A method to improve the accuracy of regional groundwater storage based on groundwater weighted fusion model

Abstract

The present invention discloses a method and system for improving the accuracy of regional groundwater reserves based on a groundwater weighted fusion model, the method comprising: randomly selecting three independent groundwater reserve anomaly data sets; performing covariance, numerical change sensitivity, and random error variance calculations based on the three independently selected groundwater reserve anomaly data sets, and constructing a groundwater weighted fusion model; based on the constructed groundwater weighted fusion model, performing groundwater reserve anomaly estimation, obtaining groundwater reserve anomaly estimation results, and outputting them. The present invention uses the square of the correlation coefficient and the error variance calculated by the ETC method with the unknown measured value to construct a weight factor, and assigns weights to GWSA based on GRACE and multiple hydrological model data, thereby fusing these GWSA results, fully retaining the advantages of each hydrological model data, and improving the accuracy of GRACE regional groundwater reserve estimation.

Description

Technical Field

[0001] The present invention belongs to the cross-technical fields of satellite gravity, hydrology, etc., and in particular relates to a method for improving the accuracy of regional groundwater reserves based on a groundwater weighted fusion model. Background Art

[0002] Groundwater is an important component of terrestrial water storage (TWS) and plays an important role in the global water cycle. Due to warming temperatures, population growth and expansion of agricultural land, the demand for water has increased, and the pressure on water resources has become increasingly greater, posing a threat to water security. More than 38% of the world's population lives in arid or semi-arid areas, and about 40% of global groundwater extraction occurs in these areas. In addition, groundwater in these areas only receives 2% of recharge, so groundwater is being rapidly depleted in these places. However, traditional groundwater storage observations are mainly monitored using groundwater level monitoring wells. This method can provide groundwater level measurements with high temporal and spatial resolution; however, in many areas, water level monitoring is limited, such as uneven distribution of monitoring wells, and the establishment and maintenance of observation stations is expensive and time-consuming, resulting in short and discontinuous observation records. Therefore, the measured data are not suitable for large-scale climate studies. Summary of the invention

[0003] The technology of the present invention solves the problem: overcomes the shortcomings of the prior art, provides a method for improving the accuracy of regional groundwater reserves based on a groundwater weighted fusion model, and proposes a new groundwater weighted fusion model. The new groundwater weighted fusion model uses the square of the correlation coefficient and the error variance between the unknown measured value calculated by the ETC method to construct a weight factor, and gives weights to the GWSA based on GRACE and multiple hydrological model data, thereby fusing these GWSA results, fully retaining the advantages of each hydrological model data, and improving the accuracy of GRACE regional groundwater reserves estimation.

[0004] In order to solve the above technical problems, the present invention discloses a method for improving the accuracy of regional groundwater reserves based on a groundwater weighted fusion model, comprising:

[0005] Three independent groundwater storage anomaly datasets were randomly selected;

[0006] According to three randomly selected independent groundwater storage anomaly data sets, the covariance, numerical change sensitivity, and random error variance are solved to obtain the correlation coefficient and random error variance of each groundwater storage anomaly data set;

[0007] According to the correlation coefficient and random error variance of each groundwater storage anomaly data set, a groundwater weighted fusion model is constructed.

[0008] Based on the constructed groundwater weighted fusion model, the groundwater reserve anomaly estimation is carried out, and the groundwater reserve anomaly estimation result is obtained and output.

[0009] In the above-mentioned method for improving the accuracy of regional groundwater reserves based on the groundwater weighted fusion model, three independent groundwater reserve anomaly data sets are randomly selected: the first groundwater reserve anomaly data set S1, the second groundwater reserve anomaly data set S2, and the second groundwater reserve anomaly data set S3 obtained by combining the gravity recovery and climate experiment satellite data with the GLDAS hydrological model, the ERA5 reanalysis data set, and the WGHM hydrological model.

[0010] In the above-mentioned method for improving the accuracy of regional groundwater storage based on the groundwater weighted fusion model, the covariance, numerical change sensitivity, and random error variance are solved according to the three independent groundwater storage anomaly data sets obtained, and the correlation coefficient and random error variance of each groundwater storage anomaly data set are obtained, including:

[0011] According to the three independent groundwater storage anomaly data sets S i , perform covariance calculation, and obtain the covariance cov(S i ,S j ), where i∈[1,2,3], j∈[1,2,3];

[0012] According to cov(S i ,S j ), calculate the sensitivity of numerical changes and the variance of random errors, and obtain S i The sensitivity of the numerical change K i and S i The random error variance

[0013] According to K i and Solve to get S i The correlation coefficient ρ i,Θ .

[0014] In the above method for improving the accuracy of regional groundwater storage based on groundwater weighted fusion model, cov(S i ,S j ) is calculated as follows:

[0015] cov(S i ,S j )=E(S i S j )-E(S i )E(S j )···(1)

[0016] Among them, E(S i S j ) indicates S i S j The mathematical expectation, E(S i ) indicates S i The mathematical expectation, E(S j ) indicates S j The mathematical expectation of .

[0017] In the above method of improving the accuracy of regional groundwater storage based on groundwater weighted fusion model, S i The sensitivity of the numerical change K i The solution process is as follows:

[0018] cov(S i ,S j ) is denoted as C ij ;

[0019] According to C ij , S is obtained by solving the following formula (2): i The sensitivity of the numerical change K i :

[0020]

[0021] In the above method of improving the accuracy of regional groundwater storage based on groundwater weighted fusion model, S i The random error variance The solution formula is as follows:

[0022]

[0023] In the above method of improving the accuracy of regional groundwater storage based on groundwater weighted fusion model, S i The correlation coefficient ρ i,Θ The solution formula is as follows:

[0024]

[0025] In the above-mentioned method for improving the accuracy of regional groundwater storage based on the groundwater weighted fusion model, a groundwater weighted fusion model is constructed according to the correlation coefficient and random error variance of each groundwater storage anomaly data set, including:

[0026] According to i,Θ and Determine the weight W of the groundwater weighted fusion model i ;

[0027] According to the time series M of each independent groundwater storage anomaly data set at the same location i and determine Wi , the following groundwater weighted fusion model Y is constructed model :

[0028] Y model =W1M1+W2M2+W3M3···(5).

[0029] In the above method for improving the accuracy of regional groundwater storage based on the groundwater weighted fusion model, the weight W of the groundwater weighted fusion model is i The solution formula is as follows:

[0030]

[0031] Correspondingly, the present invention also discloses a system for improving the accuracy of regional groundwater reserves based on a groundwater weighted fusion model, comprising:

[0032] The dataset selection module is used to randomly select three independent groundwater storage anomaly datasets;

[0033] A solution module is used to solve the covariance, value change sensitivity, and random error variance according to three randomly selected independent groundwater storage anomaly data sets, and obtain the correlation coefficient and random error variance of each groundwater storage anomaly data set;

[0034] A model building module is used to build a groundwater weighted fusion model based on the correlation coefficient and random error variance of each groundwater storage anomaly data set;

[0035] The anomaly estimation module is used to estimate the groundwater reserve anomaly based on the constructed groundwater weighted fusion model, obtain the groundwater reserve anomaly estimation result, and output it.

[0036] The present invention has the following advantages:

[0037] The present invention discloses a method for improving the accuracy of regional groundwater reserves based on a groundwater weighted fusion model, provides a method for improving the accuracy of regional groundwater reserves based on a groundwater weighted fusion model, and proposes a new groundwater weighted fusion model. The new groundwater weighted fusion model constructs weight factors using the square of the correlation coefficient and the error variance between the unknown measured values ​​calculated by the ETC method, and gives weights to GWSA based on GRACE and multiple hydrological model data, thereby fusing these GWSA results, fully retaining the advantages of each hydrological model data, and improving the accuracy of GRACE regional groundwater reserves estimation. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Figure 1 is a flowchart of the steps of a method for improving the accuracy of regional groundwater reserves based on a groundwater weighted fusion model in an embodiment of the present invention;

[0039] Figure 2 It is a location distribution map of the geographical location, digital elevation and groundwater level monitoring points of the Hexi Corridor in an embodiment of the present invention;

[0040] Figure 3 It is a schematic diagram of time series comparison between GWSA derived from GRACE-GLDAS, GRACE-ERA5, and GRACE-WGHM and GWSA based on GWFM in an embodiment of the present invention;

[0041] Figure 4 4(a) is a schematic diagram of time series comparison between GWSA from GRACE-GLDAS, GRACE-ERA5, GRACE-WGHM and GWSA based on GWFM after setting a lag time in an embodiment of the present invention; wherein, 4(a) is S1; 4(b) is S2; 4(c) is S3; 4(d) is S4; 4(e) is S5; 4(f) is the average of S1 to S5;

[0042] Figure 5 5 is a schematic diagram of the comparison of GWSA time series from GRACE-GLDAS, GRACE-ERA5, GRACE-WGHM and GWSA time series based on GWFM at different time scales in an embodiment of the present invention; wherein, 5(a) is the monthly scale; 5(b) is the annual scale; 5(c) is the seasonal scale;

[0043] Figure 6 6(a) is a vector diagram of the annual change of GWSA in the Hexi Corridor region derived from GRACE-GLDAS, GRACE-ERA5, GRACE-WGHM and GWFM in an embodiment of the present invention; wherein, 6(a) is the annual scale; 6(b) is the semi-annual scale;

[0044] Figure 7 7(a) GRACE-GLDAS; 7(b) GRACE-ERA5; 7(c) GRACE-WGHM; 7(d) GWFM;

[0045] Figure 8 8(a) is a schematic diagram of the trend spatial distribution comparison of the average results of GWSA based on GWFM and GRACE-GLDAS, GRACE-ERA5 and GRACE-WGHM in the Hexi Corridor region in an embodiment of the present invention; wherein, 8(a) is GWFM; 8(b) is Average;

[0046] Fig. 9Schematic diagram of cross wavelet transform between GWSA and precipitation (a), evapotranspiration (b) and temperature (c) in the Hexi Corridor region in an embodiment of the present invention;

[0047] Fig.10 It is a time series comparison schematic diagram of a P-ET and a GWFM-based GWSA in an embodiment of the present invention;

[0048] Fig.11 It is a schematic diagram of the time series of groundwater withdrawal in the Hexi Corridor and its sub-regions in an embodiment of the present invention. DETAILED DESCRIPTION

[0049] In order to make the objectives, technical solutions and advantages of the present invention more clear, the embodiments disclosed in the present invention will be further described in detail below with reference to the accompanying drawings.

[0050] like Figure 1 In this embodiment, the method for improving the accuracy of regional groundwater reserves based on the groundwater weighted fusion model includes:

[0051] Step 101, randomly select three independent groundwater storage anomaly data sets.

[0052] In this embodiment, the data simulated based on the hydrological model usually requires the real value of the target variable for calibration and verification. Triple combination (TC) is a statistical method for estimating the variance of random errors of three independent data sets. The use of the TC method requires four prerequisites: ① The three data sets are linearly correlated with the true value; ② The error is stable, and the mean and variance do not change over time; ③ The random errors between the three data are independent of each other; ④ The random errors are independent of the true value. Based on the above principles, this embodiment selects the gravity recovery and climate experiment satellite data to combine the GLDAS hydrological model, the ERA5 reanalysis data set, and the WGHM hydrological model to obtain the first groundwater reserve anomaly data set S1, the second groundwater reserve anomaly data set S2, and the second groundwater reserve anomaly data set S3.

[0053] Furthermore, the error model commonly used in the TC method is expressed as follows:

[0054]

[0055] Among them, S i represents three independent groundwater storage anomaly data sets; Θ represents the unknown true GWSA (groundwater storage anomaly) value; and β i Respectively represent S i The additive and multiplicative deviations of the GWSA values ​​in relative to the true GWSA values; i Indicates S iRandom error; i∈[1,2,3].

[0056] Step 102, based on three randomly selected independent groundwater storage anomaly data sets, covariance, value change sensitivity, and random error variance are calculated to obtain the correlation coefficient and random error variance of each groundwater storage anomaly data set.

[0057] In this example, covariance estimation is used to solve the random error variance of three independent groundwater storage anomaly data sets:

[0058] According to the three independent groundwater storage anomaly data sets S i , perform covariance calculation, and obtain the covariance cov(S i ,S j ):

[0059]

[0060] Among them, E(S i S j ) indicates S i S j The mathematical expectation, E(S i ) indicates S i The mathematical expectation, E(S j ) indicates S j The mathematical expectation of; j∈[1,2,3]; β j Indicates S j The multiplicative deviation of the GWSA value in relative to the true GWSA value, represents the variance of Θ, ε j Indicates S j of random errors.

[0061] According to the four prerequisites that the TC method needs to meet, cov(ε i ,ε j )=0(i≠j),cov(ε i ,Θ)=0, then:

[0062]

[0063] in, Indicates S i The random error ε i Variance K i Indicates S i The sensitivity of the value change of i The higher the S i The response to GWSA is stronger. This sensitivity can be calculated using the covariance between the datasets:

[0064]

[0065] Furthermore, The estimate of can be obtained by subtracting the sensitivity from the total variance of the data set get:

[0066]

[0067] Finally, according to K i and Solve to get S i The correlation coefficient ρ i,Θ :

[0068]

[0069] Step 103: construct a groundwater weighted fusion model based on the obtained correlation coefficients and random error variances of various groundwater reserve anomaly data sets.

[0070] In this embodiment, according to ρ i,Θ and The weight W of the groundwater weighted fusion model can be determined i :

[0071]

[0072] Furthermore, according to the time series M of each independent groundwater storage anomaly data set at the same location i and determine W i , the following groundwater weighted fusion model Y is constructed model :

[0073] Y model =W1M1+W2M2+W3M3

[0074] Step 104, based on the constructed groundwater weighted fusion model, the groundwater reserve anomaly estimation is performed to obtain the groundwater reserve anomaly estimation result and output it.

[0075] In summary, the present invention discloses a method for improving the accuracy of regional groundwater reserves based on a groundwater weighted fusion model. The groundwater reserve anomaly (GWSA) can be accurately estimated by combining the Gravity Recovery and Climate Experiment Satellite (GRACE) data with the hydrological model. First, in order to obtain a more accurate GWSA, the present invention constructs a new groundwater weighted fusion model (GWFM) based on the square of the correlation coefficient and the error variance between the unknown measured data calculated by the extended triple combination (ETC) method. Second, taking the Shiyang River Basin as an example, the GWFM is evaluated using measured data. The results show that the GWFM integrates three GWSA results based on GRACE (GRACE-GLDAS, GRACE-ERA5 and GRACE-WGHM), and the correlation coefficient (CC) and Nash coefficient (NSE) between the measured data and the measured data are increased by 9% to 40% and 23% to 657% respectively compared with these three results, and the root mean square error (RMSE) is reduced by 9% to 28%, thereby verifying the accuracy of the GWFM. Third, GWFM was used to analyze the spatiotemporal distribution and influencing factors of GWSA in the Hexi Corridor. The results showed that: ① From 2005 to 2010, GWSA decreased at a rate of -3.39±0.48 mm / yr, while the downward trend after 2011 (-1.36±1.05 mm / yr) slowed down significantly compared with that from 2005 to 2010; ② Through the analysis of spatial trend distribution, there was a large downward trend in the middle of the Shule River Basin, the northern part of the Heihe River Basin and the northern part of the Shiyang River Basin, which were -3.39 mm / yr, -3.36 mm / yr and -4.10 mm / yr, respectively. In addition, the southeastern part of the Shiyang River Basin decreased at a rate of -2.25 mm / yr; ③ The impact of climate change and human activities on groundwater reserves in the study area was analyzed. The results showed that: for the changes in groundwater reserves in the Shule River Basin and the Shiyang River Basin, climate factors dominated, while for the Heihe River Basin, with the increase in water pumping, the impact of human activities on groundwater reserves became increasingly greater. The advantages of the new groundwater weighted fusion model are high accuracy in estimating regional groundwater reserves and fast calculation speed.

[0076] Based on the above embodiments, a specific research example is described below.

[0077] 1. Verification of the new groundwater weighted fusion model

[0078] 1.1) Overview of the study area

[0079] The Hexi Corridor is located in the arid northwest region of China, spanning 1,000 km from west to east. The Hexi region includes five prefecture-level cities: Zhangye, Jiuquan, Wuwei, Jinchang and Jiayuguan. The climate in the study area is continental arid, dry, with an average annual temperature of 2 to 10°C. The average annual precipitation is only 50 to 150 mm, and the annual precipitation decreases from west to east, while the annual evaporation is as high as 1,500 to 2,500 mm. The precipitation from May to September accounts for 80% of the total annual precipitation. The terrain of the region is high in the south and low in the north ( Figure 2 ), the southern Qilian Mountains and other high mountains are covered with abundant permanent snow and prehistoric glaciers. These snow and glaciers melt at certain seasons each year, providing a steady supply of fresh water for the large number of oases and farmland in the area.

[0080] The Hexi Corridor is rich in land resources and has a well-developed irrigation agriculture. It is an important agricultural production area in Gansu Province. However, due to the dry climate and scarce precipitation in the region, the demand for irrigation water has increased. It is a typical overlapping area of ​​ecological environment fragility and climate change sensitivity. In addition, the continuous development of agriculture has led to seasonal desertification of large areas of land, and the ecological environment has lost its self-repairing ability, which not only restricts the development of the local economy and society, but also endangers the ecological security of the Hexi Corridor.

[0081] 1.2) Methods

[0082] 1.2.1) Estimation of GWSA based on GRACE and hydrological model data

[0083] Usually GWSA is estimated by subtracting the surface water component from the TWS inverted by GRACE:

[0084] GWSA=TWS-SM-SWE-CWS

[0085] Among them, GWSA represents groundwater storage anomaly, TWS represents terrestrial water storage anomaly, SM represents soil water storage anomaly, SWE represents snow water equivalent storage anomaly, and CWS represents canopy water storage anomaly. Previous studies on TWSA have shown that soil moisture and groundwater are the main factors in TWS changes, and changes in ice and snow, biological water, and surface water are relatively small. Therefore, SWE and CWS are set to zero.

[0086] 1.2.2) Multiple linear regression analysis of time series

[0087] Since the change of groundwater reserves is seasonal and cyclical, multiple linear regression analysis is used. The regression model is set as:

[0088] Y(t)=η1+η2t+η3sin(πt)+η4cos(πt)+η5sin(2πt)+η6cos(2πt)+μ

[0089] Where Y(t) represents the time series of GWSA, η1 and η2 represent the constant term and trend term, η3 and η4 represent the anniversary term, η5 and η6 represent the semi-anniversary term, and μ represents the deviation or data error.

[0090] The calculation formulas for the annual amplitude and the semi-annual amplitude are as follows:

[0091]

[0092]

[0093] 1.2.3) Evaluate the contribution of different factors to GWSA

[0094] The GWSA estimated by GRACE includes both human activities and climate change. In order to evaluate the contribution of different factors, an indicator is used to evaluate the contribution of these two factors, and its calculation formula is as follows:

[0095] GWSC C =GWSC GRACE -GWSC H

[0096] GWSC=GWSA t -GWSA t-1

[0097]

[0098]

[0099] Among them, GWSC C stands for climate-driven groundwater storage change (GWSC); GWSC GRACE represents the annual GWSC estimated from GRACE data, GWSC H Represents the GWSC caused by human activities. ξ H and C denote the contribution of human activities and climate change to GWSC respectively. If ξ H and C If it is a positive value, it means that it will have a positive impact on groundwater reserves, and vice versa.

[0100] 1.3) Data

[0101] 1.3.1) GRACE data

[0102] GRACE products mainly include spherical harmonic coefficients and mass concentration. The GRACE data used in this embodiment are products processed by the regularized mass concentration (mascon) method, rather than the traditional spherical harmonic method. The benefit of this mascon solution is that users no longer need to perform spherical harmonic conversion or empirical scale factor processing on GRACE, and it can provide accurate surface-based grid TWS information. This embodiment uses the monthly GRACEmascon solution provided by the Space Research Center of the University of Texas at Austin to derive the terrestrial water storage anomaly (TWSA) with a resolution of 0.5°×0.5°. Since GRACE data are missing in some months, they are supplemented by linear interpolation in this embodiment.

[0103] 1.3.2) Soil moisture dataset

[0104] GLDAS is jointly developed by NASA and the National Oceanic and Atmospheric Administration's Environmental Forecast Center (NOAA) to obtain high-precision, high-resolution land surface state and flux information. Currently, GLDAS includes two versions of data sets (GLDAS-1 and GLDAS-2). This embodiment mainly uses the Noah model in GLDAS-2, with a spatial resolution of 1°×1° and a temporal resolution of 1 month.

[0105] WGHM was developed by the University of Frankfurt in Germany and provides all TWS components except glaciers. It comprehensively reflects the five main components of continental water storage, namely vegetation water storage, surface water storage, snow cover, soil water and groundwater storage. This example uses the soil water product in the WGHM hydrological model, with a spatial resolution of 0.5°×0.5° and a temporal resolution of 1 month.

[0106] ERA5-Land is a reanalysis dataset produced by reanalyzing the land portion of the ERA5 data. It is one of the most modern and best reanalysis datasets produced by the European Centre for Medium-Range Weather Forecasts (ECMWF) within the Copernicus Climate Change Service, with a spatial resolution of 0.1°×0.1° and a temporal resolution of 1 month.

[0107] 1.3.3) Groundwater level data

[0108] The measured groundwater level data selected in this embodiment comes from the "China Geological Environment Monitoring Groundwater Level Yearbook", which is compiled by the Ministry of Land and Resources. Due to the small number of monitoring stations and the lack of continuous data at individual stations, this embodiment selects the groundwater level data of 5 monitoring wells in the Shiyang River Basin, mainly used to verify the reliability of GWFM.

[0109] 1.3.4) Auxiliary Data

[0110] Precipitation data are from the China Meteorological Data Service Center (CMDC). The dataset is based on precipitation data from high-density ground stations in China (2472 national meteorological observation stations). It uses the thin plate spline method of the ANUSPLIN software for spatial interpolation to generate daily gridded data from 1961 to the present. In addition, ET and T from the GLDAS model during 2005-2014 were used to evaluate the relationship between GWSA and climate factors.

[0111] To assess the impact of human activities on GWSA, annual groundwater withdrawal data were also collected in the Hexi Corridor region from 2005 to 2014. The data were obtained from the Gansu Water Resources Bulletin (http: / / slt.gansu.gov.cn / slt / c106773 / zcfg.shtml) published by the Gansu Provincial Water Resources Department of China.

[0112] 1.4) Evaluation Metrics

[0113] In order to evaluate the accuracy of GWFM, this embodiment uses the correlation coefficient CC, the root mean square error RMSE and the Nash coefficient NSE to evaluate the reference measured data and the model data:

[0114]

[0115]

[0116]

[0117] Among them, A(t) represents the reference measured data set, B(t) represents the model data set, and A mean Represents the long-term mean of the reference measured data set. Since the scales of the model data sets and the reference measured data sets are inconsistent, the two are normalized for analysis.

[0118] 1.5) Experimental verification of the new groundwater weighted fusion model

[0119] Before applying it to the study area, it is necessary to test the performance of GWFM. In this example, groundwater level data of the Shiyang River Basin were used to test the performance of the model. The GWSA time series combined with GRACE and three hydrological model data (GLDAS, ERA5 and WGHM) were introduced as the weighting factor of GWFM, and the GWFM was verified by measured data.

[0120] Figure 3 represents the time series of GWSA, while the GWSA time series based on GWFM is Figure 3In terms of the long-term changes in GWSA, the GWSA obtained from GWFM is very consistent with the GWSA based on GRACE in terms of trend, amplitude and phase.

[0121] By comparing the mean of the five observation wells’ measured data with the temporal overlap of the GWSA based on GRACE and GWFM, it is obvious that the overall trends of GRACE-GRACE, GRACE-ERA5, GRACE-WGHM, GWFM and measured water level data are roughly similar, all showing a downward trend. However, there are obvious phase differences between them (e.g. Figure 3 As shown). In many studies, there is a time lag between the GWSA estimated by GRACE and the measured groundwater data. AbouZaki et al.'s study showed that there is a certain difference in phase between the time series of groundwater reserves estimated by GRACE in the Bakhtegan Basin and the measured groundwater level data, and there is an obvious lag; Thomas et al.'s study showed that when the delay time is set to 2 months, the correlation between the groundwater reserves in the Central Valley of California from 2003 to 2016 inverted by GRACE and the measured data reaches the maximum. Since the four GWSA time series have similar amplitudes and phases, this embodiment uses GRACE-GLDAS and measured data (S1 to S5) to explore the optimal lag time. The results show that when GWSA lags behind the measured data by 4 to 5 months, the correlation coefficient can reach the highest (CC = 0.59 to 0.72).

[0122] In order to explore the reliability of GWFM, the lag time was set to 4 months, and the GRACE-based results (GRACE-GLDAS, GRACE-ERA5, GRACE-WGHM) and GWFM results were compared with the measured groundwater level data ( Figure 4 ). The results show that the seasonal cycles of GRACE-GLDAS, GRACE-ERA5, GRACE-WGHM and GWFM are consistent with field observations. The amplitude of GRACE-ERA5 is larger than that of GRACE-GLDAS and GRACE-WGHM. The amplitudes of the measured groundwater levels (S1 to S5) show large differences. For example, the amplitude of S3 decreases significantly after 2011; the amplitude of the measured groundwater level S1 varies from -2 to 3 m, and the amplitude of S5 ranges from -8 to 16 m relative to the long-term average.

[0123] To further analyze the accuracy of GWFM results, the CC, RMSE and NSE between GRACE-GLDAS, GRACE-ERA5, GRACE-WGHM and measured groundwater levels were calculated. The results show that compared with the results of a single model, all statistical indicators have been improved to a certain extent. For example, the CC between the GWFM results and the measured data in the observation well S1 area increased from 0.54-0.73 to 0.74; the RMSE decreased from 0.44-0.54 to 0.39; and the NSE increased from 0.11-0.42 to 0.54. Compared with the mean of the observation wells, the CC (9%-40%), RMSE (9%-28%) and NSE (23%-657%) also increased similarly.

[0124] 2. Application of the new groundwater weighted fusion model

[0125] Through the verification of GWFM, the results show that the new groundwater weighted fusion model combines the advantages of multiple data sets, making the results more reliable. The following mainly uses GWFM to analyze the spatiotemporal changes and influencing factors of groundwater reserves in the Hexi Corridor.

[0126] 2.1) Comparison of GWSA time series based on GWFM and GRACE

[0127] Figure 5 The time series between GRACE-based GWSA and GWFM-based GWSA in the Hexi Corridor region from 2005 to 2014 are shown, including monthly, annual and seasonal scales. The overall trends based on GRACE and GWFM results show obvious consistency, and the four results have obvious periodicity. For the interannual variation of GWSA, most GWSA time series show obvious consistency, with positive anomalies from May to September and negative anomalies in other months. The only exception is GRACE-GLDAS, which all remain negative. The main reason is that the Hexi Corridor region is located in the arid area of ​​China, with hot summers, and 80% of the rainfall occurs from May to September, which effectively recharges groundwater. The Hexi Corridor region is a famous agricultural irrigation area in Northwest China, but surface water resources are scarce, and irrigation water mainly comes from groundwater. Therefore, in spring and winter, due to irrigation needs, a large amount of groundwater is extracted, resulting in a decrease in groundwater reserves.

[0128] As for the long-term trend of GWSA, the GWSA in the Hexi Corridor region showed a significant downward trend over the past 10 years, but 2011 was a turning point, and the downward trend before and after 2011 was significantly different. The downward trend before 2011 was significantly higher than the trend after 2011. Therefore, in order to clarify the differences between the four results, a multivariate regression analysis was performed on the GWSA time series in the Hexi Corridor region from 2005 to 2010 and from 2011 to 2014. This embodiment summarizes the statistical indicator information of the GWSA time series, including long-term trends, anniversary and semi-anniversary amplitudes. From 2005 to 2010, the four downward trends of GWSA ranged from 2.14 mm / yr (GRACE-ERA5) to 5.63 mm / yr (GRACE-GLDAS), which clearly shows the depletion of groundwater in the Hexi Corridor region during 2005 to 2011. Among them, the annual amplitude of GRACE-ERA5 is significantly larger than that of other results, 27% to 36% higher than that of other results. From 2011 to 2014, the four results also showed a downward trend, with a decline rate between 0.47 mm / yr (GRACE-ERA5) and 2.64 mm / yr (GRACE-GLDAS). Comparing the two periods, the decline rate from 2011 to 2014 was significantly lower than that from 2005 to 2010, decreasing by 49% to 78%. However, the annual amplitude and the semi-annual amplitude (except GRACE-GLDAS) increased significantly compared with the previous period. This may be related to the water use policy in the region. In 2011, Gansu Province gradually implemented the most stringent water resources management system and the measures of "leading by point and taking the lead in Hexi first".

[0129] One way to analyze the annual and semi-annual variations is to present their phasor diagrams, which show their amplitude and phase relative to a reference period (2004-2009). Figure 6 ). The length of each vector on the phasor diagram indicates the magnitude of the amplitude, while the direction of the vector indicates the phase. The greater the difference between the directions of the two vectors, the more the time series are offset relative to each other. In other words, the phase difference between the two time series will affect the magnitude of the correlation. The similar lengths of the two vectors indicate that the two time series have similar amplitude magnitudes, and the amplitude difference will affect the variance consistency. From the perspective of annual amplitude and phase, the annual amplitude of GRACE-ERA5 is higher than that of other results; although there is good amplitude consistency between GWSA based on GWFM and GRACE-WGHM, the phase consistency is poor. The results based on GWFM are more consistent with GRACE-ERA5 in phase. In terms of semi-annual amplitude and phase, there is a large difference between the semi-annual phase of GRACE-ERA5 and other results. The phase and amplitude of GWSA based on GWFM show good consistency with GRACE-GLDAS.

[0130] 2.2) Comparison of GWSA spatial distribution based on GWFM and GRACE

[0131] Figure 7 The figure shows the spatial distribution of the linear trend of GWSA based on GRACE and GWFM in the Hexi Corridor region from 2005 to 2014. Among these results, GRACE-GLDAS shows a high downward trend and coverage, but the spatial distribution in the Shule River Basin and Heihe River Basin is not clear. GRACE-ERA5 shows obvious spatial variation characteristics, such as D1 and D2 areas in the Shule River Basin, D3 area in the Heihe River Basin, and D4 area in the Shiyang River Basin are the main groundwater depletion areas. GRACE-WGHM shows an overall downward trend, with higher rates in the north and lower rates in the south, but the spatial distribution is not clear compared with other results. GWFM highlights significant groundwater depletion in areas d1, d2, d3, and d4, with rates of approximately -3.05 mm / yr, -4.15 mm / yr, -3.33 mm / yr, and -3.95 mm / yr, respectively, which are similar to the GRACE-ERA5 results. However, the two results show opposite trends in the southeastern part of the Shiyang River Basin, which are −2.31 mm / yr and 1.88 mm / yr, respectively.

[0132] The results based on GWFM are as follows Figure 8 (a) shows the simple average result of GWSA based on GRACE. Figure 8 (b). Compared with simple average results, GWFM can make full use of the advantages of multiple hydrological model data, retain the characteristics of regional changes, and provide more accurate GWSA results. Df1~Df3 are areas with high groundwater depletion, with rates of approximately -3.39mm / yr, -3.36mm / yr and -4.10mm / yr, respectively. There is a large difference between GWFM and Average in the southeastern Shiyang River Basin, with trends of -2.25mm / yr and -0.22mm / yr, respectively. The Gansu Provincial Government declared Gulang County, Wuwei City, an area of ​​groundwater overexploitation, i.e. Figure 8 (a) Df4 region, which is consistent with the results of GWFM. Therefore, the GWFM results can fully retain the advantages of different data sets, especially the change characteristics of special regions.

[0133] 2.3) GWSA’s response to climate change

[0134] In the context of climate change, precipitation and evapotranspiration are the factors that have the greatest impact on groundwater. Changes in temperature lead to changes in evapotranspiration, which in turn lead to changes in groundwater. According to the water balance equation, changes in groundwater reserves represent the difference between groundwater recharge and discharge. Groundwater recharge in the piedmont plain of the Hexi Corridor mainly comes from the infiltration of surface runoff, which accounts for 80% of the total recharge, followed by recharge from precipitation and infiltration of irrigation water. Groundwater discharge is mainly the extraction of groundwater and evapotranspiration of shallow groundwater.

[0135] In order to further analyze the detailed relationship between precipitation, temperature, evapotranspiration and GWSA, this embodiment adopts cross wavelet analysis. Fig. 9 Respectively represent the cross wavelet transform between GWS and precipitation, temperature and evapotranspiration in the Hexi Corridor region. Fig. 9 (a) shows that during the period of 2005-2014, in the Hexi Corridor region, the correlation between precipitation and GWSA was strong, and at the 95% confidence level, precipitation and GWSA showed a significant positive correlation. Evapotranspiration and temperature also showed a strong positive correlation. In addition, GWSA and precipitation, temperature and evapotranspiration all had a main resonance period of about 1 month, and the significant period was longer.

[0136] In the context of climate change, precipitation is the input of water and evapotranspiration is the output of water from a region. Therefore, precipitation minus evapotranspiration (P-ET) is considered a net recharge of surface water and groundwater. Fig.10 The maximum P-ET value appears in June to August, and the minimum value appears in January to April and October to December ( Fig.10 ). P-ET is basically greater than 0, in other words, surface water and groundwater can be effectively recharged. However, between 2009 and 2011, P-ET decreased significantly, indicating that the net recharge decreased significantly. Insufficient precipitation will directly hinder the growth of vegetation and human production, and excessive evapotranspiration will further accelerate the loss of available water resources and disrupt the balance of the water cycle. Although the net recharge during this period was positive, the groundwater was still in a declining state, indicating that the groundwater was not effectively recharged.

[0137] In addition, under the background of climate warming, snow will melt earlier in the year than before. This effect means less snow cover in winter and earlier peak of spring runoff. Snowmelt is an important water source for the arid region of northwest China. The development of oases outside the mountains is of great significance to maintaining ecological balance and sustainable development. Therefore, snowmelt in mountainous areas has a greater impact on runoff. Li et al. showed that from 1960 to 2010, the average annual runoff in the arid region of northwest China continued to increase. Among them, the runoff growth rate in the northern mountainous area of ​​Qilian Mountains was 1.48×10 8 m 3 / 10a.

[0138] 2.4) GWSA’s response to human activities

[0139] In addition to climate change, the impact of human factors on groundwater cannot be ignored. In the Hexi Corridor region, water resources are scarce and unevenly distributed. During the crop water demand season, people have to intensify the extraction of water resources for irrigation. When the surface water required for irrigation is limited, the contradiction between water supply and demand will become prominent, which will inevitably lead to excessive exploitation of groundwater. In the past few decades, the Hexi Corridor region has undergone tremendous changes. Niu et al. pointed out that in the past 30 years, the increase in water use for farmland irrigation has led to an average drop of about 1.86m in groundwater. Zhou et al. showed that between 2000 and 2010, as the farmland area increased by 11%, the total irrigation water demand increased by 6.3%.

[0140] Fig.11 represents the change of groundwater extraction in the Hexi Corridor region. During the entire study period (2005-2014), groundwater extraction in the Heihe River Basin showed a continuous upward trend, while other regions showed a trend of first decreasing and then increasing, reaching the lowest level in 2009-2011, which is consistent with the change of P-ET. It is worth noting that even in years with high net recharge, GWSA is in a state of decay, and in years with low net recharge, when groundwater extraction is also lower than in previous years, the rate of decline of GWSA is also higher than that in years with relatively high net recharge. This means that even if the amount of water extracted is lower than in previous years, the amount of water extracted is much higher than the amount of recharge. Therefore, if the utilization of water resources in the basin cannot be improved, the region will continue to face the problem of groundwater depletion in the future.

[0141] 2.5) Contribution of different factors to GWSA

[0142] The contributions of climate change and human activities to GWSC were evaluated by using the proposed method. Generally, the larger the η value, the greater its contribution to the change of GWSC. Among the climate and human factors in most sub-regions, climate change is an important factor affecting groundwater storage. In the Shule River Basin and Shiyang River Basin regions, the impact of climate change on GWSC is greater than that of human activities, while in the Heihe River Basin region, the impact of human activities on GWSC gradually becomes greater than that of climate change. Groundwater extraction has a significant impact on groundwater, especially in the Heihe River Basin region, where groundwater extraction has increased year by year, indicating that with the increase in water extraction, human activities have gradually become the dominant factor affecting GWSC in the Heihe River Basin region. Therefore, in addition to climate change, overexploitation of groundwater resources caused by human activities is also an important reason for the deterioration of groundwater sustainability in the Hexi Corridor region.

[0143] On the basis of the above embodiments, the present invention also discloses a system for improving the accuracy of regional groundwater reserves based on a groundwater weighted fusion model, comprising: a data set selection module, used to randomly select three independent groundwater reserve anomaly data sets; a solution module, used to perform covariance, numerical change sensitivity, and random error variance solution based on the three randomly selected independent groundwater reserve anomaly data sets, and obtain the correlation coefficient and random error variance of each groundwater reserve anomaly data set; a model construction module, used to construct a groundwater weighted fusion model based on the obtained correlation coefficient and random error variance of each groundwater reserve anomaly data set; an anomaly estimation module, used to perform groundwater reserve anomaly estimation based on the constructed groundwater weighted fusion model, obtain the groundwater reserve anomaly estimation result, and output it.

[0144] As for the system embodiment, since it corresponds to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the description of the method embodiment part.

[0145] Although the present invention has been disclosed as above in the form of a preferred embodiment, it is not intended to limit the present invention. Any person skilled in the art may make possible changes and modifications to the technical solution of the present invention by using the methods and technical contents disclosed above without departing from the spirit and scope of the present invention. Therefore, any simple modifications, equivalent changes and modifications made to the above embodiments based on the technical essence of the present invention without departing from the content of the technical solution of the present invention shall fall within the protection scope of the technical solution of the present invention.

[0146] The contents not described in detail in the specification of the present invention belong to the common knowledge of the professionals in this field.

Claims

1. A method for improving the accuracy of regional groundwater reserves based on a groundwater weighted fusion model, characterized in that: include: Three independent groundwater storage anomaly datasets were randomly selected; According to three randomly selected independent groundwater storage anomaly data sets, the covariance, numerical change sensitivity, and random error variance are solved to obtain the correlation coefficient and random error variance of each groundwater storage anomaly data set; According to the correlation coefficient and random error variance of each groundwater storage anomaly data set, a groundwater weighted fusion model is constructed. Based on the constructed groundwater weighted fusion model, the groundwater reserve anomaly estimation is performed to obtain the groundwater reserve anomaly estimation result and output it; The three randomly selected independent groundwater storage anomaly datasets are: the first groundwater storage anomaly dataset S1, the second groundwater storage anomaly dataset S2, and the second groundwater storage anomaly dataset S3, which are obtained by combining the gravity recovery and climate experiment satellite data with the GLDAS hydrological model, the ERA5 reanalysis dataset, and the WGHM hydrological model; Based on the three independent groundwater storage anomaly data sets obtained, the covariance, numerical change sensitivity, and random error variance are solved to obtain the correlation coefficient and random error variance of each groundwater storage anomaly data set, including: According to the three independent groundwater storage anomaly data sets S i , perform covariance calculation, and obtain the covariance cov(S i ,S j ), where i∈[1,2,3], j∈[1,2,3]; According to cov(S i ,S j ), calculate the sensitivity of numerical changes and the variance of random errors, and obtain S i The sensitivity of the numerical change K i and S i The random error variance According to K i and Solve to get S i The correlation coefficient ρ i,Θ ; cov(S i ,S j ) is calculated as follows: the(S i ,S j )=E(S i S j )-E(S i )E(S j )···(1) Among them, E(S i S j ) indicates S i S j The mathematical expectation, E(S i ) indicates S i The mathematical expectation, E(S j ) indicates S j The mathematical expectation of S i The sensitivity of the numerical change K i The solution process is as follows: cov(S i ,S j ) is denoted as C ij ; According to C ij , S is obtained by solving the following formula (2): i The sensitivity of the numerical change K i :

2. The method for improving the accuracy of regional groundwater reserves based on the groundwater weighted fusion model according to claim 1 is characterized in that: S i The random error variance The solution formula is as follows:

3. The method for improving the accuracy of regional groundwater reserves based on the groundwater weighted fusion model according to claim 2 is characterized in that: S i The correlation coefficient ρ i,Θ The solution formula is as follows:

4. The method for improving the accuracy of regional groundwater reserves based on the groundwater weighted fusion model according to claim 3 is characterized in that: According to the correlation coefficient and random error variance of the groundwater storage anomaly data sets, a groundwater weighted fusion model was constructed, including: According to i,Θ and Determine the weight W of the groundwater weighted fusion model i ; According to the time series M of each independent groundwater storage anomaly data set at the same location i and determine W i , the following groundwater weighted fusion model Y is constructed model : Y model =W1M1+W2M2+W3M3···(5)。 5. The method for improving the accuracy of regional groundwater reserves based on the groundwater weighted fusion model according to claim 4 is characterized in that: The weight W of the groundwater weighted fusion model i The solution formula is as follows:

Images