Non-consistent hydrological drought attribution method based on successive runoff reduction

Abstract

The invention provides a non-consistent hydrological drought attribution method based on successive runoff restoration. The method comprises the following three steps: fusing multi-source data to successively restore runoff, constructing a non-consistent drought index, and calculating contribution degrees of various influence factors and interaction of the influence factors. Specifically, multi-source data and fusion algorithms such as terrain, land utilization, remote sensing images, global scale published water consumption, water resource announcement and the like are adopted to gradually restore an actually measured runoff series; calculating a corresponding inconsistent hydrological drought index by adopting an actual measurement runoff series, a primary reduction runoff series and a secondary reduction runoff series; hydrological drought indexes of different reduction series and periods are adopted, an elastic coefficient method is combined, and the contribution degree of reservoir regulation and storage, water taking and using, climate changes and interaction of all factors to hydrological drought is calculated. According to the method, the runoff is gradually restored by fusing multi-source data, and the non-consistent hydrological drought index is constructed, so that the problem of hydrological drought attribution under the comprehensive influence of multiple factors and interaction thereof is solved.

Description

Technical Field

[0001] The present invention relates to the technical field of extreme hydrological event analysis, and in particular to an inconsistent hydrological drought attribution method based on successive runoff reduction. Background Art

[0002] As a natural disaster, drought has caused many negative impacts on my country's food security, water resources supply, ecological environment, social economy and energy supply. The frequency, duration, intensity and impact range of droughts are also on the rise worldwide. Affected by climate change and strong human activities, hydrological drought refers to abnormal water shortage caused by the imbalance between precipitation and surface water or groundwater. Therefore, in a changing environment, clarifying the cause mechanism of hydrological drought is a theoretical prerequisite for improving the comprehensive response capacity to extreme droughts.

[0003] Constructing a drought index to assess hydrological drought and using attribution methods to calculate the contribution of influencing factors to hydrological drought are important ways to conduct attribution analysis of hydrological drought. Commonly used hydrological drought indices include: surface water supply index (SWSI), streamflow drought index (SDI), standardized streamflow index (SSI) and standardized runoff index (SRI). Commonly used attribution methods include: comparative test basin method, before and after disturbance method, upstream and downstream comparison method and simulation observation method.

[0004] However, the existing hydrological drought attribution methods have the following shortcomings. First, most of the existing attribution methods rely on constructing hydrological models for the base period and the impact period to calculate the contribution of climate change and human activities. However, when the basin is large and there are many influencing factors, the cost of building the model is high; and it is difficult to calculate the interaction of multiple influencing factors. Therefore, there is an urgent need for a method suitable for calculating the contribution of multiple influencing factors and their interactions to hydrological drought; and this method should be simple and direct, and be able to make full use of multi-source data for attribution analysis without the need to build complex models. Second, the changing environment has caused inconsistency in the runoff time series. Therefore, in a changing environment, the reliability of SRI estimates based on the premise of consistency of the runoff time series has been questioned. Summary of the invention

[0005] The purpose of the present invention is to address the deficiencies of the above-mentioned prior art and provide a non-consistent hydrological drought attribution method based on successive runoff restoration, so as to achieve refined and dynamic management of water resources and prevent and alleviate drought events.

[0006] To achieve the above object, the present invention adopts the following technical solutions:

[0007] The present invention provides a non-consistent hydrological drought attribution method based on successive runoff reduction, comprising:

[0008] S1. Fusion of multi-source data to restore runoff one by one, i.e., using multi-source data such as topography, land use, remote sensing images, global-scale water use and water resources bulletins and fusion algorithms to restore the measured runoff series one by one;

[0009] S2. Construct a non-consistent drought index, that is, use the measured runoff series and the first restored runoff series to calculate the corresponding non-consistent hydrological drought index, and use the second restored runoff series to calculate the hydrological drought index.

[0010] Furthermore, in S1, fusing multi-source data to restore runoff successively includes:

[0011] According to the measured runoff series R obs , the reservoir water storage change data obtained by remote sensing inversion, using water balance, considering the impact of water conservancy projects on the runoff series, and obtaining the primary reduction runoff series R re1 , calculated as:

[0012]

[0013] Where ΔRV t is the storage variable of the upstream reservoir group in the section within the period t;

[0014] According to the primary reduction runoff series R re1 , the water withdrawal data obtained by multi-source fusion, based on the primary reduction, further consider the impact of water withdrawal on the runoff series using water balance, and obtain the secondary reduction runoff series R re2 , calculated as:

[0015]

[0016] Among them, WU t is the water consumption of the upstream section in time period t; ζ is the water consumption rate.

[0017] Furthermore, the reservoir water storage capacity change data obtained by remote sensing inversion is obtained by:

[0018] S101. Construct a water level-area relationship curve based on the digital elevation model (DEM) data: starting from the lowest water level of the reservoir, count the number of raster pixels in the digital elevation model (DEM) that are less than or equal to the water level, and multiply the number of pixels by the pixel area to obtain the water surface area corresponding to the water level; repeat the calculation at intervals of 1 meter from the lowest water level to the highest water level of the reservoir to obtain a series of water level-area values; and use polynomial fitting to obtain a water level-area relationship curve based on the series of water level-area values.

[0019] S102. Extract the change process of reservoir water surface area using remote sensing images: On the Google Earth Engine GEE platform, use the image series and median synthesis to reduce the impact of clouds, and calculate the improved normalized water index MNDWI as follows:

[0020]

[0021] Among them, ρ(GREEN) is the green band, ρ(SWIR1) is the shortwave infrared band;

[0022] According to the normalized water body index MNDWI and threshold, the change process of reservoir water surface area is extracted.

[0023] S103. According to the reservoir water level area relationship curve and the reservoir water surface area change process, the reservoir water storage capacity change process is calculated as follows:

[0024]

[0025] Among them, RA is the surface area of ​​the reservoir and RE is the water level of the reservoir.

[0026] Furthermore, the water use data obtained by multi-source fusion is obtained by:

[0027] S104. Publish water use data on a global scale R As a reference, the water resources bulletin water use data WU G The calculation for interpolation to make up for missing data is:

[0028]

[0029] in, is the water use data of the water resources bulletin for period t, and i <t<j;R j and R i They represent the ratios of the public water consumption data and the reference water consumption data in time period j and time period i respectively.

[0030] S105, assuming that the time variation of water use is an S-shaped curve, the water use data WU G Perform forward extension to make up for the missing data, and the calculation is:

[0031]

[0032] Water Resources Bulletin Water Use Data WU G Backward extension is performed to make up for the missing data, and the calculation is:

[0033]

[0034] Among them, α is the corrected growth rate, and its value range is [0.1, 0.2]; K is the corrected water use change rate.

[0035] S106, assuming that water is only used for construction land and cultivated land, using the land use data of each period, obtain the water use data WU of the water resources bulletin after being processed by S104 and S105 G Spatial distribution data Calculated as:

[0036]

[0037] Among them, A LU Indicates the area of ​​construction land or cultivated land;

[0038] Spatial distribution data of water use based on water resources bulletin Know the water consumption WU in any section within the above period t t .

[0039] Further, in S2, the calculation of the corresponding inconsistent hydrological drought index is specifically as follows:

[0040] S201, according to the measured runoff series R obs , taking reservoir index RI and water intake index WUI as explanatory variables, the first model is constructed to calculate the corresponding measured runoff series R obs The first distribution function parameter μ obs and the second distribution function parameter σ obs :

[0041]

[0042] Among them, g(.) represents the link function; k is the coefficient vector; X is the explanatory variable;

[0043] According to the first distribution function parameter μ obs and the second distribution function parameter σ obs , calculate the corresponding measured runoff series R obs The SRI is a non-uniform hydrological drought index. obs .

[0044] S202, according to the primary reduction runoff series R re1 , taking the water withdrawal index WUI as the explanatory variable, the second model is constructed to calculate the corresponding first-time restored runoff series R re1 The third distribution function parameter μ re1 and the fourth distribution function σ re1 :

[0045]

[0046] According to the third distribution function parameter μre1 and the fourth distribution function parameter σ re1 , calculate the corresponding primary reduction runoff series R re1 The SRI is a non-uniform hydrological drought index. re1 .

[0047] S203, according to the secondary reduction runoff series R re2 , calculate the corresponding secondary reduction runoff series R re2 The Hydrological Drought Index (SRI) re2 .

[0048] Furthermore, it also includes calculating the contribution of multiple influencing factors and their interactions, specifically:

[0049] S301, secondary reduction runoff series R re2 Perform mutation point analysis and divide the analysis period into two periods before and after the impact of climate change, namely the first period and the second period, to obtain SRI obs_1 、SRI obs_2 、SRI re1_1 、SRI re1_2 、SRI re2_1 and SRI re2_2 ;

[0050] Among them, SRI obs_1 and SRI obs_2 Represents the SRI of the first period and the second period respectively obs , SRI obs_1 Affected by water conservancy projects and water withdrawal, SRI obs_2 Affected by water conservancy projects, water withdrawal, climate change, the interaction between water conservancy projects and climate change, and the interaction between water withdrawal and climate change;

[0051] SRI re1_1 and SRI re1_2 Represents the SRI of the first period and the second period respectively re1 , SRI re1_1 Affected by water withdrawal, SRI re1_2 Affected by water withdrawal, climate change, and the interaction between water withdrawal and climate change;

[0052] SRI re2_1 and SRI re2_2 Represents the SRI of the first period and the second period respectively re2 , SRI re2_1 Nearly natural state, SRI re2_2 Affected by climate change.

[0053] S302. Calculate the contribution of water conservancy project operation to hydrological drought α R, Contribution of water withdrawal to hydrological drought α W 、Contribution of climate change to hydrological drought α C , the contribution of the interaction between water conservancy projects and climate change to hydrological drought α RC And the contribution of the interaction between water withdrawal and climate change to hydrological drought α WC :

[0054]

[0055] Among them, ΔSRI R , ΔSRI W , ΔSRI C , ΔSRI RC and ΔSRI WC They respectively represent the changes in the inconsistent hydrological drought index (SRI) caused by water conservancy project scheduling and operation, water withdrawal, climate change, the interaction between water conservancy projects and climate change, and the interaction between water withdrawal and climate change.

[0056] S303. Calculate the variation of the non-consistent hydrological drought index SRI, including ΔSRI R , ΔSRI W , ΔSRI C , ΔSRI RC and ΔSRI WC :

[0057]

[0058] Among them, ΔSRI R1 and ΔSRI R2 represents the change of hydrological drought index caused by water conservancy projects in the first period and the second period, respectively, and ΔSRI R1 =ΔSRI R2 ;

[0059] ΔSRI W1 and ΔSRI W2 represents the change of hydrological drought index caused by water withdrawal in the first period and the second period, respectively, and ΔSRI W1 =ΔSRI W2 ;

[0060] f is SRI re2_1 About atmospheric circulation factor a i The expression f is constructed by regression equation, proxy model or machine learning. A is the set of atmospheric circulation factors, Δa i is the atmospheric circulation factor a affected by climate change i The average value and the atmospheric circulation factor a before the impact of climate change i The difference in means.

[0061] The beneficial effects of the present invention are: using multi-source data and fusion algorithms such as topography, land use, remote sensing images, global-scale published water use, and water resources bulletins to gradually restore the measured runoff series and construct a non-consistent hydrological drought index, solving the problem of hydrological drought attribution under the comprehensive influence of multiple factors and their interactions;

[0062] Constructing an inconsistency index under a changing environment is conducive to improving the reliability of hydrological drought attribution and reducing the error of hydrological drought attribution. BRIEF DESCRIPTION OF THE DRAWINGS

[0063] Figure 1 It is a structural block diagram of a non-consistent hydrological drought attribution method based on successive runoff restoration according to an embodiment of the present invention.

[0064] Figure 2 This is the change process of water storage capacity of four large reservoirs (Wudongde, Baihetan, Xiluodu and Xiangjiaba) in the lower reaches of Jinsha River.

[0065] Figure 3 This is the water use process above the Yichang section.

[0066] Figure 4 This is the spatial distribution of water consumption above the Yichang section in 2020.

[0067] Figure 5 It is a comparison diagram of the consistency and inconsistency distribution functions of spring natural runoff according to an embodiment of the present invention.

[0068] Figure 6 This is a dynamic change diagram of the non-consistent hydrological drought index according to an embodiment of the present invention.

[0069] Table 1 shows the optimal distribution function types of the annual runoff in each season according to the embodiment of the present invention.

[0070] Table 2 is the attribution analysis results of various influencing factors of the embodiment of the present invention. DETAILED DESCRIPTION

[0071] The present embodiment is described in detail below, and examples of the embodiment are shown in the accompanying drawings. The embodiments described below with reference to the accompanying drawings are exemplary and intended to be used to explain the present invention, but should not be understood as limiting the present invention.

[0072] As described in the background technology, the existing attribution methods have the following shortcomings:

[0073] (1) Most existing attribution methods rely on constructing hydrological models for the base period and the impact period to calculate the contribution of climate change and human activities to hydrological drought. When the basin is large and there are many influencing factors, the cost of building the model is high; and it is difficult to calculate the interaction of multiple influencing factors. (2) The consistency assumptions based on most existing hydrological indices are not conducive to the reliability of hydrological drought attribution under a changing environment.

[0074] To this end, this embodiment provides a non-consistent hydrological drought attribution method based on successive runoff restoration. This method uses multi-source data such as topography, land use, remote sensing images, global-scale published water use, and water resources bulletins and fusion algorithms to successively restore the measured runoff series and construct a non-consistent hydrological drought index, solving the problem of hydrological drought attribution under the combined influence of multiple factors and their interactions, and improving the reliability of hydrological drought attribution under a changing environment, thereby helping to deeply understand the evolution mechanism of hydrological drought under non-consistent conditions, and realize the refined and dynamic management of water resources, and prevent and alleviate drought events.

[0075] Attached Figure 1 FIG. 1 is a structural block diagram of a non-uniform hydrological drought attribution method based on successive runoff reduction according to an embodiment of the present invention. Figure 1 Taking the hydrological drought attribution analysis of Yichang Station as an example, the non-consistent hydrological drought attribution method based on successive runoff reduction is described.

[0076] (1) Fusion of multi-source data to restore runoff one by one:

[0077] (1.1) Primary runoff reduction:

[0078] The large and medium-sized reservoirs above the Yichang section are determined to be Liyuan, Ahai, Jin'anqiao, Longkaikou, Ludila, Guanyinyan, Wudongde, Baihetan, Xiluodu, Xiangjiaba, Lianghekou, Jinping I, Ertan, Zipingpu, Houziyan, Changheba, Dagangshan, Pubugou, Goupitan, Silin, Shatuo, Pengshui, Bikou, Baozhusi, Tingzikou, Caojie, Jiangkou, Wudu and Sanxia. For each reservoir, starting from the lowest water level of the reservoir, the number of grid pixels of SRTM-DEM less than or equal to the water level is counted, and the number of pixels is multiplied by the pixel area (i.e. 30m×30m) to obtain the water surface area corresponding to the water level; the above calculation process is repeated every 1m from the lowest water level to the highest water level of the reservoir to obtain a series of water level-area values; based on the series of water level-area values, a polynomial fitting is used to obtain a water level-area relationship curve.

[0079] On the GEE platform, the modified normalized difference water index (MNDWI) is calculated using the Landsat5, Landsat7 and Landsat8 image series (using median synthesis to reduce the impact of clouds), as follows:

[0080]

[0081] Among them, ρ(Green) is the green band, and ρ(SWIR1) is the shortwave infrared band. According to MNDWI and threshold, the water surface area change process of the above reservoirs is extracted.

[0082] According to the reservoir water level area relationship curve and the reservoir water surface area change process, the water storage capacity change process of each reservoir is calculated as follows:

[0083]

[0084] Among them, RA is the surface area of ​​the reservoir, and RE is the water level of the reservoir. Taking the four reservoirs in the lower reaches of the Jinsha River as an example, the calculation results are shown in the attached Figure 2

[0085] According to the long series of measured flow R at Yichang Station from 1956 to 2023 obs As well as the change process of water storage capacity of the controlling reservoir above Yichang Station, the water balance method is used to consider the impact of water conservancy projects on the runoff series, and the primary reduction runoff series R is obtained. re_1 , calculated as follows:

[0086]

[0087] Where ΔRV t is the total storage variable of the Yichang upstream reservoir group during period t.

[0088] (1.2) Secondary reduction:

[0089] According to the water use data published by the Food and Agriculture Organization of the United Nations (WU R ) as a reference, and the water resources bulletin water use data (WU G ) is interpolated to make up for the missing data, and the calculation is as follows:

[0090]

[0091] in, is the water use data of the water resources bulletin for period t, and i <t<j;R j and R i They represent the ratios of the water use data published in the bulletin to the water use data published by the Food and Agriculture Organization of the United Nations in period j and period i respectively.

[0092] Assuming that the time course of water use is an S-shaped curve, G Forward extension is performed to make up for the missing data, and the calculation is as follows:

[0093]

[0094] About WU G Backward extension is performed to make up for the missing data, and the calculation is as follows:

[0095]

[0096] Among them, α is used to correct the growth rate, and its value range is [0.1, 0.2].

[0097] Assuming that water withdrawal only exists in construction land and cultivated land, the global land cover type product (Climate Change Initiative-Land Cover, CCI-LC) released by ESA every year is used to obtain the corresponding year after time interpolation. Spatial distribution data Since the time range of CCI-LC is from 1992 to 2020, The spatial discretization adopts the CCI-LC of 1992, and the CCI-LC after 2020 The spatial discretization adopts the CCI-LC of 2020. The calculation is as follows:

[0098]

[0099] Among them, A LU Indicates the area of ​​construction land or cultivated land. It can be seen that the water consumption WU in the above period t of Yichang section t The water usage changes above the Yichang section are shown in the attached Figure 3 The spatial distribution of water use above the Yichang section in 2020 is shown in the attached Figure 4 .

[0100] According to the primary reduction runoff series R re1 Based on the water use data of the Yichang section, the water balance is used to further consider the impact of water use on the runoff series, and the secondary reduction runoff series R is obtained. re2 , calculated as follows:

[0101]

[0102] Among them, WU t is the water consumption above the Yichang section in time period t, and ζ is the water consumption rate.

[0103] (2) Constructing a non-consistent drought index:

[0104] In order to construct the non-uniform SRI, the semi-parametric additive formula of the GAMLSS model is used to estimate the parameters of the runoff distribution function of the Yichang station, which is calculated as follows:

[0105]

[0106] Among them, θ k and β k is a vector of length n, θ k is the runoff distribution function parameter of Yichang Station. k (.) is a known link function used to link the distribution function parameters with the explanatory variables. k For the known n×J' k Design matrix, The length is J' k The explanatory variable vector, Z jk For the known n×m kj The design matrix, γ jk is m kj A random variable of dimension X. k η k is a parameter component composed of linear or nonlinear explanatory variables. jk γ jk is the additive component representing the random effect.

[0107] θ k Take the parameters μ and σ in the GAMLSS model. η k There are differences for different GAMLSS models. Specifically, for the measured runoff series, η k Including reservoir index and water withdrawal index; for the single reduction runoff series, η k Includes water withdrawal index.

[0108] In different seasons, for the measured annual runoff series R obs and the once-reduced annual runoff series R re_1 , respectively fit GAMLSS model 1 and GAMLSS model 2, and estimate the distribution function parameter μ of the corresponding annual runoff obs , σ obs , μ re_1 and σ re_1 The fitting method adopts the Rigby and Stasinopoulos (RS) algorithm, and the objective function is to minimize the global deviation. The five distributions that the annual runoff may obey are pre-set as follows: Gumbel (GU), Gamma (GA), Weibull (WEI), Logistic (LO) and Lognormal (LOGNO). Finally, the distribution that minimizes the Akaike information criterion (AIC) is selected, as shown in Table 1.

[0109] Table 1

[0110]

[0111] Taking the spring natural runoff at Yichang Station as an example, the non-uniform distribution obtained by fitting based on GAMLSS model 1 and its comparison with the consistent distribution are shown in the attached figure. Figure 5 Among them, the five curves from top to bottom represent the 95%, 75%, 50%, 25% and 5% quantiles respectively, and the red dots are the measured runoff series. The results show that compared with the consistent distribution, the inconsistent distribution can better fit the measured runoff series; especially after the measured runoff series has a mutation, that is, after 2000, the inconsistent distribution can capture the upward trend of runoff, thereby better fitting the extreme values ​​of the runoff series.

[0112] In different seasons, according to the distribution function parameter μ of annual runoff obs , σ obs , μ re_1 and σ re_1 , calculate the distribution function F of annual runoff respectively obs (x) and F re_1 (x), and the corresponding non-consistent hydrological drought index SRI is obtained obs and SRI re_1 For the secondary reduction annual runoff series, the conventional method is used to calculate its distribution parameter μ re_2 , σ re_2 、F re_2 (x) and SRI re_2 The formula for calculating SRI based on F(x) is as follows:

[0113]

[0114] Among them, C 0 =2.516, C 1 =0.803, C 2 =0.010, d 1 =1.433,d 2 =0.189, d 3 = 0.001, x is the annual runoff. The calculated non-consistent hydrological drought index SRI of Yichang station obs and SRI re_1 and the Hydrological Drought Index (SRI) re_2 See attached Figure 6 According to the calculated SRI, the hydrological drought levels can be divided into: normal (0.5 ≥ SRI ≥ -0.5), mild drought (-1 ≤ SRI < 0.5), moderate drought (-1.5 ≤ SRI < -1), severe drought (-2 ≤ SRI < -1.5) and extreme drought (SRI < -2).

[0115] (3) Calculate the contribution of multiple influencing factors and their interactions:

[0116] For the secondary reduction series R re2The mutation point analysis was carried out, and the analysis period was divided into two periods: before the impact of climate change (1956-1995) and after the impact (1996-2022), and then the SRI was obtained. obs_1 、SRI obs_2 、SRI re1_1 、SRI re1_2 、SRI re2_1 and SRI re2_2 , where SRI re2_1 Nearly natural state, SRI re2_2 Affected by climate change, SRI re1_1 Affected by water withdrawal, SRI re1_2 SRI is affected by water withdrawal, climate change, and the interaction between water withdrawal and climate change. obs_1 Affected by water conservancy projects and water withdrawal, SRI obs_2 Affected by water conservancy projects, water use, climate change, the interaction between water conservancy projects and climate change, and the interaction between water use and climate change, the results are shown in the attached Figure 6 Attached Figure 6 The vertical lines before and after the middle are period 1 and period 2. The above analysis is based on the assumption that there is no interaction between reservoir regulation and water withdrawal.

[0117] SRI obs_1 With SRI re1_1 The absolute value of the difference represents the change in the hydrological drought index ΔSRI caused by the operation of the water conservancy project in period 1. R1 , see formula (11); and SRI obs_2 With SRI re1_2 The absolute value of the difference represents the change in hydrological drought index caused by the water conservancy project in period 2 and the interaction between the water conservancy project and climate change (respectively represented by ΔSRI R2 and ΔSRI RC (expressed), see formula (12). Since the inconsistency of the effects of water conservancy projects has been taken into account by constructing the inconsistency hydrological drought index, it can be considered that ΔSRI R1 =ΔSRI R2 =ΔSRI R In summary, ΔSRI can be calculated R =0.303, ΔSRI RC =0.056.

[0118] |SRI obs_1 -SRI re1_1 |=ΔSRI R1 (11)

[0119] |SRI obs_2 -SRI re1_2 |=ΔSRI R2+ΔSRI RC (12)

[0120] SRI re1_1 With SRI re2_1 The absolute value of the difference represents the change in the hydrological drought index ΔSRI caused by water withdrawal in period 1 W1 , see formula (13); and SRI re1_2 With SRI re2_2 The absolute value of the difference represents the change in the hydrological drought index caused by water withdrawal in period 2 and the interaction between water withdrawal and climate change (respectively denoted by ΔSRI W2 and ΔSRI WC (expressed as ΔSRI), see formula (14). Since the inconsistency of water withdrawal has been taken into account by constructing the inconsistency hydrological drought index, it can be considered that ΔSRI W1 =

[0121] ΔSRI W2 =ΔSRI W In summary, ΔSRI can be calculated W =0.256, ΔSRI WC =0.020.

[0122] |SRI re1_1 -SRI re2_1 |=ΔSRI W1 (13)

[0123] |SRI re1_2 -SRI re2_2 |=ΔSRI W2 +ΔSRI WC (14)

[0124] Through the atmospheric circulation factor a iTo describe the impact of climate change, the commonly used Western Pacific Subtropical high intensity index (WPSHII), South China Sea subtropical high intensity index (SCSSHII), North Pacific pattern (NP), North Atlantic oscillation index (NAO), El Nino-Southern Oscillation index ENSO, total Sunspot number index (TSNI) and Southern Oscillation index (SOI) are selected as atmospheric circulation factors. Taking a relatively simple linear regression as an example, SRI is constructed. re2_1 About atmospheric circulation factor a i The expression of is shown in formula (15):

[0125]

[0126] Referring to the elasticity coefficient method, the change in hydrological drought index ΔSRI caused by climate change is calculated C , see formula (16):

[0127]

[0128] Among them, A is the set of atmospheric circulation factors, Δa i a in period 2 i Mean and a in period 1 i The difference between the means is used to calculate ΔSRI C =0.249.

[0129] The contributions of water conservancy project operation, water withdrawal, climate change, the interaction between water conservancy projects and climate change, and the interaction between water withdrawal and climate change to hydrological drought are calculated as follows:

[0130]

[0131] Calculated: α R =34.3%, α W =30.0%, α C =28.2%, α RC =6.3%, α WC =2.3%, the results are summarized in Table 2.

[0132] Table 2

[0133]

[0134] In summary, the present invention solves the problem of hydrological drought attribution under the combined influence of multiple factors and their interactions by fusing multi-source data to restore the measured runoff one by one and constructing an inconsistent hydrological drought index, which helps to deeply understand the evolution mechanism of hydrological drought under inconsistent conditions, realize the refined and dynamic management of water resources, and prevent and alleviate drought events.

Claims

1. A non-uniform hydrological drought attribution method based on successive runoff reduction, characterized in that: include: S1. Fusion of multi-source data to restore runoff one by one, i.e., using multi-source data such as topography, land use, remote sensing images, global-scale water use and water resources bulletins and fusion algorithms to restore the measured runoff series one by one; S2. Construct a non-consistent drought index, that is, use the measured runoff series and the first restored runoff series to calculate the corresponding non-consistent hydrological drought index, and use the second restored runoff series to calculate the hydrological drought index.

2. The non-uniform hydrological drought attribution method based on successive runoff reduction according to claim 1, characterized in that: In S1, fusing multi-source data to restore runoff successively includes: According to the measured runoff series R obs , the reservoir water storage change data obtained by remote sensing inversion, using water balance, considering the impact of water conservancy projects on the runoff series, and obtaining the primary reduction runoff series R re1 , calculated as: Where ΔRV t is the storage variable of the upstream reservoir group in the section within the period t; According to the primary reduction runoff series R re1 , the water withdrawal data obtained by multi-source fusion, based on the primary reduction, further consider the impact of water withdrawal on the runoff series using water balance, and obtain the secondary reduction runoff series R re2 , calculated as: Among them, WU t is the water consumption of the upstream section in time period t; ζ is the water consumption rate.

3. The non-uniform hydrological drought attribution method based on successive runoff reduction according to claim 2 is characterized in that: The method for obtaining the reservoir water storage change data obtained by remote sensing inversion is: S101. Construct a water level-area relationship curve based on the digital elevation model (DEM) data: starting from the lowest water level of the reservoir, count the number of raster pixels in the digital elevation model (DEM) that are less than or equal to the water level, and multiply the number of pixels by the pixel area to obtain the water surface area corresponding to the water level; repeat the calculation at intervals of 1 meter from the lowest water level to the highest water level of the reservoir to obtain a series of water level-area values; and use polynomial fitting to obtain a water level-area relationship curve based on the series of water level-area values. S102. Extract the change process of reservoir water surface area using remote sensing images: On the Google Earth Engine GEE platform, use the image series and median synthesis to reduce the impact of clouds, and calculate the improved normalized water index MNDWI as follows: Among them, ρ(GREEN) is the green band, ρ(SWIR1) is the shortwave infrared band; According to the normalized water body index MNDWI and threshold, the change process of reservoir water surface area is extracted. S103. According to the reservoir water level area relationship curve and the reservoir water surface area change process, the reservoir water storage capacity change process is calculated as follows: Among them, RA is the surface area of ​​the reservoir and RE is the water level of the reservoir.

4. The non-uniform hydrological drought attribution method based on successive runoff reduction according to claim 2 is characterized in that: The method for obtaining the water use data obtained by multi-source fusion is: S104. Publish water use data on a global scale R As a reference, the water resources bulletin water use data WU G The calculation for interpolation to make up for missing data is: in, is the water use data of the water resources bulletin for period t, and i <t<j;R j and R i They represent the ratios of the public water consumption data and the reference water consumption data in time period j and time period i respectively. S105, assuming that the time variation of water use is an S-shaped curve, the water use data WU G Forward extension is performed to make up for the missing data, and the calculation is: Water Resources Bulletin Water Use Data WU G Backward extension is performed to make up for the missing data, and the calculation is: Among them, α is the corrected growth rate, and its value range is [0.1, 0.2]; K is the corrected water use change rate. S106, assuming that water is only used for construction land and cultivated land, using the land use data of each period, obtain the water use data WU of the water resources bulletin after being processed by S104 and S105 G Spatial distribution data Calculated as: Among them, A LU Indicates the area of ​​construction land or cultivated land; Spatial distribution data of water use based on water resources bulletin Know the water consumption WU in any section within the above period t t .

5. The non-uniform hydrological drought attribution method based on successive runoff reduction according to claim 1, characterized in that: In S2, the calculation of the corresponding inconsistent hydrological drought index is specifically as follows: S201, according to the measured runoff series R obs , taking reservoir index RI and water intake index WUI as explanatory variables, the first model is constructed to calculate the corresponding measured runoff series R obs The first distribution function parameter μ obs and the second distribution function parameter σ obs : Among them, g(.) represents the link function; k is the coefficient vector; X is the explanatory variable; According to the first distribution function parameter μ obs and the second distribution function parameter σ obs , calculate the corresponding measured runoff series R obs The SRI is a non-uniform hydrological drought index. obs . S202, according to the primary reduction runoff series R re1 , taking the water withdrawal index WUI as the explanatory variable, the second model is constructed to calculate the corresponding first-time restored runoff series R re1 The third distribution function parameter μ re1 and the fourth distribution function σ re1 : According to the third distribution function parameter μ re1 and the fourth distribution function parameter σ re1 , calculate the corresponding primary reduction runoff series R re1 The SRI is a non-uniform hydrological drought index. re1 . S203, according to the secondary reduction runoff series R re2 , calculate the corresponding secondary reduction runoff series R re2 The Hydrological Drought Index (SRI) re2 .

6. The non-uniform hydrological drought attribution method based on successive runoff reduction according to claim 1, characterized in that: It also includes calculating the contribution of multiple influencing factors and their interactions, specifically: S301, secondary reduction runoff series R re2 Perform mutation point analysis and divide the analysis period into two periods before and after the impact of climate change, namely the first period and the second period, to obtain SRI obs_1 、SRI obs_2 、SRI re1_1 、SRI re1_2 、SRI re2_1 and SRI re2_2 ; Among them, SRI obs_1 and SRI obs_2 Represents the SRI of the first period and the second period respectively obs , SRI obs_1 Affected by water conservancy projects and water use, SRI obs_2 Affected by water conservancy projects, water withdrawal, climate change, the interaction between water conservancy projects and climate change, and the interaction between water withdrawal and climate change; SRI re1_1 and SRI re1_2 Represents the SRI of the first period and the second period respectively re1 , SRI re1_1 Affected by water withdrawal, SRI re1_2 Affected by water withdrawal, climate change, and the interaction between water withdrawal and climate change; SRI re2_1 and SRI re2_2 Represents the SRI of the first period and the second period respectively re2 , SRI re2_1 Nearly natural state, SRI re2_2 Affected by climate change. S302. Calculate the contribution of water conservancy project operation to hydrological drought α R , Contribution of water withdrawal to hydrological drought α W 、Contribution of climate change to hydrological drought α C , the contribution of the interaction between water conservancy projects and climate change to hydrological drought α RC And the contribution of the interaction between water withdrawal and climate change to hydrological drought α WC : Among them, ΔSRI R , ΔSRI W , ΔSRI C , ΔSRI RC and ΔSRI WC They respectively represent the changes in the inconsistent hydrological drought index (SRI) caused by water conservancy project scheduling and operation, water withdrawal, climate change, the interaction between water conservancy projects and climate change, and the interaction between water withdrawal and climate change. S303. Calculate the variation of the non-consistent hydrological drought index SRI, including ΔSRI R , ΔSRI W , ΔSRI C , ΔSRI RC and ΔSRI WC : Among them, ΔSRI R1 and ΔSRI R2 represents the change of hydrological drought index caused by water conservancy projects in the first period and the second period, respectively, and ΔSRI R1 =ΔSRI R2 ; ΔSRI W1 and ΔSRI W2 represents the change of hydrological drought index caused by water withdrawal in the first period and the second period, respectively, and ΔSRI W1 =ΔSRI W2 ; f is SRI re2_1 About atmospheric circulation factor a i The expression f is constructed by regression equation, proxy model or machine learning. A is the set of atmospheric circulation factors, Δa i is the atmospheric circulation factor a affected by climate change i The average value and the atmospheric circulation factor a before the impact of climate change i The difference in means.

Images