CN-121997831-A - Groundwater model method for constructing snow melting and supplying effects based on CMA-RA/Land data

CN121997831ACN 121997831 ACN121997831 ACN 121997831ACN-121997831-A

Abstract

The invention belongs to the technical field of mine engineering, and particularly discloses a groundwater model method for constructing snow melting and supplying effects based on CMA-RA/Land data. The method comprises the steps of obtaining a CMA-RA/Land data set of a research area, creating data extraction points, extracting a meteorological variable time sequence corresponding to each data extraction point from the data set, performing spatial interpolation and unit conversion, determining source and sink items based on the meteorological variable time sequence and combining groundwater burial depth conditions, establishing a groundwater model, setting initial and boundary conditions, combining the source and sink items as input parameters and the set conditions, simulating a groundwater dynamic process by using the groundwater model, analyzing output characteristics of the groundwater model by using a SHAP method based on simulation results, optimizing the groundwater model according to analysis, and obtaining the groundwater model with snow melting supply influence. The invention has the characteristics of easy data acquisition, definite physical meaning of parameters and strong certainty of simulation results.

Inventors

ZHANG QIUYUAN
FANG XIN
YU LEI
DONG QUAN
Tong Xinlei
XIONG XIAOFENG
LIU JINGCHAO
XU ZHIMIN
YAO HENG
CHEN WEIXIAO
LI BIN
CHEN TIANCI
Xing Ruishu

Assignees

内蒙古平庄煤业(集团)有限责任公司元宝山露天煤矿
中国矿业大学

Dates

Publication Date: 20260508
Application Date: 20260129

Claims (10)

1. The groundwater model method for constructing the snow melting and supplying effect based on CMA-RA/Land data is characterized by comprising the steps of acquiring data, creating data extraction points, processing the data, determining source and sink items, dynamically simulating and optimizing the model, wherein the specific contents of the steps are as follows: A. acquiring data, namely acquiring drilling data of a research area and the snow melting infiltration amount and the groundwater level monitored on site, and acquiring a CMA-RA/Land data set corresponding to the research area from a national weather science data center; B. creating data extraction points, namely creating the data extraction points of the research area by adopting geographic information system software; C. b, data processing, namely extracting weather variable time sequences corresponding to the data extraction points created in the step B in batches from a CMA-RA/Land data set, and performing spatial interpolation and unit conversion processing; D. C, determining source and sink items, namely determining the source and sink items comprising rainfall, evaporation and equivalent water amount of snow melting based on the meteorological variable time sequence extracted in the step C and by combining with the underground water burial depth condition of a research area, judging the relation of underground water supply, runoff and excretion; E. establishing a three-dimensional geological structure model based on drilling data of a research area, establishing a groundwater model based on the three-dimensional geological structure model, taking the source and sink items determined in the step D as input parameters of the groundwater model, setting initial conditions and boundary conditions of the groundwater model, freely combining the input parameters and the set conditions to obtain a plurality of groups of parameter combinations, executing calculation batch operation on the groundwater model based on the plurality of groups of parameter combinations to obtain groundwater dynamic data, and then adjusting snow melting and infiltration replenishment coefficients of the groundwater model according to the on-site monitoring actual measurement values obtained in the step A; F. And E, carrying out importance and influence degree analysis on the output characteristic y=f (p 1 , p 2 , p 3 …, p n ) with n characteristics of the groundwater model by adopting a SHAP method based on the groundwater dynamic data and the adjusted groundwater model obtained in the step E, quantifying the influence degree of each parameter in the groundwater model on the predicted result of the water inflow of the pit according to the analysis, and optimizing the groundwater model according to the quantified influence degree to obtain the groundwater model influenced by snow melting and supplying for predicting the water inflow of the pit.
2. The method for constructing a groundwater model for snow melt replenishment influence based on CMA-RA/Land data according to claim 1, wherein in the step A, a CMA-RA/Land dataset corresponding to a research area is retrieved and downloaded through a national weather science data center website, and then variable names and units contained in the dataset are identified by adopting geoscience data visualization software.
3. The method for constructing the groundwater model for the snow melting and replenishing effects based on CMA-RA/Land data according to claim 1, wherein in the step B, a geographic information system software is adopted to arrange scattered points which are spatially distributed in a research area, the distance between any two scattered points is not smaller than the spatial resolution of the CMA-RA/Land data set, and then an extraction point file extrac _point.shp in a shape format is generated.
4. The method for constructing the groundwater model influenced by the snow melt replenishment based on CMA-RA/Land data according to claim 1, wherein the step C is characterized by comprising the following specific processes: C10, reading extrac _Point.shp file, and assigning unique identifier to each extraction point, then reading time, SWE_inst, SWE_tavg, evap_ tavg, snowCover _ tavg, snowDepth _ tavg, totalPrecip _ tavg variables in CMA-RA/Land dataset; C20, calculating Euclidean distance between an extraction point P s (λ s , φ s ) and each grid point (lambda i , φ j ) in the CMA-RA/Land data set by adopting a nearest neighbor interpolation method, selecting a grid point index (i \* ,j \* ) with the smallest distance, and directly assigning a physical quantity value V i\*,j\* (t k of the nearest grid point index (i \* ,j \* ) to the extraction point V s (t k ); Wherein, (lambda s , φ s ) is the longitude and latitude of the extraction point P s , (lambda i , φ j ) is the longitude and latitude of the ith row and jth column lattice points of the CMA-RA/Land dataset, V s (t k ) is the value of the corresponding physical quantity of the nearest lattice point at the lattice point and the moment t k , and the subscript "i \* ,j \* " represents the lattice point in the CMA-RA/Land dataset nearest to the extraction point; C30, the unit of instantaneous value SWE_inst and average value SWE_ tavg of equivalent water amount of snow melting is converted into mm by kg/m 2 , and the unit of rainfall TotalPrecip _ tavg and evaporation Evap_ tavg is converted into mm/d by kg/(m 2 s).
5. The method for constructing a groundwater model for snow-melting replenishment impact based on CMA-RA/Land data according to claim 4, wherein in step C10, python is used to write a program code, then GeoPandas is used to read extrac _point.shp file containing geographic coordinates and to assign unique identifiers to each extraction point, and then netCDF4 is used to read time, SWE_inst, SWE_tavg, evap_ tavg, snowCover _ tavg, snowDepth _ tavg, totalPrecip _ tavg variables in the CMA-RA/Land dataset; in the step C20, first, the euclidean distance of the extraction point P s (λ s , φ s ) from each grid point (λ i , φ j ) in the CMA-RA/Land dataset is calculated: Then find the nearest grid point index to the extraction point P s (λ s , φ s ): Then, the physical quantity value V i\*,j\* (t k corresponding to the nearest grid point index (i \* ,j \* ) is directly assigned to the extraction point V s (t k )=V i\*,j\* (t k ), wherein the superscript(s) indicates the time series at the extraction point Ps.
6. The method for constructing a groundwater model affected by snow melt replenishment based on CMA-RA/Land data according to claim 1, wherein in the step D, rainfall, evaporation and equivalent water amount of snow melt in the extracted meteorological variable time sequence are compared, groundwater burial depth conditions of a research area are combined, the relationship between groundwater replenishment, runoff and excretion is judged, and source and sink item constitution is determined, wherein if the research area belongs to arid-semiarid regions and the groundwater burial depth is greater than 5m, the source and sink item constitution may not include rainfall.
7. The method for constructing a groundwater model for snow melt replenishment influence based on CMA-RA/Land data according to claim 1, wherein in the step E, groundwater flows of the groundwater model are represented as three-dimensional transient darcy flow equations: , Wherein, V is a Nabula operator, S s is water storage rate in 1/m, H is water head in m, K is osmotic coefficient tensor in m/d, W is source and sink items in unit time including infiltration replenishment, river replenishment, pit boundary drainage and well pumping in 1/d, t is time in d; Wherein the source sink term W per unit time is expressed at grid cells (i, j, k) in the groundwater model as: , Where i, j, k is the grid point number, Q p (i, j, k, t) is the rainfall/snow melt infiltration make-up volume, Q h (i, j, k) is the given head boundary make-up flow, Q w (i, j, k) is the pumped well flow, Q d (i, j, k) is the pit boundary displacement, and DeltaV i,j,k is the grid cell volume.
8. The method for constructing a groundwater model for snow melt replenishment as defined in claim 7, wherein in the step E, setting initial conditions and boundary conditions of the groundwater model comprises: Initial conditions: , Given a head boundary: , water-barrier boundary: , Wherein H 0 is the initial water level of the model, the unit m, the initial water level is set to be the ground elevation for a steady-state model, the initial water level is the result of the last step for a non-steady-state model step-by-step simulation, H river is the water level of a river, the units m, Γ 1 and Γ 2 are Dirichlet and Newman boundaries respectively, and n is the normal unit vector of the boundary.
9. The method for constructing the groundwater model for the snow melting and replenishing effect based on CMA-RA/Land data according to any one of claims 1 to 8, wherein in the E step, the source and sink items determined in the D step are subjected to Z-score standardization treatment to eliminate dimension effects and then serve as input parameters of the groundwater model, and then a numerical model for replacing the groundwater model is constructed by adopting a random forest machine learning method based on a groundwater dynamic simulation result combined by multiple groups of parameters.
10. The method for constructing a groundwater model for snow melt replenishment influence based on CMA-RA/Land data according to claim 9, wherein in the step F, the importance and influence degree analysis is performed on the output characteristic y=f (p 1 , p 2 , p 3 …, p n ) with n characteristics of the numerical model by adopting the SHAP method, the influence degree of each parameter in the numerical model on the predicted result of the water inflow of the pit is quantized according to the analysis, and the numerical model is optimized according to the quantized influence degree, so as to obtain the groundwater model considering the snow melt replenishment influence, and the specific process is as follows: F10, first, the SHAP value of each feature i in the output feature y=f (p 1 , p 2 , p 3 …, p n ) is calculated as: , Where m= {1, 2, 3,..n }, represents a set of all features; The method comprises the steps of selecting a feature subset which does not contain an ith feature, f S (p S ) predicting output of a numerical model based on the feature subset S, f S∪{i} (p S∪{i} ) predicting increment of the numerical model after adding the feature i, namely marginal contribution of the i, and f and p are importance index functions and corresponding feature items respectively; And F20, taking the average absolute value of all samples for each feature I to obtain a global importance index I i : , f30, then normalizing the global importance index I i : , Wherein I i is the global importance of feature I; The method comprises the steps of obtaining a global importance normalization value of a feature i, wherein N is the total number of samples, j is a sample index, phi i,j is the SHAP value of the ith feature in the jth sample, and N is the total number of features; F40, finally, utilizing the global importance index of normalization processing And measuring the average marginal contribution of each feature to the prediction result of the groundwater model in different feature subset combinations, realizing the importance quantitative sequencing of each parameter in the groundwater model, and obtaining the groundwater model considering the influence of snow melting and supplying according to sequencing optimization.

Description

Groundwater model method for constructing snow melting and supplying effects based on CMA-RA/Land data Technical Field The invention belongs to the technical field of mine engineering, and particularly relates to a groundwater model method for constructing snow melting and supplying effects based on CMA-RA/Land data, which is easy to obtain data, has definite physical meaning of parameters and strong certainty of simulation results. Background In the cold and dry mountain-basin area, snow melt replenishment (including glacier water and seasonal snow water) is one of the most important and marked replenishment sources for the groundwater system, and has remarkable seasonal and spatial heterogeneity. For an opencast coal mine and surrounding aquifers thereof, the spring snow melting infiltration determines the recovery process of the groundwater level in the dead water period, and the dynamic change of the water inflow of a pit is directly influenced. Therefore, the accurate depiction of the snow melt replenishing process and the response mechanism thereof to the groundwater system is a key premise for constructing a high-precision and high-reliability groundwater model. In current engineering practice, the estimation of the equivalent supply of snow melt depends on the holiday factor method (device-day method) or the energy balance method (Energy balance method). The method has the advantages that the method is widely applied because of few required parameters and simple and convenient calculation, but the core parameters (such as the holiday factor and the snow melting threshold temperature) are obviously influenced by regional climate, topography and underlying conditions and are difficult to keep stable in space-time, while the method has a more perfect physical mechanism, but is highly dependent on high space-time resolution meteorological elements such as solar radiation, wind speed, humidity and the like, and is often limited by observation data deletion in mining area scale, so that the method is difficult to popularize into actual groundwater modeling scenes. In addition, the existing method generally simplifies snow melting and supplementing into experience infiltration coefficient multiplied by total snow melting amount, ignores complex regulation and control effects of air covering belt thickness, soil freezing-melting process, vegetation coverage and human engineering activities (such as pit drainage and seepage-proof curtain) on infiltration paths and efficiency, and causes strong subjectivity and weak physical foundation of underground water model source and sink item setting. In recent years, with the development of analytical technology, a Land surface analytical data set (CMA-RA/Land) of the chinese weather bureau provides continuous weather and hydrologic variables with high space-time resolution (about 9 km, hour by hour) by fusing multisource observation with a Land surface process model, including key elements such as Snow Water Equivalent (SWE), precipitation, vapor emission, snow melting equivalent water amount, and the like. As the data set is verified by a system in China, the data set has good precision and consistency, and is widely applied to the fields of climate change, watershed hydrologic simulation and the like. However, related work is mostly used for climate change analysis and river basin scale surface hydrologic simulation at present, a technical path for directly and efficiently converting snow melting information in CMA-RA/Land into input parameters of a groundwater model is not available, namely, on one hand, netCDF-format grid data are difficult to accurately dock with a space discrete structure (such as a MODIFLOW grid) or a monitoring point position of the groundwater model, on the other hand, an original variable unit (such as kg/(m 2 s)) is inconsistent with a common unit (such as mm/d) of the groundwater model, and a time scale is required to be converted from an instantaneous/average value to a daily scale, so that a standardized treatment flow is needed. More importantly, when the source and sink items are constructed by the existing groundwater modeling method, experience constants or simplified rainfall infiltration parameterization are mostly adopted, and the 'rainfall-evaporation-snow melting' total element cooperative information provided by CMA-RA/Land cannot be fully utilized, so that the physical meaning of the source and sink items is not clear. Particularly in arid-semiarid mining areas, the underground water has large burial depth and thick air-covering zone, the rainfall infiltration efficiency is extremely low, and the snow melting process is slow and has long duration, so that effective replenishment is easier to form. If the traditional rainfall-dominated replenishment hypothesis is still used, the actual contribution of the melted snow will be severely underestimated, resulting in a predicted deviation of pit water inflow. Meanwhile, the pit exc