Runoff change attribution analysis method based on station partition rainfall treatment and multi-method cooperation

Abstract

The invention discloses a runoff change attribution analysis method based on station partition precipitation processing and multi-method cooperation, comprising the following steps: reading and preprocessing basin hydrological data, and calculating station partition precipitation; detecting a runoff sequence mutation point by adopting a cumulative distance-leveling method, and determining a reference period and a change period; carrying out Mann-Kendall trend test on the basis of the annual scale data, and calculating a Sen slope quantitative change trend; and implementing three attribution analysis methods in parallel: performing weighted fusion on attribution results of the three methods, calculating contribution rates and dominant factors of climate changes and human activities, and visually displaying runoff process lines, mutation points, interannual trends and multi-method attribution result comparison. According to the method, the contribution of rainfall and evapotranspiration change is independently quantified by improving a Budyko elastic coefficient method, three attribution analysis methods are further integrated, and quantification of factors such as climate change and human activity contribution and the like is efficiently realized by combining automatic detection and trend analysis of mutation points.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of hydrology and water resources simulation and management, and particularly relates to a runoff change attribution analysis method based on site partitioned precipitation processing and multi-method cooperation. BACKGROUND

[0002] Runoff change attribution analysis is a core problem of water resources management, and is crucial for flood control, drought resistance, ecological protection and water conservancy planning. Existing methods mainly include improved Budyko hypothesis method, double cumulative curve method and hydrological simulation method. However, single method has significant limitations: the traditional Budyko hypothesis method depends on the accuracy of parameter ω, the double cumulative curve method is sensitive to mutation points, and the hydrological simulation method is limited by the uncertainty of parameter calibration. In addition, the existing technology lacks a multi-method cooperation framework, and it is difficult to comprehensively evaluate the interactive effects of climate change and human activities. Especially in the basins with frequent human activities (such as urbanization and agricultural irrigation areas), single method is easy to cause error in contribution rate evaluation, and there is an urgent need for a comprehensive and reliable runoff change attribution analysis method suitable for complex hydrological conditions in the basin. SUMMARY

[0003] The application proposes a runoff change attribution analysis method based on site partitioned precipitation processing and multi-method cooperation. The method independently quantifies the contribution of precipitation and evapotranspiration change by improving the Budyko elasticity coefficient method, avoids mixing climate factors and human activity impacts, further integrates the improved Budyko hypothesis method, double cumulative curve method and hydrological simulation method, and combines automatic detection of mutation points and trend analysis to efficiently realize quantification of climate change and human activity contribution elements.

[0004] Technical scheme: The runoff change attribution analysis method based on site partitioned precipitation processing and multi-method cooperation proposed by the application comprises the following steps:

[0005] S1, reading and preprocessing the basin hydrological data, including data integrity check, missing value filling and runoff depth conversion, and calculating the site partitioned precipitation amount;

[0006] S2, detecting the mutation point of the runoff sequence by using the cumulative anomaly method, and determining the reference period and the change period;

[0007] S3, performing Mann-Kendall trend test based on annual scale data, and calculating Sen slope to quantify the change trend;

[0008] S4, implementing three attribution analysis methods in parallel:

[0009] The improved Budyko hypothesis method calculates the climate elasticity coefficient;

[0010] The double cumulative curve method establishes the statistical relationship between precipitation and runoff;

[0011] Hydrological simulation method to calibrate runoff parameters;

[0012] S5, the attribution results of the three methods are weighted and fused, the contribution rates of climate change and human activities and the dominant factors are calculated, and the runoff process line, mutation point, annual trend and multi-method attribution results are visualized and displayed.

[0013] Further, the runoff depth conversion formula in S1 is:

[0014]

[0015] Wherein, R is the runoff depth, mm / d, Q is the daily average flow, m 3 / s, A is the area of the basin, km 2 .

[0016] Further, the mutation point detection in S2 is specifically:

[0017] Calculate the runoff cumulative anomaly sequence:

[0018]

[0019] Determine the mutation point:

[0020] t change = argmax |C k |

[0021] Wherein, R i is the daily runoff depth, is the sequence mean.

[0022] Further, the annual scale runoff trend analysis in S3 is specifically:

[0023] By constructing the statistical quantity S of rank order sequence to evaluate the trend direction, for a time sequence with length n, calculate the sum of the difference sign function of all subsequent values and previous values, that is:

[0024]

[0025] The corrected variance Var(S) is:

[0026]

[0027] In the formula: m is the number of knots; t k represents the length of the kth knot, that is, the number of repeated values;

[0028] The calculation formula of the standardized statistical quantity Z is:

[0029]

[0030] Based on the Z value to determine the trend direction and significance level, p < 0.05 is significant.

[0031] Further, the improved Budyko hypothesis method in S4 specifically includes:

[0032] The calculation of potential evapotranspiration PET uses the Hargreaves-Samani formula to estimate:

[0033]

[0034] In the formula: R a is daily extraterrestrial solar radiation, MJ / m 2 / day, calculated by latitude and day number; T max , T min , and T mean are daily maximum, minimum, and average air temperatures, respectively;

[0035] Based on the baseline period data, the optimal ω value is calibrated using the nonlinear least squares method, and the objective function is defined as:

[0036]

[0037] In the formula: Q total is total runoff; Q base is base flow; λ = 0.3 is the base flow weight coefficient; the function g represents the relationship between base flow and climate factors, and the partial derivative method is used to calculate the elasticity coefficient of precipitation ε P and potential evapotranspiration ε PET :

[0038]

[0039] Further, the double mass curve method in S4 is specifically:

[0040] Establish the linear relationship between precipitation and runoff in the baseline period:

[0041] ∑R = β0 + β1∑P

[0042] Calculate the cumulative runoff under natural conditions:

[0043] ∑R natural = β0 + β1∑P total

[0044] Climate change contribution rate:

[0045]

[0046] Where R base is the cumulative runoff in the baseline period, and R obs is the measured cumulative runoff.

[0047] Further, the hydrological simulation method in S4 specifically comprises:

[0048] Constructing a water tank model:

[0049]

[0050] Optimizing the parameter vector:

[0051] θ=[α,C,d] T

[0052] Calibrating parameters by a Nash efficiency coefficient objective function:

[0053]

[0054] Further, the weighted fusion formula in S5 is:

[0055]

[0056] Where ω1, ω2, ω3 are weight coefficients, and the dominant factor determination rule is:

[0057]

[0058] Beneficial effects: Compared with the prior art, the present application has the following significant advantages:

[0059] Firstly, multiple methods are used in cooperation, and the improved Budyko hypothesis method, double cumulative curve method and hydrological simulation method are integrated, the uncertainty of a single method is reduced through weighted fusion, and the contribution rate evaluation accuracy is improved;

[0060] Secondly, full-process automation is realized, full-process automation from data preprocessing, mutation point detection to attribution analysis is realized, subjective intervention is avoided, visualization driving is realized, multi-dimensional icons are used to intuitively display runoff change characteristics and attribution results, and water resource management decision-making is supported;

[0061] Finally, it is widely applicable and suitable for complex basin scenarios with climate change and human activities, and has been verified through a case of the Fenghe River Basin. BRIEF DESCRIPTION OF DRAWINGS

[0062] Figure 1 It is the overall flowchart of the method of the present application;

[0063] Figure 2 It is a runoff hydrograph and mutation point position diagram of Dayu Station and Qinduzhen Station in the Fenghe River Basin;

[0064] Figure 3 It is a graph of annual runoff depth variation trend;

[0065] Figure 4 It is a pie chart of site attribution results;

[0066] Figure 5 A bar chart comparing the climate contribution rates of the three methods;

[0067] Figure 6 This is a comparison chart of the overall attribution results. Detailed Implementation

[0068] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0069] The effectiveness of this method is verified using the Fenghe River Basin as an example, including the following steps:

[0070] S1: Construct a watershed hydrological analysis framework. Verify the site-specific precipitation treatment and multi-method synergistic runoff change attribution analysis method of this invention in the Fenghe River basin, such as... Figure 1 As shown. The watershed includes Dayu Station (area 54 km²). 2 ) and Qinduzhen Station (area 566km) 2 Two hydrological zones were defined. First, daily hydrological data from 2001 to 2020 were acquired, primarily including daily runoff and precipitation data from Dayu Station and Qindu Town Station. Then, runoff volume was converted to runoff depth.

[0071]

[0072] Where R is the runoff depth (mm / d) and Q is the average daily flow rate (m³ / d). 3 / s), A is the drainage area (km²) 2 ).

[0073] S2: Hydrological sequence mutation point detection.

[0074] Automatic identification of abrupt changes in runoff sequences using the cumulative anomaly method:

[0075]

[0076] Mutation point: t change =argmax|C k |

[0077] The results of the examples are as follows Figure 2 As shown, the year of abrupt change in runoff at Dayu Station was 2005, and the year of abrupt change in runoff at Qinduzhen Station was 2009. Then, based on the abrupt change points, the study period was divided into the baseline period and the change period.

[0078] S3: Trend Quantification and Annual Scale Analysis.

[0079] This method assesses the trend direction by constructing a statistic S for the rank sequence: for a time series of length n, the sum of the sign functions of the differences between all subsequent values ​​and preceding values ​​is calculated, i.e.:

[0080]

[0081] Since there are often repeated values (called "knots") in hydrological data, the variance calculation needs to be corrected to improve the accuracy of the test. The corrected variance Var(S) is:

[0082]

[0083] In the formula: m is the number of knots; t k represents the length of the kth knot (i.e. the number of repeated values).

[0084] The calculation formula of the standardized statistic Z is:

[0085]

[0086] Based on the Z value to determine the trend direction (up / down) and significance level (p<0.05 is significant);

[0087] Calculate the Sen slope to estimate the rate of change.

[0088] Example results (Appendix Figure 3 ):

[0089] Annual runoff depth at Dayu Station: showing a non-significant upward trend (Sen slope 2.6853 mm / year, p=0.8337);

[0090] Annual runoff depth at Qindu Town Station: showing a significant upward trend (Sen slope 11.4915 mm / year, p=0.0689).

[0091] S4: Multi-method collaborative attribution analysis.

[0092] Quantify the impact of climate change and human activities on the changes in hydrological data in the basin through the integration of the three methods.

[0093] Figure 4 Pie chart for station attribution results, showing the contribution proportion of climate change and human activities at Dayu Station and Qindu Town Station; Figure 5 Comparison of climate contribution rates of the three methods: improved Budyko hypothesis method, double cumulative curve method, and hydrological simulation method, showing the differences in climate contribution rates at the two stations.

[0094] Improved Budyko Hypothesis Method:

[0095] Based on daily observation data, first calculate the average precipitation (P) and runoff (Q) of the basin. Missing values are repaired using linear interpolation, and daily data is aggregated into annual values to improve analysis stability. The potential evaporation (PET) is estimated using the Hargreaves-Samani formula:

[0096]

[0097] where R a is the daily extraterrestrial solar radiation (MJ / m 2 / day), calculated from latitude and day number; T max , T min , and T mean are daily maximum, minimum, and mean air temperature, respectively.

[0098] The optimal ω value was calibrated by nonlinear least square method based on the baseline data. The objective function was defined as:

[0099]

[0100] where Q total is the total runoff; Q base is the baseflow; λ = 0.3 is the baseflow weighting coefficient; and function g represents the relationship between baseflow and climate factors.

[0101] To avoid local optimal solution, multiple initial values (ω0= 1.5, 2.0, 2.5, 3.0, 3.5) were used for iterative calculation, and the solution with the minimum residual sum of squares was selected as the global optimal parameter. The elasticity coefficient reflects the sensitivity of runoff to climate factors. The partial derivative method was used to calculate the elasticity coefficients of precipitation (ε P ) and potential evapotranspiration (ε PET ):

[0102]

[0103]

[0104] The climate contribution is determined by the changes in precipitation and potential evapotranspiration:

[0105]

[0106] The human activity contribution rate is calculated by the residual method:

[0107] ΔQ human = ΔQ - ΔQ climate

[0108] Double mass curve method:

[0109] Establish the baseline cumulative precipitation-runoff relationship:

[0110] ∑R = β0+ β1∑P

[0111] Predict the cumulative natural runoff R natural :

[0112] ∑Rnatural = β0+ β1∑P total

[0113] Calculate the contribution rate of climate change:

[0114]

[0115] Where R base is the cumulative runoff of the reference period, R obs is the measured cumulative runoff. Hydrological simulation method:

[0116] Build a water tank model:

[0117]

[0118] Optimize the parameter vector:

[0119] θ = [α, C, d] T

[0120] Calibrate the parameters by the Nash efficiency coefficient objective function:

[0121]

[0122] S5: result fusion and verification.

[0123] Weighted fusion contribution rate:

[0124]

[0125] In this example, take Determine the dominant factor:

[0126]

[0127] Finally, visualize the runoff hydrograph, abrupt point, interannual trend and multi-method attribution results of Dayu Station and Qinduzhen Station.

[0128] Figure 6 The comprehensive attribution result comparison chart is obtained by comparing the comprehensive contribution rates of climate change and human activities of the two stations through the bar chart.

[0129] The above implementation shows that the improved Budyko hypothesis method, double cumulative curve method and hydrological simulation method are integrated to effectively quantify the contribution rates of climate change and human activities to runoff. Compared with a single method, the comprehensive analysis method significantly improves the reliability of the attribution results, and is suitable for quantitatively evaluating the contribution of climate change and high-intensity human activities to the runoff change of a complex hydrological condition basin.

Claims

1. A method for runoff change attribution analysis based on site partition precipitation processing and multi-method coordination, characterized in that, Comprising the following steps: S1, read the hydrological data of the catchment and pre-process, including data integrity check, missing value filling and runoff depth conversion, calculate the site partition precipitation; S2, adopt the cumulative anomaly method to detect the mutation point of runoff sequence, determine the reference period and change period; S3, based on annual scale data, carry out Mann-Kendall trend test, calculate Sen slope to quantify the change trend; S4, implement three kinds of attribution analysis methods in parallel: Improved Budyko hypothesis method to calculate climate elasticity coefficient; Double cumulative curve method to establish the statistical relationship between precipitation and runoff; Hydrological simulation method to calibrate runoff parameters; S5, the attribution results of the three methods are weighted and fused to calculate the contribution rate of climate change and human activities and the dominant factor, and the runoff process line, mutation point, interannual trend and multi-method attribution result comparison are visualized.

2. The method of claim 1, wherein the method is based on the partitioning of the site into subareas, and the analysis of the changes in the runoff is performed in coordination with the use of multiple methods. The runoff depth conversion formula in S1 is: where R is the runoff depth, mm / d, and Q is the daily average flow, m 3 / s, and A is the drainage area, km 2 .

3. The method of claim 1, wherein the method is characterized by, The mutation point detection in S2 is: Calculate the runoff cumulative anomaly sequence: Determine the mutation point: t change = argmax |C k | where R i is the daily runoff depth, is the sequence mean.

4. The method of claim 1, wherein the method is based on a site- specific precipitation processing and multi-method synergy for runoff change attribution analysis. The annual scale runoff trend analysis in S3 is: By constructing the statistical quantity S of rank sequence to evaluate the trend direction, for a time series with length n, calculate the sum of the difference sign function of all subsequent values and previous values, that is: The corrected variance Var(S) is: wherein: m is the number of junctions; t k represents the length of the kth junction, i.e. the number of repeating values; The calculation formula of standardized statistics Z is: Determine the trend direction and significance level based on Z value, p<0.05 is significant.

5. The method of claim 4, wherein the method is based on the partitioning of the site into subareas, and the analysis of the runoff change is performed for each subarea. The improved Budyko hypothesis method in S4 includes: The calculation of potential evaporation PET uses Hargreaves-Samani formula to estimate: where: R a is the extraterrestrial solar radiation, MJ / m 2 / day, calculated from latitude and day number; T max ,T min ,T mean are the daily maximum, minimum and average air temperature, respectively; And based on the reference period data, the optimal ω value is calibrated by nonlinear least squares method, and the objective function is defined as: where Q total is the total runoff; Q base is the baseflow; λ = 0.3 is the baseflow weighting coefficient; the function g characterizes the baseflow dependence on the climatic factors, calculated by the partial derivative method with respect to the precipitation ε P and the potential evapotranspiration ε PET elasticity coefficient:

6. The method of runoff change attribution analysis based on site-division precipitation processing and multi-method synergy according to claim 5, characterized in that, The double cumulative curve method in S4 is: Establish the linear relationship between precipitation and runoff in the reference period: ∑R=β0+β1∑P Calculate the runoff cumulative amount under natural conditions: ∑R natural = β0+ β1∑P total Climate change contribution rate: where R base is the cumulative runoff for the reference period, R obs is the measured cumulative runoff.

7. The method of claim 6, wherein the method is characterized by, The hydrological simulation method in S4 includes: Build a water tank model: Optimize the parameter vector: θ = [a, C, d] T Calibrate the parameters by Nash efficiency coefficient objective function:

8. The method of runoff change attribution analysis based on site-division precipitation processing and multi-method synergy according to claim 7, characterized in that, The weighted fusion formula in S5 is: Where ω1, ω2, ω3 are weight coefficients, and the dominant factor determination rule is:

Images