key human factors affecting hydrological drought according to claim 2, characterized in that: Based on a preset study period, the hydrological drought determination threshold is compared with the natural runoff data of the study area at different time scales to obtain the frequency of natural hydrological drought in the study period, including: Based on the preset research period and the cumulative natural runoff series at different time scales, obtaining the number of time periods in which the natural runoff data of the research area is less than the hydrological drought determination threshold; Based on the total number of time periods in the preset research period and the number of time periods in which the natural runoff data of the research area is less than the hydrological drought determination threshold, the frequency of occurrence of natural hydrological drought in the research period is obtained.

5. The method for identifying key human factors affecting hydrological drought according to claim 4, characterized in that: Obtain measured runoff data for the study area, compare the hydrological drought determination threshold with the measured runoff data for the study area at different time scales, and obtain the measured frequency of hydrological drought within the study period, including: Based on the measured runoff data of the study area at different time scales, calculating the cumulative measured runoff series at different time scales; Based on the preset research period and the cumulative measured runoff series at different time scales, obtaining the number of time periods in which the measured runoff data of the research area is less than the hydrological drought determination threshold; Based on the total number of time periods in the preset research period and the number of time periods in which the measured runoff data in the research area is less than the hydrological drought determination threshold, the measured frequency of occurrence of hydrological drought in the research period is obtained.

6. The method for identifying key human factors affecting hydrological drought according to claim 1, characterized in that: The difference between the observed frequency of hydrological drought during the study period and the frequency of natural hydrological drought during the study period was calculated, and the degree of influence of human factors on hydrological drought in the study area was divided into aggravating effect, mitigating effect and neutral effect, including: Calculate the difference between the measured frequency of hydrological drought during the study period and the frequency of natural hydrological drought during the study period to obtain the degree of influence of human factors on hydrological drought in the study area; If the degree of influence of human factors on hydrological drought in the study area is greater than a preset impact threshold, the degree of influence of human factors on hydrological drought in the study area is classified as an aggravating effect; If the degree of influence of human factors on hydrological drought in the study area is less than a preset impact threshold, the degree of influence of human factors on hydrological drought in the study area is classified as a mitigation effect; If the degree of influence of human factors on hydrological drought in the study area is equal to a preset influence threshold, the degree of influence of human factors on hydrological drought in the study area is classified as a neutral effect.

7. The method for identifying key human factors affecting hydrological drought according to claim 6, characterized in that: Based on the degree of impact of human factors on hydrological drought in the study area and the pre-acquired basic data of the study area, a classification model of the impact of human factors on hydrological drought at different time scales was established using a random forest model, and the relative contribution rates of different human factors at different time scales were calculated, including: Obtain basic data on human factors in the area to be studied and calculate the average annual change rate of each human factor; Taking the average annual change rate of each of the human factors as a prediction factor, and the degree of impact of the human factors on hydrological drought at different time scales as a classification result, the prediction factors are input into a preset decision tree ensemble learning model for classification learning to obtain an initial classification model; Based on the initial classification model and the prediction factors, the degree of influence of human factors on hydrological drought is predicted; Based on the hydrological drought determination threshold and the prediction factor, obtaining the degree of influence of measured human factors on hydrological drought; Obtaining a model training accuracy rate based on the predicted degree of influence of human factors on hydrological drought and the measured degree of influence of human factors on hydrological drought; Optimizing the initial classification model until the model training accuracy meets the preset training requirements to obtain a target classification model; Based on the target classification model, the relative contribution rates of different human factors at different time scales are calculated.

8. The method for identifying key human factors affecting hydrological drought according to claim 7, characterized in that: Based on the target classification model, the relative contribution rates of different human factors at different time scales are calculated, including: Based on the average annual change rate of each human factor, the characteristic data of human factors in the area to be studied are obtained; Inputting the human factor characteristic data of the area to be studied into the target classification model to obtain a prediction contribution evaluation result of each decision tree to the human factor characteristic data of the area to be studied; The predicted contribution evaluation results are weighted averaged to obtain the characteristic contribution rate of the human factor to be identified.

9. The method for identifying key human factors affecting hydrological drought according to claim 8, characterized in that: Based on the contribution rate of the human factors to be identified, the key human factors affecting hydrological drought are identified, including: By calculating the contribution rates of different human factor characteristics at different time scales, the contribution rates of several human factor characteristics to be identified are obtained; The contribution rates of the characteristics of the human factors to be identified are ranked, and the characteristics of the human factors to be identified that meet the preset contribution rate requirements are selected as the key human factors affecting hydrological drought.

10. A system for identifying key human factors affecting hydrological drought, characterized by: Executing a method for identifying key human factors affecting hydrological drought according to any one of claims 1 to 8, comprising: Hydrological drought determination threshold calculation module, natural hydrological drought occurrence frequency calculation module, measured hydrological drought occurrence frequency calculation module, human factor influence degree classification module, contribution rate calculation module and key human factor identification module; The hydrological drought determination threshold calculation module is used to obtain natural runoff data in the study area, calculate the standardized runoff index at different time scales, and use the runoff value corresponding to the standardized runoff index being equal to a preset threshold as the hydrological drought determination threshold; wherein the different time scales include: month, season, and year; The natural hydrological drought occurrence frequency calculation module is used to compare the hydrological drought determination threshold and the natural runoff data of the study area at different time scales based on a preset study period to obtain the natural hydrological drought occurrence frequency within the study period; The measured hydrological drought occurrence frequency calculation module is used to obtain measured runoff data of the study area, compare the hydrological drought determination threshold with the measured runoff data of the study area at different time scales, and obtain the measured hydrological drought occurrence frequency within the study period; The human factor impact degree classification module is used to calculate the difference between the measured hydrological drought occurrence frequency during the study period and the natural hydrological drought occurrence frequency during the study period, and classify the impact degree of human factors on hydrological drought in the study area into aggravating effect, mitigating effect and neutral effect; The contribution rate calculation module is used to establish a classification model of the impact of human factors on hydrological drought at different time scales through a random forest model based on the impact of human factors on hydrological drought in the study area and pre-acquired basic data of the study area, and calculate the relative contribution rates of different human factors at different time scales; The key human factor identification module is used to identify the key human factors affecting hydrological drought based on the relative contribution rates of different human factors at different time scales.

Images