CN111241473B - Method for improving estimation accuracy of regional groundwater reserves - Google Patents

Abstract

The invention discloses a handleA method of high area groundwater reserve estimation accuracy, comprising: acquiring a change delta TWS of land water reserves on a month scale ₀ The method comprises the steps of carrying out a first treatment on the surface of the Method for extracting global month-scale soil water content change delta SM by using GLDAS hydrologic model ₁ Change in snow Water equivalent ΔSWE ₁ And vegetation canopy water reserves change ΔPCSW ₁ The method comprises the steps of carrying out a first treatment on the surface of the Extracting global month-scale soil water content change delta SM by utilizing WGHM hydrologic model ₂ Change in snow Water equivalent ΔSWE ₂ Vegetation canopy water reserves change ΔPCSW ₂ The method comprises the steps of carrying out a first treatment on the surface of the Solving to obtain the change delta GWS of the groundwater reserve of the month scale ₁ And delta GWS ₂ The method comprises the steps of carrying out a first treatment on the surface of the Measured month-scale groundwater reserve change ΔGWS from study area ₀ For DeltaGWS respectively ₁ And delta GWS ₂ Evaluating; selecting the groundwater reserve change delta GWS with the optimal month scale according to the evaluation result _{Excellent (excellent)} And outputting the result of the change of the groundwater reserve of the month scale of each pixel of the research area. According to the invention, based on a statistical selection method and combining GRACE satellite gravity data and a global hydrologic model, the estimation accuracy of regional groundwater reserves is improved.

Description

A method to improve the accuracy of regional groundwater reserve estimation

Technical field

The invention belongs to the intersection technical field of satellite gravity and hydrology, and in particular relates to a method for improving the accuracy of regional groundwater reserve estimation.

Background technique

Groundwater (GWS) is the largest freshwater resource in the global hydrological cycle, providing approximately 50% of the world's drinking water. In recent years, extreme climate, population growth, and over-exploitation of groundwater resources have led to serious consumption of groundwater resources. Therefore, understanding the dynamic changes of groundwater is crucial to water resources management and human survival.

Traditional groundwater change monitoring methods mainly rely on observation wells. Although the results can provide high-resolution groundwater level estimates, they have many limitations in practical applications. First, the construction and maintenance of observation wells is time-consuming and labor-intensive; second, the observation wells are unevenly distributed; and finally, single-point observation data cannot represent the results of a large area. The Gravity Retrieval and Climate Experiment Satellite (GRACE) program is jointly implemented by NASA and the German Space Flight Center (DLR). It was successfully launched in March 2002. The satellite can obtain terrestrial water reserves at all depths in the region. So far, the GRACE satellite is the only remote sensing method that can monitor TWS changes at all depths under any conditions. However, its main disadvantage is the inability to separate individual hydrological components from GRACE data.

In order to separate groundwater storage from terrestrial water storage, previous studies mainly used auxiliary information from individual hydrological models to vertically decompose GRACE data. The Global Land Data Assimilation System (GLDAS) provides hydrological flux estimates at 0.25° spatial resolution and has been applied in various hydrological studies. For example, groundwater decline in the North China Plain, groundwater recharge rate in China's Loess Plateau, and runoff assessment in the Qinghai-Tibet Plateau, etc.

Currently, many hydrological models and land surface models have been developed to describe various hydrological fluxes on land, such as WaterGAP Global Hydrology Model (WGHM), Community Atmosphere Biosphere LandExchange (CABLE) and World-Wide Water Resources Assessment (W3RA). Due to differences in model structure, parameter settings, and driving data, there are some differences in the output results of hydrological models. Typically, these models are developed on a global scale and therefore have pros and cons. For example, the global hydrological data of the GLDAS model is publicly available, but it does not simulate the groundwater component; the AWAR model simulates the groundwater component, but cannot describe the massive water resource consumption that occurs during drought periods. Therefore, the biggest problem with using a single hydrological model is uncertainty about whether the model output is suitable for a specific region.

Tasmania is located in southern Australia, with a total area of approximately 68,000km ² . Although it makes up less than 1% of Australia's surface area, the area accounts for approximately 12% of Australia's freshwater resources. Groundwater extraction rates in Tasmania are relatively low, with total consumption estimated at 38GL/yr. However, 90% of the extraction areas are in the northwest and north-central parts of the state, meaning groundwater extraction can cause localized problems in these areas. In addition, much of Tasmania's west coast is protected as a World Heritage Area, with native vegetation covering 50% of the state. Therefore, understanding the dynamic changes of GWS is of great significance to the local ecological environment. Moreover, due to the uneven distribution of groundwater level monitoring wells, it is difficult to use well observations to reliably estimate groundwater reserves. Especially between 2001 and 2009, a long-term drought occurred in southeastern Australia, the so-called "Millennium Drought", which affected the environment, agriculture and economic activities. There have been many studies showing that the drought has affected Tasmanian groundwater.

Contents of the invention

The technology of the present invention solves the problem by overcoming the shortcomings of the existing technology and providing a method to improve the estimation accuracy of regional groundwater reserves. Based on the statistical selection method combined with GRACE satellite gravity data and the global hydrological model, the estimation accuracy of regional groundwater reserves is improved.

In order to solve the above technical problems, the present invention discloses a method for improving the accuracy of regional groundwater storage estimation, including:

Obtain monthly scale terrestrial water storage changes ΔTWS ₀ ;

The GLDAS hydrological model was used to extract global monthly-scale changes in soil moisture content ΔSM ₁ , changes in snow water equivalent ΔSWE ₁ and changes in vegetation canopy water storage ΔPCSW ₁ ;

The WGHM hydrological model was used to extract global monthly-scale changes in soil moisture content ΔSM ₂ , changes in snow water equivalent ΔSWE ₂ , and changes in vegetation canopy water storage ΔPCSW ₂ ;

According to ΔTWS ₀ , ΔSM ₁ , ΔSWE ₁ and ΔPCSW ₁ , the monthly scale groundwater storage change ΔGWS ₁ is calculated; according to ΔTWS ₀ , ΔSM ₂ , ΔSWE ₂ and ΔPCSW ₂ , the monthly scale groundwater storage change ΔGWS ₂ is calculated;

According to the measured monthly groundwater storage change ΔGWS ₀ in the study area, ΔGWS ₁ and ΔGWS ₂ were evaluated respectively;

According to the evaluation results, the optimal monthly-scale groundwater storage change ΔGWS _{is selected} as the monthly-scale groundwater storage change result output for each pixel in the study area.

In the above method to improve the accuracy of regional groundwater storage estimation, the monthly scale terrestrial water storage change ΔTWS ₀ is obtained, including:

m monthly-scale land water storage changes ΔTWS ₁ , ΔTWS ₂ ...ΔTWS _m calculated based on spherical harmonic coefficients are obtained from m data sources; among them, m≥3;

Determine the regression models Z ₁ (t), Z ₂ (t), ... Z _m (t) of the time series corresponding to ΔTWS ₁ , ΔTWS ₂ ...ΔTWS _i ;

Solve Z ₁ (t), Z ₂ (t),...Z _m (t) respectively to obtain the value of the linear trend term in each regression model;

According to the comparison results of the values of the linear trend terms in each regression model, the optimal monthly terrestrial water storage change ΔTWS ₀ is selected from ΔTWS ₁ , ΔTWS ₂ ...ΔTWS _m , and output.

In the above method to improve the accuracy of regional groundwater reserve estimation, the general expression of the regression model of the corresponding time series of ΔTWS ₁ , ΔTWS ₂ ...ΔTWS _i is:

Among them, i∈m, β _i1 represents the constant term of the i-th regression model, β _i2 represents the linear trend term of the i-th regression model, β _i3 represents the annual sinusoidal signal of the i-th regression model, and β _i4 represents the i-th regression model. The annual cosine signal of the regression model, β _i5 represents the half-year sine signal of the i-th regression model, β _i6 represents the half-year cosine signal of the i-th regression model, and ε _i represents the data error of the i-th regression model.

In the above method to improve the accuracy of regional groundwater storage estimation, the solution formula for monthly scale groundwater storage changes is as follows:

ΔGWS ₁ =ΔTWS ₀ -ΔSM ₁ -ΔSWE ₁ -ΔPCSW ₁

ΔGWS ₂ =ΔTWS ₀ -ΔSM ₂ -ΔSWE ₂ -ΔPCSW ₂ .

In the above method to improve the accuracy of regional groundwater storage estimation, ΔGWS ₁ and ΔGWS ₂ are evaluated respectively based on the measured monthly groundwater storage change ΔGWS ₀ in the study area, including:

Determine the correlation coefficient PR ₁ and root mean square error RMSE ₁ between ΔGWS _{0 and} ΔGWS 1, and determine the correlation coefficient PR ₂ and root mean square error RMSE 2 between ΔGWS ₀ and ΔGWS _{2 ;}

Determine the respective slopes Tr ₀ , Tr ₁ and Tr ₂ of ΔGWS ₀ , ΔGWS ₁ and ΔGWS ₂ ;

The evaluation result Y ₁ of ΔGWS ₁ and the evaluation result Y ₂ of ΔGWS ₂ are obtained by solving:

Among them, F ₁₁ , F ₁₂ , F ₂₁ and F ₂₂ represent the weight coefficients of PR ₁ , RMSE ₁ , PR ₂ and RMSE ₂ respectively, and Consist(·) represents the trend consistency judgment function.

In the above method to improve the accuracy of regional groundwater reserve estimation, determine the correlation coefficient PR _{1 and} root mean square error RMSE ₁ between ΔGWS 0 and ΔGWS ₁ , determine the correlation coefficient PR ₂ and root mean square error RMSE ₂ between ΔGWS ₀ and ΔGWS ₂ , include:

Obtain the time series X(t) of ΔGWS ₀ , the time series Y ₁ (t) of ΔGWS ₁ , and the time series Y ₂ (t) of ΔGWS ₂ ;

The correlation coefficient is calculated as follows:

The root mean square error is calculated as follows:

Among them, n represents the length of the time series.

Among the above methods to improve the accuracy of regional groundwater reserve estimation,

When the trend of Tr ₀ is consistent with the trend of Tr ₁ , then Consist(Tr ₁ , Tr ₀ )=1; otherwise, Consist(Tr ₁ , Tr ₀ )=0;

When the trend of Tr ₀ is consistent with the trend of Tr ₂ , then Consist(Tr ₂ , Tr ₀ )=1; otherwise, Consist(Tr ₂ , Tr ₀ )=0.

Among the above methods to improve the accuracy of regional groundwater reserve estimation,

The invention has the following advantages:

The invention discloses a method for improving the estimation accuracy of regional groundwater reserves. Based on the statistical selection method combined with GRACE satellite gravity data and the global hydrological model, the estimation accuracy of regional groundwater reserves is improved, and an effective solution is provided for selecting appropriate hydrological models for hydrological applications. .

Description of the drawings

Figure 1 is a flow chart of a method for improving the accuracy of regional groundwater reserve estimation in an embodiment of the present invention;

Figure 2 is a schematic diagram of the regional average results of Tasmania's terrestrial water storage changes from 2003 to 2015 calculated using different GRACE products in the embodiment of the present invention;

Figure 3 is a schematic diagram of Tasmania's groundwater storage changes from 2003 to 2015 obtained by combining GRACE-GLDAS, GRACE-WGHM and WGHM in the embodiment of the present invention; wherein, 3a: monthly scale, 3b: seasonal scale, 3c: Correlation coefficient and RMSE between GRACE-GLDAS and GRACE-WGHM;

Figure 4 is a schematic diagram of the distribution position and number of groundwater level monitoring wells in Tasmania in an embodiment of the present invention;

Figure 5 is a schematic diagram comparing the GLDAS-GLDAS, GRACE-WGHM and WGHM results of four grids with the measured data in the embodiment of the present invention;

Figure 6 is a spatial distribution map of groundwater storage changes from 2003 to 2015 in the embodiment of the present invention; wherein, 6a: GLDAS-GLDAS, 6b: GRACE-WGHM; D and d represent areas with a downward trend, and R and r represent areas with an upward trend. trend area;

Figure 7 is a schematic diagram comparing the estimated results of GRACE-GLDAS and GRACE-WGHM groundwater storage changes in the R2 and d4 areas with measured data, as well as with temperature and rainfall in the embodiment of the present invention;

Figure 8 is a schematic diagram of the combined results of GRACE-GLDAS and GRACE-WGHM in the embodiment of the present invention; 8a: optimal selection result, 8b: simple average result;

Figure 9 is a schematic diagram of groundwater storage changes and corresponding rainfall changes from 2003 to 2015 in the embodiment of the present invention; 9a: monthly scale, 9b: inter-annual scale, 9c: average seasonal scale; gray bars represent real rainfall data, Black bars represent rainfall data adjusted with a 7-month lag;

Figure 10 is a schematic diagram of groundwater storage changes from 2003 to 2015 in the embodiment of the present invention; 10a: average results of the entire region, 10b drought index and rainfall integration results.

Detailed ways

In order to make the purpose, technical solutions and advantages of the present invention clearer, the disclosed embodiments of the present invention will be described in further detail below with reference to the accompanying drawings.

In the present invention, a method for improving the accuracy of regional groundwater reserve estimation disclosed in this embodiment is mainly explained through the following aspects.

1. Overview of the study area

Tasmania is located 240km south of mainland Australia, with a latitude and longitude range of 40°S-44°S and 144°E-148°E, and an area of 64,500 km ² . Most of the area consists of mountains and hills. The center is the highest part of the area, with peaks exceeding 1,500m above sea level. The central and eastern areas and coastal areas are relatively flat.

Tasmania's climate type is a temperate maritime climate, which is cool and mild. The average annual temperature is as high as 15.7℃ and as low as 4.5℃. Due to the influence of terrain, there is a large difference in precipitation between the east and west. The annual precipitation in most areas in the west exceeds 2000mm and reaches 4000mm in mountainous areas. The average annual precipitation in the eastern region is less than 750mm, and in some areas it is less than 400mm. The northeastern highlands receive less rainfall than surrounding areas. The amount is also very high, partly due to snowfall, with the annual precipitation being about 900mm; the annual precipitation in the southeastern region is evenly distributed, mostly less than 800mm. The region has approximately 150,000km of waterways, 8,800 wetlands and 94,000 water bodies. The river basin area of the island is 685~11700km ² .

2. Data source

2.1) GRACE land water storage anomaly data

The terrestrial water storage anomaly data used in the embodiment of the present invention is the Level3 grid product and the Mascon product. These two products are provided by the Center for Space Research (CSR) of the University of Texas in the United States. Level3 grid products are processed by the algorithms of Swenson and Wahr (2006) and Landerer and Swenson (2012), with a spatial resolution of 1° × 1°. Due to the sampling and post-processing of GRACE observation data, the surface mass change signal at smaller spatial scales tends to attenuate.

The scaling factor method is used to recover signal leakage (called CSR-scaled), the scaling factor is provided by ftp://podaac-ftp.jpl.nasa.gov/allData/tellus/L3/land_mass/.

Mascon is another basic equation for gravity field solution, with a spatial resolution of 0.5° × 0.5°. CSR Mascon (CSR-M) is constrained by a time-varying regularization matrix and is derived only from GRACE information without applying other models or data for constraints. Therefore, the CSR-M solution has no significant banding errors and captures the GRACE signal within the measurement noise level.

2.2) Hydrological model surface water storage data

The Global Land Data Assimilation System (GLDAS) was jointly developed by NASA, the National Center for Environmental Prediction (NCEP) and the National Oceanic and Atmospheric Administration (NOAA). It also established a global hydrological model and publicly released real-time satellite remote sensing observation data. With surface observation data, these data drive four land surface process models: CLM, MOS, VIC and NOAH, and 28 items of land surface meteorological data can be generated. The present invention adopts NOAH2.1 spatial resolution of 0.25°×0.25° and resampling of 0.5°×0.5°.

The Water GAP Global Hydrology Model (WGHM) was developed by the Institute of Physical Geography (IPG) of Frankfurt University in Germany and provides global 0.5° × 0.5° water resources information except Antarctica and Greenland. The WGHM model not only considers the groundwater component, but also considers the impact of human activities on water resource consumption.

2.3) Rainfall data

TRMM 3B43 is a standard monthly precipitation product that combines precipitation data sets, including TMI (TRMM Microwave Imager), PR (Precipitation Radar), VIRS (Visible and Infrared Scanner), SSM/I (Special Sensor Microwave Imager) and rain gauge data. TRMM 3B43 is derived by averaging the TRMM 3B42V6 precipitation product and is widely used in climatological applications. It provides estimates of total monthly rainfall recorded from 1988 to the present with a spatial resolution of 0.25°.

2.4) Ground observation data

The measured data of groundwater monitoring wells comes from the Global Ground Monitoring Network (GGMN). GGMN was initiated by UNESCO and implemented by IGRAC (International Groundwater Resources Assessment Centre), aiming to improve the quality and accessibility of groundwater monitoring information. Website (https://ggmn.un-igrac.org/) provides global spatiotemporal groundwater monitoring data, and the data is collected 1 to 2 times a day. This invention averages monthly water level information to obtain monthly water level information. Since the water supply degree is unknown, the water level is not converted into reserves.

The Australian Bureau of Meteorology (BoM, http://www.bom.gov.au/climate/data/) provides precipitation and temperature data based on surface observations. Despite their limited range and inherent errors, climate stations remain the most direct and precise measurement tools available. Therefore, in the following discussion ground-based measurements are considered “true precipitation” and “true temperature”.

3. Method

In this embodiment, as shown in Figure 1, the method for improving the accuracy of regional groundwater reserve estimation includes:

Step 101: Obtain monthly terrestrial water storage change ΔTWS ₀ .

In this embodiment, first, m monthly-scale land water storage changes ΔTWS ₁ , ΔTWS ₂ ...ΔTWS _m calculated based on spherical harmonic coefficients can be obtained from m data sources; where, ΔTWS ₁ , ΔTWS m are determined ΔTWS ₂ ...ΔTWS _i 's corresponding time series regression models Z ₁ (t), Z ₂ (t),...Z _m (t); then, for Z ₁ (t), Z ₂ (t) ,...Z _m (t) are solved respectively to obtain the value of the linear trend term in each regression model; finally, based on the comparison results of the values of the linear trend term in each regression model, from ΔTWS ₁ , ΔTWS ₂ . ..Screen out the optimal monthly terrestrial water storage change ΔTWS ₀ in ΔTWS _m and output it.

Preferably, the general expression of the regression model of the time series corresponding to ΔTWS ₁ , ΔTWS ₂ ...ΔTWS _i can be as follows:

Among them, m≥3, i∈m, β _i1 represents the constant term of the i-th regression model, β _i2 represents the linear trend term of the i-th regression model, β _i3 represents the annual sinusoidal signal of the i-th regression model, β _i4 represents the annual cosine signal of the i-th regression model, β _i5 represents the half-year sine signal of the i-th regression model, β _i6 represents the half-year cosine signal of the i-th regression model, and ε _i represents the data error of the i-th regression model.

Step 102: Use the GLDAS hydrological model to extract global monthly soil moisture content changes ΔSM ₁ , snow water equivalent changes ΔSWE ₁ and vegetation canopy water storage changes ΔPCSW ₁ .

Step 103: Use the WGHM hydrological model to extract global monthly-scale changes in soil moisture content ΔSM ₂ , changes in snow water equivalent ΔSWE ₂ , and changes in vegetation canopy water storage ΔPCSW ₂ .

Step 104: According to ΔTWS ₀ , ΔSM ₁ , ΔSWE ₁ and ΔPCSW ₁ , the monthly-scale groundwater storage change ΔGWS ₁ is calculated; based on ΔTWS ₀ , ΔSM ₂ , ΔSWE ₂ and ΔPCSW ₂ , the monthly-scale groundwater storage change ΔGWS is calculated ₂ .

In this embodiment, the calculation formula for monthly scale changes in groundwater storage is as follows:

ΔGWS ₁ =ΔTWS ₀ -ΔSM ₁ -ΔSWE ₁ -ΔPCSW ₁

ΔGWS ₂ =ΔTWS ₀ -ΔSM ₂ -ΔSWE ₂ -ΔPCSW ₂

Step 105: Evaluate ΔGWS ₁ and ΔGWS ₂ respectively based on the measured monthly groundwater storage change ΔGWS ₀ in the study area.

In this embodiment, the specific evaluation process is as follows:

1) Determine the correlation coefficient PR ₁ and root mean square error RMSE ₁ between ΔGWS ₀ and ΔGWS ₁ , and determine the correlation coefficient PR ₂ and root mean square error RMSE ₂ between ΔGWS ₀ and ΔGWS ₂ .

In this embodiment, the time series X(t) of ΔGWS ₀ , the time series Y ₁ (t) of ΔGWS ₁ and the time series Y ₂ (t) of ΔGWS ₂ can be obtained, then:

The correlation coefficient is calculated as follows:

The root mean square error is calculated as follows:

Among them, n represents the length of the time series.

2) Determine the respective slopes Tr ₀ , Tr ₁ and Tr ₂ of ΔGWS ₀ , ΔGWS ₁ and ΔGWS ₂ .

3) Solve to obtain the evaluation result Y ₁ of ΔGWS ₁ and the evaluation result Y ₂ of ΔGWS ₂ :

Among them, F ₁₁ , F ₁₂ , F ₂₁ and F ₂₂ represent the weight coefficients of PR ₁ , RMSE ₁ , PR ₂ and RMSE ₂ respectively.

Preferably, Consist(·) represents the trend consistency judgment function. Among them, when the trend of Tr ₀ is consistent with the trend of Tr ₁ , then Consist(Tr ₁ , Tr ₀ )=1; otherwise, Consist(Tr ₁ , Tr ₀ )=0. When the trend of Tr ₀ is consistent with the trend of Tr ₂ , then Consist(Tr ₂ , Tr ₀ )=1; otherwise, Consist(Tr ₂ , Tr ₀ )=0.

Step 106: Select the optimal monthly groundwater storage change ΔGWS _optimal according to the evaluation results and output it as the monthly groundwater storage change result for each pixel in the study area.

In this embodiment, the greater the evaluation value obtained according to step 105, the smaller the difference between the output result of the selected hydrological model and the actual measured data. Right now:

4. Results and analysis

4.1) Changes in terrestrial water reserves

The regional average results of Tasmania's terrestrial water storage changes from 2003 to 2015 calculated using different GRACE products are shown in Figure 2. The CSR-scaled result is larger than the CSR-SH and CSR-M results, and the difference in amplitude from these two results is 42.6mm and 21.34mm respectively. This indicates that in the Tasmanian region, the signal corrected with the scale factor is overestimated, possibly due to a certain error in the TWS in the CLM4.0 model. Between 2003 and 2015, all TWS change sequences showed an upward trend, with slopes ranging from 0.33mm/yr (CSR-SH) to 1.49mm/yr (CSR-scaled). The shading in the figure indicates the uncertainty of the CSR-scaled and CSR-Mascon results, with values of 46.26mm and 21.34mm.

There is a large uncertainty in the land water storage anomaly of CSR-SH, and the rough spatial resolution is one of the main shortcomings of the data; in addition, CSR-M can clearly define land and ocean areas, which can effectively reduce the impact of leakage errors and Noise is suppressed during processing, with virtually no post-processing requirements. Therefore, the CSR-M results are selected to describe the anomaly characteristics of terrestrial water storage in the following discussion.

4.2) Analysis of long-term changes in groundwater reserves

Figure 3 shows the overall changes in groundwater reserves in Tasmania from 2003 to 2015 obtained by GRACE-GLDAS, GRACE-WGHM and WGHM. The overall trends of the GRACE-GLDAS and GRACE-WGHM results are basically consistent, with a correlation coefficient of 0.82, and both results show obvious periodicity (Figure 3a). The annual amplitudes of GRACE-GLDAS and GRACE-WGHM are 40.75mm and 65.41mm respectively, and the annual phases are 76.37° and 69.91° respectively. However, in terms of seasonal characteristics, WGHM is opposite to the other two results. When the WGHM result reaches its peak, the other two results trough. In addition, the amplitude of WGHM results is undetermined over approximately 30% of Tasmania, which is quite inaccurate for groundwater estimates.

Changes in groundwater reserves show strong cyclicality, with a surplus from January to May and a loss from June to November. In the colder season, the difference between GRACE-GLDAS and GRACE-WGHM is larger, with the largest difference reaching 42.48mm in September (Fig. 3b). Generally speaking, the GWS changes of GRACE-GLDAS are consistent with those of GRACE-WGHM in grids other than G23 and G24, with correlation coefficients ranging from 0.59 to 0.85. The RMSE between GRACE-GLDAS and GRACE-WGHM is about 55mm, and the maximum and minimum values appear in G21 and G40, which are 61.61mm and 27.61mm respectively (Fig. 3c).

4.3) Verification of groundwater storage changes based on GRACE satellite gravity and hydrological model

Tasmania is spatially divided into 48 grids based on 0.5° longitude and latitude 0.5°, as shown in Figure 4. The dots represent monitoring wells, which do not completely cover the entire study area. The comparison between the estimated results of groundwater storage changes by GRACE-GLDAS, GRACE-WGHM and WGHM and the measured data is shown in Figure 5. This invention only selects 4 grids. In each grid, the seasonality and periodicity of GRACE-GLDAS and GRACE-WGHM are relatively close to the measured data. In addition, in the G8 grid, the amplitude of the GRACE-WGHM results is significantly larger than that of GRACE-GLDAS, while in G32 GRACE-GLDAS has a larger amplitude of change. The vibration amplitude of the measured data changes greatly, for example, it changes from -5 to 5m in G14 (Figure 5b) and -0.3 to 0.3m in G34 (Figure 5d). The WGHM results show opposite seasonal characteristics, with almost no changing trend in G35. This shows that the groundwater storage estimation of the WGHM model in Tasmania needs to be greatly improved.

In eastern Tasmania, GRACE-GLDAS results are highly correlated with measured data, with correlation coefficients ranging from 0.64 (G41) to 0.85 (G33). In the northern region, the GRACE-WGHM results are highly correlated with the measured data, with correlation coefficients ranging from 0.69 (G21) to 0.88 (G26). Except for G27 and G41, the RMSE results are consistent with the correlation coefficient conclusions.

For the time series trend items, except for G21 and G27, both the GRACE-hydrological model results and the measured data show an upward trend. Furthermore, the slope of GRACE-GLDAS is typically 1.8 times that of GRACE-WGHM. In G21 and G27, both GRACE-WGHM and measured data show a downward trend, while GRACE-GLDAS has an opposite trend. The reason may be that the GLDAS model cannot produce accurate estimates of hydrological variables at high altitudes due to shortcomings in meteorological forcing data and model parameters.

4.4) Spatial distribution of groundwater storage change trends

GRACE-GLDAS and GRACE-WGHM were used to calculate the spatial distribution of groundwater storage change rates in Tasmania from 2003 to 2015. The results are shown in Figure 6. D1-D3 and d1-d4 are the main areas showing a downward trend shown by the GRACE-GLDAS and GRACE-WGHM results respectively. R1-R2 and r1 are the main areas showing an upward trend shown by the GRACE-GLDAS and GRACE-WGHM results respectively. As can be seen from Figure 6, the spatial distribution of groundwater storage change rates in the study area obtained by GRACE-GLDAS and GRACE-WGHM inversions is relatively consistent. The results of both models show that: from 2003 to 2015, the western coastal and southwestern Tasmania The groundwater reserves in the southern and southern regions show a downward trend. The fastest declining areas are mainly in the western coastal areas, and the fastest increasing areas are mainly in the central and northern regions. The GRACE-GLDAS results show that the area of groundwater reserves with an upward trend is larger than that of GRACE-WGHM, and the rate of increase is large. The area of area with a downward trend is smaller than that of GRACE-WGHM, and the rate of decrease is small. That is, in most areas, GRACE-GLDAS The rate of change is high. The largest difference occurs in the central plateau region (R2 and d4), where the two results show opposite trends, with rates of 2.93mm/yr and -2.36mm/yr respectively.

According to Figure 7, the reason why the two results show opposite trends in R2 and d4 can be explained. Figure 7a shows that the correlation between GRACE-WGHM and measured data is higher than that between GRACE-GLDAS and measured data, with correlation coefficients of 0.70 and 0.41 respectively. The main reasons leading to the opposite trend include: (1) This area is located in the central plateau, and the groundwater supply source mainly comes from snow and glacier meltwater, so GWS changes are greatly affected by temperature. Figure 7b shows that the GRACE-WGHM results are more sensitive to temperature changes, while the GRACE-GLDAS results cannot capture the dynamic changes in temperature, especially from May to July. The correlation coefficients of GRACE-WGHM and GRACE-GLDAS with temperature data are 0.89 and 0.55 respectively; (2) The SM changes obtained by WGHM are larger in amplitude than GLDAS, as shown in Figure 7c. The main reason is that the rainfall driving data of GLDAS and WGHM are different and the soil layer and soil depth defined by the model are different.

4.5) Spatial distribution of optimal estimates of groundwater storage changes

According to the statistical selection method, the overall indicators show that YGRACE-WGHM and GRACE-GLDAS are larger in the gray area (Figure 4). This shows that the GRACE-WGHM and GRACE-GLDAS results are in good agreement with the measured data in the gray area. For areas lacking measured data, the correlation coefficients and RMSE values between GRACE-GLDAS and GRACE-WGHM are 0.74 and 44.17 respectively, which are better than the entire Tasmanian average (0.68mm and 45.15mm). Therefore, the GRACE-WGHM results are used as the final estimate for the central and northern regions, the eastern coastal and southeastern regions use GRACE-GLDAS as the final estimate, and the average of the two is taken for other regions. The spatial distribution results of the final groundwater storage change rate in the study area are obtained as shown in Figure 8a shown. In addition, Figure 8b shows the simple average results of the two results over the entire area.

Compared with the simple average result in Figure 8b, the improved result can comprehensively utilize different model results and retain specific regional characteristics, as shown in Figure 8a. DF1-DF4 are the main areas where groundwater reserves show a downward trend, with the decreasing rates being approximately -2.21mm/yr, -3.37mm/yr, -3.19mm/yr and -2.36mm/yr respectively. RF1 is the main area showing an upward trend, with an increase rate of approximately 5.43mm/yr. The dynamic changes of GWS are mainly affected by factors such as rainfall, human activities, geological and terrain conditions. The areas with a very clear downward trend (DF1-DF2) mainly include rugged mountains and vast forests. Therefore, minimal human activity and interannual variability in rainfall may play an important role in the dramatic decline of groundwater in these areas.

Figure 9 compares the DF2 groundwater storage change results and actual precipitation results. It can be seen that the change results of groundwater storage lag behind the rainfall data because rainfall takes time to recharge groundwater. This invention counts the lag correlation coefficients of the time series of rainfall and groundwater reserves in four regions. The lag periods are respectively set to 0 to 8 months. It can be seen that the lag periods of the four regions are 7 months (the lag correlation coefficients are 0.49 to 0.63 ). From 2003 to 2015, both groundwater storage and rainfall showed a significant downward trend, with the decline rates being -3.36mm/yr and -39.04mm/yr respectively (Figure 9b), which indicates that the decline in groundwater storage is mainly caused by reduced precipitation. For the seasonal cycle, rainfall mainly occurs from September to May, accounting for 60% of the annual rainfall, which is consistent with changes in groundwater storage (Fig. 9c).

4.6) Analysis of time change trends of groundwater reserves and climate impact

In the measured water well data area, the groundwater storage change results estimated by the GRACE satellite gravity and hydrology model are highly consistent with the measured data, which shows that the GRACE satellite gravity and hydrology model can be used to estimate groundwater storage changes in the entire Tasmanian region ( Figure 10a). The change results of groundwater storage showed a downward trend from January 2003 to September 2010, with a decline rate of -2.57mm/yr. It also showed that Tasmania was affected by the "Millennium Drought". The present invention uses the adaptive Palmer Drought Index (scPDSI) as a meteorological drought indicator. scPDSI data comes from ClimaticResearch (http://www.cru.uea.ac.uk/cru/data). PDSI less than -2 indicates severe drought. As can be seen from Figure 10b, from 2006 to 2009, scPDSI was lower than -2, indicating a severe drought climate in the region. This arid climate has an impact on groundwater reserves, which caused a serious decline in groundwater reserves at this stage, with a decline rate of -9.47mm/yr. During the drought phase, rainfall is below average and shows a downward trend, which may be the main cause of drought. From 2011 to 2015, groundwater reserves recovered at a rate of 3.94mm/yr, mainly due to increased rainfall.

Although the present invention has been disclosed above in terms of preferred embodiments, they are not intended to limit the present invention. Any person skilled in the art can utilize the methods and technical contents disclosed above to improve the present invention without departing from the spirit and scope of the present invention. Possible changes and modifications are made to the technical solution. 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, all belong to the technical solution of the present invention. protected range.

Contents not described in detail in the specification of the present invention belong to the well-known techniques of those skilled in the art.

Claims (6)

1. A method for improving accuracy of regional groundwater reserve estimation, comprising:

acquiring a change delta TWS of land water reserves on a month scale ₀ ；

Method for extracting global month-scale soil water content change delta SM by using GLDAS hydrologic model ₁ Change in snow Water equivalent ΔSWE ₁ And vegetation canopy water reserves change ΔPCSW ₁ ；

Extracting global month-scale soil water content change delta SM by utilizing WGHM hydrologic model ₂ Change in snow Water equivalent ΔSWE ₂ Vegetation canopy water reserves changeConversion of ΔPCSW ₂ ；

According to DeltaTWS ₀ 、ΔSM ₁ 、ΔSWE ₁ And ΔPCSW ₁ Solving to obtain the change delta GWS of the groundwater reserve of the month scale ₁ The method comprises the steps of carrying out a first treatment on the surface of the According to DeltaTWS ₀ 、ΔSM ₂ 、ΔSWE ₂ And ΔPCSW ₂ Solving to obtain the change delta GWS of the groundwater reserve of the month scale ₂ ；

Measured month-scale groundwater reserve change ΔGWS from study area ₀ For DeltaGWS respectively ₁ And delta GWS ₂ Evaluating;

selecting the groundwater reserve change delta GWS with the optimal month scale according to the evaluation result _{Excellent (excellent)} Outputting a result of the change of the groundwater reserve of the month scale of each pixel of the research area;

wherein:

measured month-scale groundwater reserve change ΔGWS from study area ₀ For DeltaGWS respectively ₁ And delta GWS ₂ Performing the evaluation, comprising:

determining ΔGWS ₀ And delta GWS ₁ Related coefficient PR of (2) ₁ Root mean square error RMSE ₁ Determining Δgws ₀ And delta GWS ₂ Related coefficient PR of (2) ₂ Root mean square error RMSE ₂ ；

Determining ΔGWS ₀ 、ΔGWS ₁ And delta GWS ₂ Respective slopes Tr ₀ 、Tr ₁ And Tr ₂ ；

Resolving to obtain delta GWS ₁ Is the evaluation result Y of (2) ₁ And delta GWS ₂ Is the evaluation result Y of (2) ₂ ：

Wherein F is ₁₁ 、F ₁₂ 、F ₂₁ And F ₂₂ Respectively represent PR ₁ 、RMSE ₁ 、PR ₂ And RMSE ₂ Consist (·) represents a trend consistency judgment function;

determining ΔGWS ₀ And delta GWS ₁ Related coefficient PR of (2) ₁ Root mean square error RMSE ₁ Determining Δgws ₀ And delta GWS ₂ Related coefficient PR of (2) ₂ Root mean square error RMSE ₂ Comprising:

acquisition of ΔGWS ₀ Time series X (t), ΔGWS of (2) ₁ Time series Y of (2) ₁ (t) and ΔGWS ₂ Time series Y of (2) ₂ (t)；

The correlation coefficient is calculated as follows:

the root mean square error is calculated as follows:

where n represents the length of the time series.

2. The method for improving accuracy of regional groundwater reserve estimation according to claim 1, wherein a terrestrial water reserve variation Δtws of a lunar scale is obtained ₀ Comprising:

obtaining m lunar-scale land water reserve changes delta TWS based on spherical harmonic coefficient calculation from m data sources ₁ 、ΔTWS ₂ ...ΔTWS _m The method comprises the steps of carrying out a first treatment on the surface of the Wherein,,m≥3；

determining ΔTWS ₁ 、ΔTWS ₂ ...ΔTWS _i Regression model Z of the respective time series ₁ (t)、Z ₂ (t)、...Z _m (t)；

For Z ₁ (t)、Z ₂ (t)、...Z _m (t) respectively carrying out solution to obtain the value of the linear trend term in each regression model;

from ΔTWS based on the comparison of the values of the linear trend terms in the regression models ₁ 、ΔTWS ₂ ...ΔTWS _m Obtaining the optimal terrestrial water reserve change delta TWS of the month scale through medium screening ₀ And output.

3. The method for improving accuracy of regional groundwater reserve estimation according to claim 2, wherein Δtws ₁ 、ΔTWS ₂ ...ΔTWS _i The general expression of the regression model of the respective corresponding time series is:

wherein i is m, beta _i1 Constant term, beta, representing the ith regression model _i2 Linear trend term, beta, representing the ith regression model _i3 Annual sinusoidal signal, beta, representing the ith regression model _i4 Annual cosine signal, beta, representing the ith regression model _i5 Half-year sinusoidal signal, beta, representing the ith regression model _i6 Half year cosine signal, epsilon, representing the ith regression model _i Representing the data error of the ith regression model.

4. The method for improving accuracy of regional groundwater reserve estimation according to claim 1, wherein a solution formula for a change in groundwater reserve on a monthly scale is as follows:

ΔGWS ₁ ＝ΔTWS ₀ -ΔSM ₁ -ΔSWE ₁ -ΔPCSW ₁

ΔGWS ₂ ＝ΔTWS ₀ -ΔSM ₂ -ΔSWE ₂ -ΔPCSW ₂ 。

5. the method for improving accuracy of regional groundwater reserve estimation according to claim 1,

when Tr is ₀ Trend and Tr of (2) ₁ Is consistent with the trend of Consist (Tr) ₁ ,Tr ₀ ) =1; otherwise, consist (Tr ₁ ,Tr ₀ )＝0；

When Tr is ₀ Trend and Tr of (2) ₂ Is consistent with the trend of Consist (Tr) ₂ ,Tr ₀ ) =1; otherwise, consist (Tr ₂ ,Tr ₀ )＝0。

6. The method for improving accuracy of regional groundwater reserve estimation according to claim 1,