CN114812496A - Regional ground settlement early warning method based on multi-source heterogeneous data - Google Patents

Abstract

The invention relates to the technical field of engineering geology/geological disaster monitoring, in particular to a regional ground settlement early warning method based on multisource heterogeneous data, which comprises a data acquisition module, a data transmission module, a resolving module, an analysis module and an early warning module, wherein the data acquisition module is used for identifying and quantifying main control factors of ground settlement and screening and evaluating multisource monitoring technology, the resolving module comprises monitoring data resolving and early warning index system construction, the information acquisition module extracts a remote sensing construction land index IBI based on indexes, firstly, three index wave bands are superposed to generate a new image, the spectral characteristics of each wave band of the generated new image are further analyzed, the maximum values of a construction land, vegetation and a water body are found under NDBI, SAVI and MNDWI conditions respectively, and the construction land index IBI based on the indexes is constructed by utilizing the three index wave bands. The system is stable, low in cost and strong in real-time performance.

Description

A regional land subsidence early warning method based on multi-source heterogeneous data

technical field

The invention relates to the technical field of engineering geology/geological disaster monitoring, in particular to a regional land subsidence early warning method based on multi-source heterogeneous data.

Background technique

Land subsidence, as a kind of "slowly changing geological disaster", has the characteristics of long duration, slow generation, wide influence, complex genetic mechanism and difficult prevention and control. It has become a global environmental geological problem.

With the intensification of land subsidence, its impact on urban planning development and economic construction has gradually become prominent, such as inducing ground fissures, causing damage to buildings (structures), and causing loss of use value of land resources within a certain range; reducing ground elevation , resulting in a decline in the defense capability of urban flood control facilities, resulting in cracking and damage of underground tunnels, affecting the safe operation of subways and increasing maintenance costs.

The current land subsidence early warning only considers surface deformation information such as regional cumulative subsidence and subsidence rate, and cannot reflect the layered deformation characteristics and distribution of land subsidence, and cannot accurately identify the main subsidence layers. The relevant norms of land subsidence in my country lack a clear land subsidence early warning index system, which makes it difficult to achieve accurate early warning for disaster-stricken bodies built at different levels of underground space, which increases the difficulty of implementing land subsidence prevention and control measures. At the same time, the monitoring and early warning facilities are unevenly laid out, and there is a lack of effective unit restrictions, which reduces the early warning accuracy. Therefore, it is urgent to establish a stratified and regional early warning index system for land subsidence to carry out high-precision monitoring and early warning of land subsidence.

SUMMARY OF THE INVENTION

In view of this, the purpose of the present invention is to provide a regional land subsidence early warning method based on multi-source heterogeneous data to solve the problems mentioned in the background art.

The present invention is a regional land subsidence early warning method based on multi-source heterogeneous data. Based on each index factor in the land subsidence early warning index system, the land subsidence unit is divided; Multi-type monitoring methods include: precision leveling, GNSS and InSAR; classify and optimize multi-type monitoring methods, and establish a land subsidence information calculation model; obtain the data to be monitored, and calculate each land subsidence through the land subsidence information calculation model. The layered settlement value in the unit; according to the calculated layered settlement value, the risk level of land subsidence is determined, and the early warning is carried out. This method is composed of an evaluation system and a module. The evaluation system of this method includes identification and quantification of main controlling factors of land subsidence, screening and evaluation of multi-source monitoring technology, calculation of monitoring data, and construction of early warning index system. Specific modules include data acquisition module, data transmission module, solution module, analysis module and early warning module

Further, the land subsidence early warning index system includes: target layer, standard and criterion layer, and index layer.

The index layer includes: land subsidence zonal index and land subsidence layered index. The stratified index of land subsidence reflects: the vertical development depth of land subsidence, the main contribution layer and sensitive layer of land subsidence, the land subsidence index reflecting the characteristics of surface deformation, and the degree, depth and scale of underground space development and utilization. The land subsidence information calculation model includes: leveling point correction model, deformation projection transformation model, and groundwater-land subsidence coupling model.

Before dividing the land subsidence unit based on each index factor in the land subsidence early warning index system, the land subsidence early warning index system is established, and the weight of each index is obtained through PCA. Design the framework structure of the land subsidence early warning index system, and carry out the index system screening based on the research results of regional land subsidence. Before obtaining the data to be monitored and calculating the layered subsidence value in each land subsidence unit through the land subsidence information solution model, after classifying and optimizing the multi-type monitoring methods and establishing the land subsidence information solution model, design the land subsidence The framework structure of the early warning index system, based on the research results of regional land subsidence, carry out the selection of the index system. Combined with the classification standard of land subsidence early warning, the risk level of land subsidence is determined, and after early warning, cloud computing, big data, and WebGIS technologies are comprehensively applied to realize the display of early warning results.

Further, the main controlling factors of land subsidence are identified and quantified, starting from the influencing factors of land subsidence, and selecting the main controlling factors of land subsidence from the perspectives of natural factors and human factors. Natural factors mainly include: basement tectonic movement, stratum lithology and structural characteristics; man-made factors mainly include: groundwater level variation, static loads of tall buildings, dense buildings, and dynamic loads of linear work such as rail transit. Based on statistical analysis methods, the relationship between different factors and land subsidence responses was obtained.

Further, the amplitude data of groundwater level is obtained by the water level measuring instrument placed in the groundwater monitoring well; the static load of tall buildings and dense building groups: the method based on the remote sensing building land index is used for reference, and the remote sensing images covering the study area are selected and used in NDBI and MNDWI. , SAVI index inversion, the Erdas Modeler tool is further used to obtain the spatial and temporal evolution information of building loads. Remote sensing image preprocessing includes: geometric correction, radiometric calibration, atmospheric correction, and image masking. Then the vegetation index SAVI extraction formula is as formula (1):

SAVI=[(NIR-Red)(1+L)](NIR+Red+L) Formula (1)

In the formula, NIR is the pixel brightness value in the near-infrared band, Red is the pixel brightness value in the red band, and L is the soil adjustment factor, which ranges from 0 to 1. When L=0, it means that the vegetation coverage is zero; when L=1, it means that the influence of the soil background is zero, that is, the vegetation coverage is the highest, and the influence of the soil background is zero. Only in tree-covered areas. In urban areas, Huete recommends L=0.5, which can reduce the influence of the soil background. In this paper, the L value is taken as 0.5 to calculate the soil-adjusted vegetation index SAVI in Beijing.

Modified normalized water body index MNDWI extraction:

MNDWI=(Green-MIR)/(Green+MIR) Formula (2)

In the formula, Green represents the pixel brightness value in the green band, and MIR is the pixel brightness value in the mid-infrared band.

Normalized building land index NDBI extraction:

NDBI=(MIR-NIR)/(MIR+NIR) Formula (3)

In the formula, MIR is the gray value of the pixel in the mid-infrared band, NIR is the brightness value of the pixel in the near-infrared band, and NDBI is -1 to 1. The spectral characteristics of building land and dry land show that the average value of the mid-infrared band is greater than that of the near-infrared band, while the water body, forest land and farmland show the opposite characteristics. Where the NDBI value is greater than 0 is urban land, and where the NDBI is less than 0 is non-building land.

Further, based on the index IBI extraction of the remote sensing building land index, firstly, the three index bands are superimposed to generate a new image, and the spectral characteristics of each band of the generated new image are further analyzed. to get the maximum value. Therefore, the three index bands are used to construct the index-based building land index IBI, and the formula is as formula (4);

IBI=[NDBI-(SAVI+MNDWI)2]/[NDBI+(SAVI+MNDWI)/2] Formula (4)

Dynamic load of linear work such as rail transit: select the average static load of rail transit engineering, and the calculation formula is as follows:

In the formula, P _E : static load of rail traffic; V _E vehicle running speed. _PES : vehicle average static load; n: vehicle number; Mi: contribution rate; Pi vehicle average static load PVS linear engineering average static load.

Statistical analysis methods mainly include: SRCC, random forest, and the following formula is used to calculate the above data, such as formula (6)

Among them: Pk indicates that the proportion of the k-th type of samples in the current sample set D is Pk.

Further, in the screening and evaluation of multi-source monitoring technologies, from the perspective of spatial measurement technology, the three-dimensional monitoring technology of "sky and ground" land subsidence is selected. The main technical means include: InSAR, GPS, leveling, layered markers, and distributed optical fibers. From the perspectives of spatial resolution, temporal resolution and monitoring accuracy, each source monitoring technology was evaluated, and monitoring methods suitable for the frequency and accuracy of land subsidence early warning were screened out, which provided technical support for the subsequent development of land subsidence monitoring and early warning.

Further, monitor data calculation and information collection, establish a data calculation framework for each source after screening, and build a monitoring data-land subsidence intuitive response data calculation platform to obtain high-precision calculation data in real time. InSAR, GNSS, and distributed optical fiber are selected as the dominant monitoring methods for land subsidence monitoring and early warning. InSAR data has high spatial resolution, which can obtain regional land subsidence deformation information and control the trend of land subsidence deformation. GNSS can acquire ground deformation information in real time, and can be used to carry out long-term monitoring of the deformation characteristics of the subsidence center, and to carry out early warning of the land subsidence deformation of the subsidence center. Distributed optical fiber has the advantage of combining transmission and measurement, and can obtain soil strain information in the full monitoring depth in real time, which can be used for subsidence central area and subsidence edge area, layered subsidence monitoring, and realizes layered monitoring and early warning of land subsidence.

Further, the InSAR solution mainly adopts: Small Baseline Interferometry (SBAS-InSAR) technology, and the formula can be expressed as Equation (7):

where ν( ) and β( ) are the average velocity and nonlinear components in the residual deformation of the high-resolution single-level differential interferogram, respectively, and ΔZ(x, r) is the high-resolution single-level differential interferogram The terrain component in , Δn(x, r) is the noise error, and for the estimation of ν( ) and ΔZ(x, r), they must satisfy the maximum time coherence factor, which is expressed as formula (8):

Among them, _δφmo is the analog phase, which is expressed as follows, such as formula (9):

Subtract formula (8) and formula (9) to obtain a new phase, including β(·) and Δm(x, r). Using singular value decomposition method, the nonlinear deformation rate β(·) is removed, and the overall shape is The variable can be expressed as equation (10): d( _tn ,x,r) _=dL (tn,x,r)+( _{tn- t0} )v(x,r)+β( _tn ,x, r), n=0, 1, 2...N formula (10)

Further, the GNSS solution adopts the GAMIT baseline solution:

式(11)、(12)中f₁为L₁的载波频率；f₂为L₂的载波频率；

为卫星到接收机间的几何距离；

为电离层延迟；δT_i ^p为对流层延迟；δt_i为接收机钟差；δt^p为卫星钟差；

为初始整周模糊度；

为残差。GAMIT uses the double-difference method to process the original observations. Double-difference observations can completely eliminate the influence of satellite clock errors and receiver clock errors, and can also significantly weaken the influence of systematic errors such as orbital errors and atmospheric refraction. Assuming that satellite p is observed at station i at time t, the linearized dual-frequency carrier phase observation equation is:

In formulas (11) and (12), f ₁ is the carrier frequency of L ₁ ; f ₂ is the carrier frequency of L ₂ ;

is the geometric distance between the satellite and the receiver;

is the ionospheric delay; δT _i ^p is the tropospheric delay; δt _i is the receiver clock error; δt ^p is the satellite clock error;

is the initial integer ambiguity;

is the residual.

Assuming that satellites p and q are observed at stations i and j at time t, the linearized double-difference carrier phase observation equation is equation (13):

In formula (13), the process delay can be weakened by parameter estimation or model correction; the ionospheric refraction is affected by various factors and it is difficult to use a specific method to deal with it. At present, dual-frequency phase observations are often used to eliminate the ionosphere. The combined LC weakens the influence of the first-order ionospheric refraction, as shown in equation (14).

In Equation (14), the ionospheric influence of the LC observations is eliminated after the double-difference combination, but the ambiguity of the LC observations no longer has integer characteristics. In order to accurately fix the integer ambiguity of the LC observations, the wide The LC ambiguity is decomposed by the combined observations of lane WL and narrow lane NL.

The strain amount and the Brillouin frequency shift of the fiber can be expressed by the following formula:

为比例系数，约为493MHz(/％strain)；ε为光纤的应变量。Among them, v _B (ε) is the drift of the Brillouin frequency when the strain is ε; v _B (0) is the drift of the Brillouin frequency when the strain is 0;

is the proportional coefficient, about 493MHz (/%strain); ε is the strain amount of the optical fiber.

Further, the construction of the early warning index system and the determination of the threshold value, according to the identification and quantification results of the main control factors of land subsidence, determine the various evaluation factors and their weights required for the construction of the land subsidence early warning index system, and use cluster analysis and fuzzy mathematics methods. The single-factor evaluation index map is coupled and superimposed to establish a land subsidence early warning evaluation index system.

Further, a land subsidence monitoring and early warning device, comprising one or more processors; a storage device for storing one or more programs, when the one or more programs are executed by the one or more processors, so that The one or more processors implement the method for early warning of regional land subsidence based on multi-source heterogeneous data according to any one of the above.

Based on the comprehensive unit of land subsidence, an early warning standard is established. The early warning standard integrates the “National Land Subsidence Prevention and Control Plan”, the “Regional Land Subsidence Prevention and Control Plan”, the threshold of each building/linear engineering deformation caused by land subsidence, and the “regional land subsidence disaster risk assessment factor level”, and comprehensively compares various thresholds. Yes, divide the warning levels from small to large. Regional early warning indicators

1) Groundwater level

The annual prevention and control indicators of land subsidence are decomposed into quarters, months and days according to the development law of the year, and the compression amount of key subsidence strata and the groundwater decline of adjacent aquifers that cause subsidence form a clear alternative model on the basis of continuous iteration. The subsidence value is decomposed in detail in the vertical space, and the groundwater level is used as the main early warning threshold and index, and the real-time monitoring and early warning is carried out with the results of the dynamic monitoring data of the groundwater level.

2) Surface deformation

The annual prevention and control indicators of land subsidence are decomposed into quarters, months, and days according to the development law of the year, and the monitoring results are implemented by using GPS continuous stations in typical areas. Compared with the prevention and control indicators, real-time monitoring and early warning are carried out.

(3) Classification of early warning levels

By constructing a dual early-warning index system that combines the subsidence index represented by the dynamic cumulative subsidence value during the year and the groundwater level index, the early-warning and forecasting is realized. The early warning risk prevention and control levels are divided into three levels according to the size of the risk. Each level corresponds to the corresponding subsidence and groundwater level indicators. The first level is the highest, the display color is red, the second level is the second, the display color is orange, and the third level is the lowest. Display color is yellow.

The present invention proposes a ground subsidence monitoring and early warning system composed of an evaluation system and modules, which can acquire high-precision surface and underground three-dimensional monitoring data in real time, and realize monitoring and early warning within the scale of the entire ground subsidence. It solves the defects of low precision and low spatial resolution of traditional land subsidence model early warning, which can only carry out regional (small scale) land subsidence early warning, and cannot achieve small regional scale land subsidence early warning. The system integrates multiple processes such as multi-source monitoring technology screening and real-time processing of information data, identification and quantification of main control factors, early warning indicator system, information system, results expression form and real-time release, etc. It can carry out real-time, one-stop early warning of land subsidence disasters, simplify the intermediate process, and be more conducive to promotion and use. In view of the defect that traditional early warning methods can only perform early warning based on surface deformation information, and cannot achieve stratified and regional early warning, the design is based on the identification of the main controlling factors of land subsidence, and a multi-source land subsidence monitoring technology framework system is constructed to obtain and real-time subsidence. To overcome the limitations of existing monitoring technologies, establish a land subsidence early warning evaluation index system based on regional and key project land subsidence evaluation standards, and comprehensively apply cloud computing, big data, and WebGIS technologies to realize land subsidence monitoring and early warning.

Detailed ways

The following are specific embodiments to describe the present invention in detail. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, those of ordinary skill in the art will All other embodiments obtained under the premise of creative work fall within the scope of protection of this application.

It consists of an evaluation system and modules. The evaluation system includes: identification and quantification of main controlling factors of land subsidence, screening and evaluation of multi-source monitoring technology, calculation of monitoring data, and construction of early warning index system. Modules include: data acquisition module, data transmission module, solution module, analysis module and early warning module

In this embodiment, the main control factors of land subsidence are identified and quantified, and the main control factors of land subsidence are selected from the perspectives of natural factors and human factors, starting from the influencing factors of land subsidence. Natural factors mainly include: basement tectonic movement, stratum lithology and structural characteristics; man-made factors mainly include: groundwater level variation, static loads of tall buildings, dense buildings, and dynamic loads of linear work such as rail transit. Based on statistical analysis methods, the relationship between different factors and land subsidence responses was obtained.

In this embodiment, the groundwater level fluctuation data is obtained by the water level measuring instrument placed in the groundwater monitoring well; the static load of tall buildings and dense building groups: refer to the method based on the remote sensing building land index, select the remote sensing image covering the research area, On the basis of NDBI, MNDWI, and SAVI index inversion, Erdas Modeler tool is further used to obtain the spatial and temporal evolution information of building loads. Remote sensing image preprocessing includes: geometric correction, radiometric calibration, atmospheric correction, and image masking. Then the vegetation index SAVI extraction formula is as formula (1):

SAVI=[(NIR-Red)(1+L)]/(NIR+Red+L) Formula (1)

In the formula, NIR is the pixel brightness value in the near-infrared band, Red is the pixel brightness value in the red band, and L is the soil adjustment factor, which ranges from 0 to 1. When L=0, it means that the vegetation coverage is zero; when L=1, it means that the influence of the soil background is zero, that is, the vegetation coverage is the highest, and the influence of the soil background is zero. Only in tree-covered areas. In urban areas, Huete recommends L=0.5, which can reduce the influence of the soil background. In this paper, the L value is taken as 0.5 to calculate the soil-adjusted vegetation index SAVI in Beijing.

Modified normalized water body index MNDWI extraction:

MNDWI=(Green-MIR)/(Green+MIR) Formula (2)

In the formula, Green represents the pixel brightness value in the green band, and MIR is the pixel brightness value in the mid-infrared band.

Normalized building land index NDBI extraction:

NDBI=(MIR-NIR)/(MIR+NIR) Formula (3)

In the formula, MIR is the gray value of the pixel in the mid-infrared band, NIR is the brightness value of the pixel in the near-infrared band, and NDBI is -1 to 1. The spectral characteristics of building land and dry land show that the average value of the mid-infrared band is greater than that of the near-infrared band, while the water body, forest land and farmland show the opposite characteristics. Where the NDBI value is greater than 0 is urban land, and where the NDBI is less than 0 is non-building land.

In this embodiment, the index-based IBI extraction of the remote sensing building land index firstly superimposes the three index bands to generate a new image, and further analyzes the spectral characteristics of each band of the generated new image. and MNDWI conditions to obtain the maximum value. Therefore, using these three index bands to construct the index-based building land index IBI, the formula is as formula (4):

IBI=[NDBI-(SAVI+MNDWI)/2]/[NDBI+(SAVI+MNDWI)/2] Formula (4)

Dynamic load of linear work such as rail transit: select the average static load of rail transit engineering, and the calculation formula is as follows:

In the formula, P _E : static load of rail traffic; V _E vehicle running speed. _PES : vehicle average static load; n: vehicle number; Mi: contribution rate; Pi vehicle average static load PVS linear engineering average static load.

Statistical analysis methods mainly include: SRCC, random forest, and the following formula is used to calculate the above data, such as formula (6)

Among them: Pk indicates that the proportion of the k-th type of samples in the current sample set D is Pk.

In this embodiment, the multi-source monitoring technology is screened and evaluated. From the perspective of space measurement technology, the three-dimensional monitoring technology of "sky and earth" land subsidence is selected. The main technical means include: InSAR, GPS, leveling, layered marking, distributed optical fiber . From the perspectives of spatial resolution, temporal resolution and monitoring accuracy, each source monitoring technology was evaluated, and monitoring methods suitable for the frequency and accuracy of land subsidence early warning were screened out, which provided technical support for the subsequent development of land subsidence monitoring and early warning.

In this embodiment, the monitoring data calculation and information collection are performed, the screened source data calculation framework is established, and the monitoring data-land subsidence intuitive response data calculation platform is established, so as to obtain high-precision calculation data in real time. InSAR, GNSS, and distributed optical fiber are selected as the dominant monitoring methods for land subsidence monitoring and early warning. InSAR data has high spatial resolution, which can obtain regional land subsidence deformation information and control the trend of land subsidence deformation. GNSS can obtain ground deformation information in real time, and can be used to carry out long-term monitoring of the deformation characteristics of the subsidence center, and to carry out early warning of the land subsidence deformation of the subsidence center. Distributed optical fiber has the advantage of combining transmission and measurement, and can obtain soil strain information in the full monitoring depth in real time, which can be used for subsidence central area and subsidence edge area, layered subsidence monitoring, and realizes layered monitoring and early warning of land subsidence.

InSAR solution mainly adopts: Small Baseline Interferometry (SBAS-InSAR) technology, the formula can be expressed as Equation (7):

where ν( ) and β( ) are the average velocity and nonlinear components in the residual deformation of the high-resolution single-level differential interferogram, respectively, and ΔZ(x, r) is the high-resolution single-level differential interferogram The terrain component in , Δn(x, r) is the noise error, and for the estimation of ν( ) and ΔZ(x, r), they must satisfy the maximum time coherence factor, which is expressed as formula (8):

Among them, _δφmo is the analog phase, which is expressed as follows, such as formula (9):

Subtract formula (8) and formula (9) to obtain a new phase, including β(·) and Δm(x, r). Using singular value decomposition method, the nonlinear deformation rate β(·) is removed, and the overall shape is The variable can be expressed as equation (10): d(t _n , x, r)=d _L (t _n , x, r)+(t _n -t ₀ )ν(x,r)+β(t _n ,x , r), n=0, 1, 2...N Formula (10)

In this embodiment, the GNSS calculation adopts the GAMIT baseline calculation:

式(11)、(12)中f₁为L₁的载波频率；f₂为L₂的载波频率；

为卫星到接收机间的几何距离；

为电离层延迟；δT_i ^p为对流层延迟；δt_i为接收机钟差；δt^p为卫星钟差；

为初始整周模糊度；

为残差。GAMIT uses the double-difference method to process the original observations. Double-difference observations can completely eliminate the influence of satellite clock errors and receiver clock errors, and can also significantly weaken the influence of systematic errors such as orbital errors and atmospheric refraction. Assuming that satellite p is observed at station i at time t, the linearized dual-frequency carrier phase observation equation is:

In formulas (11) and (12), f ₁ is the carrier frequency of L ₁ ; f ₂ is the carrier frequency of L ₂ ;

is the geometric distance between the satellite and the receiver;

is the ionospheric delay; δT _i ^p is the tropospheric delay; δt _i is the receiver clock error; δt ^p is the satellite clock error;

is the initial integer ambiguity;

is the residual.

Assuming that satellites p and q are observed at stations i and j at time t, the linearized double-difference carrier phase observation equation is equation (13):

In formula (13), the process delay can be weakened by parameter estimation or model correction; the ionospheric refraction is affected by various factors and it is difficult to use a specific method to deal with it. At present, dual-frequency phase observations are often used to eliminate the ionosphere. The combined LC weakens the influence of the first-order ionospheric refraction, as shown in equation (14).

In Equation (14), the ionospheric influence of the LC observations is eliminated after the double-difference combination, but the ambiguity of the LC observations no longer has integer characteristics. In order to accurately fix the integer ambiguity of the LC observations, the wide The LC ambiguity is decomposed by the combined observations of lane WL and narrow lane NL.

The strain amount and the Brillouin frequency shift of the fiber can be expressed by the following formula:

where v _B (ε) is the shift of the Brillouin frequency when the strain is ε;

为比例系数，约为 493MHz(/％strain)；ε为光纤的应变量。v _B (0) is the drift of the Brillouin frequency when the strain is 0;

is the proportional coefficient, about 493MHz (/%strain); ε is the strain amount of the optical fiber.

In this embodiment, the early warning index system is constructed and the threshold is determined. According to the identification and quantification results of the main control factors of land subsidence, various evaluation factors and their weights required for the construction of the land subsidence early warning index system are determined. Cluster analysis and fuzzy mathematics methods are used. Coupling and superposition analysis of various single-factor evaluation index maps is carried out to establish a land subsidence early warning evaluation index system.

In this embodiment, a land subsidence monitoring and early warning device includes one or more processors; and a storage device for storing one or more programs, when the one or more programs are executed by the one or more processors The execution causes the one or more processors to implement the method for early warning of regional land subsidence based on multi-source heterogeneous data according to any one of the above.

In the second embodiment of the present invention, a method for early warning of regional land subsidence based on multi-source heterogeneous data includes:

Design the framework structure of the land subsidence early warning index system, and carry out the index system screening based on the research results of regional land subsidence. The framework structure includes three parts, the target layer, the standard and criterion layer, and the index layer.

Among them, the target layer is to realize the monitoring and early warning of land subsidence, and the standard and guideline layer. The standard is to follow the current relevant norms and standards related to the monitoring and evaluation of land subsidence. Each index reflects the consolidation and compression of underground loose rock layers caused by natural factors or human engineering activities. And lead to the slow-moving geological disasters that reduce the ground elevation in a certain area. In the process of index selection, (1) the selection index is expressed in numerical form (quantitative); (2) it highlights the characteristics of land subsidence and deformation; (3) each index is irreplaceable and unique. The indicator layer is a variety of indicators that reflect the temporal and spatial variation characteristics of land subsidence and support the early warning of land subsidence.

The index layer consists of two parts: land subsidence zoning and land subsidence stratification. The selection of land subsidence zoning indicators should highlight the following four parts: (1) The natural factors that affect the development of land subsidence mainly include: geological structure, the thickness of the compressible layer, and the characteristics of the aquifer; (2) The human factors that affect the development of land subsidence mainly include: Underground resource mining level, underground resource mining volume, building static load and vehicle dynamic load; (3) urban planning and major projects, lifeline projects, etc.; (4) existing land subsidence disaster areas. The selection of the stratified index of land subsidence should highlight the following four parts: (1) the vertical development depth of land subsidence; (2) the main contribution layer and sensitive layer of land subsidence; (3) the land subsidence index reflecting the characteristics of surface deformation; (4) ) The degree, depth and scale of underground space development and utilization. Establish a land subsidence early warning indicator system and obtain the weights of each indicator.

The early warning index system includes land subsidence zonal index and land subsidence stratified index. This step mainly adopts statistical analysis methods to comprehensively determine the early warning indicator system. The specific determination steps are as follows:

Based on the research results of urban planning and land subsidence mechanism, indicators that can fully reflect the characteristics of regional land subsidence are selected and distributed to experts in relevant fields. Different indicators are scored from multiple perspectives such as geological background and structural damage. Very important (10-8 points), relatively important (8-6 points), generally important (6-4 points), not important (4-2 points). Remove the unimportant parts, quantify the indicators selected from the remaining three levels, and analyze the relationship between each indicator and land subsidence (including the impact on the development of land subsidence and the degree of influence by land subsidence) based on statistical principles. Analysis methods Mainly through the correlation analysis method combined with the principal component analysis (PCA), the comprehensive screening of each index was carried out again. Finally, it is selected as the monitoring and early warning index of land subsidence, and the monitoring and early warning index system of land subsidence is established.

Based on each index factor, the land subsidence unit is divided.

Based on the screening results and weights of the zoning indicators, the index factors representing natural factors are selected, and the factors are superimposed and analyzed by the linear weighted synthesis method (equation 1) to obtain the ground foundation subsidence unit. This unit mainly reflects the ground subsidence foundation pattern, and the results are superimposed. Human factors and land subsidence disaster index factors are used to obtain the dynamic unit of land subsidence, which represents the severity and development range of land subsidence. Finally, the index factors reflecting the deformation characteristics of major projects are superimposed to obtain a comprehensive unit of land subsidence, which represents the impact of land subsidence on urban construction, as shown in Equation (16);

In the formula: y——comprehensive evaluation value; _xj ——evaluation index; _wj ——evaluation index; corresponding weight coefficient of _xj ; n——the number of evaluation index.

In this embodiment, in each ground subsidence unit, multiple types of monitoring means are set up, and each monitoring means can obtain surface deformation information or vertical deformation characteristics of the formation.

Based on the results of the comprehensive unit of land subsidence, and according to the internal main control impact indicators/receptors, monitoring methods or means that can effectively reflect the characteristics of each indicator are selected. The main means include distributed optical fiber, CMT multi-level monitoring wells, GNSS, InSAR and other monitoring technologies that can obtain real-time data information.

In this embodiment, various types of monitoring means are classified, integrated and optimized, and a land subsidence information solution model is established.

According to different monitoring scales, they are generally classified into surface monitoring systems and vertical monitoring systems. Surface monitoring systems include regional monitoring methods and small-scale monitoring methods for key projects. For different projects, the monitoring methods are slightly different, and refer to the monitoring standards for different engineering methods. specification. At present, regional monitoring methods mainly include precision leveling, GNSS and InSAR. Among them, precision leveling obtains surface deformation information by laying out a multi-level leveling network, and obtaining surface deformation information through adjustment calculation and spatial interpolation. This method cannot meet the requirements of real-time dynamic monitoring of land subsidence. Requirements, precision leveling is usually used to verify the accuracy of land subsidence monitoring technology due to its high precision. GNSS measurement technology has the advantages of short cycle, high positioning accuracy, rapid network deployment, and all-weather. However, what GNSS measurement obtains is the point-like distribution of ground monitoring point deformation information, so this method is mainly used for ground subsidence information monitoring in subsidence center areas or areas with severe differential subsidence. Synthetic aperture radar differential interferometry technology can be real-time, fast, large-scale and high-precision, and the monitoring accuracy of obtaining vertical deformation information can reach mm level, which can be used for real-time monitoring of regional settlement. However, in terms of horizontal deformation monitoring, its detection ability is limited and it is not sensitive to horizontal deformation. In addition, the phase unwrapping is seriously affected by atmospheric delay and space-time de-correlation, so it is necessary to eliminate the influence of these errors in the solution. Based on the above analysis, the comprehensive multi-interpretation method of InSAR data is selected as the main method for real-time dynamic monitoring of regional land subsidence, and the results are corrected by using GNSS and precision leveling results.

Benchmark Correction Model

Using the spatial projection transformation model, the precision leveling point and the PS point are corrected to the same coordinate system. Taking the leveling point as the center of the circle, the cumulative settlement of the PS point within a range of 100m is averaged, and the measured value of the precision leveling point is compared. Calculate the correlation coefficient and root mean square error. Specific steps are as follows:

Convert the PS point line-of-sight deformation variable into the vertical direction deformation variable by formula a.

Calculate the error value (E _vari ) between the level point and the transformed PS deformation variable, which includes orbital error, atmospheric delay, and random error.

Atmospheric delay and terrain-induced errors (E _s ) are calculated using a systematic error model, and regional error estimates (E _a ) are calculated in combination with variograms.

Use formula b to perform PS point correction.

V _v =V _los /cosθ Equation (17)

E _vair =V _v -V _b formula (18)

E _vair =E _s +E _a Formula (19)

V _verif =V _v -E _vair formula (20)

In the above formula, V _v is the deformation amount in the vertical direction of the PS value, V _Los is the deformation amount in the LOS direction of the PS value, ΔH is the displacement of the ground surface in the vertical direction, and θ is the imaging side view of the SAR satellite.

In this embodiment, the deformation projection conversion model of GNSS and InSAR

(1) GNSS data projection transformation model

Suppose the actual deformation vector of the surface is V, the deformation components along the east, north and vertical directions are (ΔE, ΔN, ΔH), and the horizontal deformation components in the east and north directions obtained by GNSS observation are (ΔE _GNSS , ΔN _GNSS ), let V _GPS =(ΔE _GNSS , ΔN _GNSS ). Assuming that the included angle between the SAR satellite track direction (SAR image azimuth) and the north direction is Φ, the GNSS horizontal displacement (ΔE _GNSS , ΔN _GNSS ) is converted to its range and azimuth in the SAR image by means of the Φ angle. The deformation projection component (ΔR, ΔA) to (Azimuth), let V _SAR = (ΔR, ΔA), the conversion relationship between the two is:

The azimuth direction is perpendicular to the range direction and the LOS direction, so the deformation variable ΔA in the azimuth direction is projected to be zero in the LOS direction, and the angle between the deformation variable ΔR in the range direction and the LOS direction is θ, so ΔR The deformation projection component in the LOS direction is (ΔE _GNSS cosΦ+ΔN _GNSS sinΦ) sinθ, the deformation component of the vertical deformation ΔH in the LOS direction is ΔH cosθ, the sum of the two is the projection of the three-dimensional surface deformation to the radar Line of sight direction (LOS deformation Δr, therefore:

Δr=ΔH·cosθ+(ΔE _GNSS cosΦ+ΔN _GNSS sinΦ)·sinθ Equation (22)

In the above formula, Δr is the deformation amount in the LOS direction measured by InSAR, ΔH is the displacement of the ground surface in the vertical direction, and θ is the imaging side view of the SAR satellite. Rewrite the above formula into the following form:

The above formula establishes the conversion relationship between the LOS deformation of the radar and the deformation in the vertical direction of the surface. It can be seen from this formula that if the horizontal displacement of the ground surface (ΔE _GNSS , ΔN _GNSS ) and the included angles Φ and θ calculated based on the satellite orbit parameters have been measured, using (3), the InSAR can be placed in the radar line-of-sight direction ( LOS) The deformation amount Δr is converted into the deformation amount ΔH of the ground surface in the vertical direction.

In this embodiment, the atmospheric delay mean model

①Using the GNSS calculation results to calculate the high-precision tropospheric zenith total delay (ZTD) during the transit time of the SAR satellite. The zenith static delay (ZHD) is calculated using the surface meteorological data, and then the zenith wet delay (ZWD) is obtained by ZTD-ZHD. The zenith wet delay GNSS (ZWD) calculated by GNSS combined with ground meteorological data and the zenith wet delay MODIS (ZWD) obtained by inversion of MODIS data are used for regression fitting to realize the correction of MODIS (ZWD) by GPS (ZWD).

为GPS站点纬度，H为GPS站点高程值。where P _s is the surface atmospheric pressure,

is the latitude of the GPS site, and H is the elevation value of the GPS site.

② Calculate the differential atmospheric delay phase at the time of acquisition of the main and auxiliary SAR images. The formula is:

进行低通滤波。where θ _inc is the radar incident angle. In order to reduce the influence of noise and operating errors, it is necessary to

Perform low-pass filtering.

与GPS/MODIS数据联合反演得到的大气延迟相位

做均值融合处理，建立高精度、高时空分辨率的大气延迟均值模型。Atmospheric delay phase separated from PS-InSAR residual phase

Atmospheric delay phase obtained by joint inversion with GPS/MODIS data

Perform mean fusion processing to establish an atmospheric delay mean model with high precision and high spatial and temporal resolution.

(3) Atmospheric delay correction model

相位图与初始差分干涉相位图进行配准，达到像元尺度的统一，才能进行栅格计算，求取消除大气延迟影响的高精度差分干涉相位图。Since the removal of the atmospheric delay phase is carried out for the pixel, it is also necessary to

The phase map is registered with the initial differential interferometric phase map to achieve the unity of the pixel scale, and then the grid calculation can be performed to obtain a high-precision differential interferometric phase map that eliminates the influence of atmospheric delay.

为大气延迟改正后的差分干涉相位图，

为初始差分干涉相位图。In the formula,

is the differential interferometric phase diagram after atmospheric delay correction,

is the initial differential interference phase map.

The vertical monitoring system currently mainly includes bedrock-layered markers, distributed optical fibers and groundwater monitoring wells. The bedrock marker-layered marker monitoring method realizes the acquisition of high-precision vertical layered ground subsidence deformation information by placing the benchmarks in different subsidence layers, and its accuracy reaches 0.01-0.1mm. However, due to the complex operation, large area, high construction technology, and high cost, this method is only used for the verification of the ground subsidence layered monitoring model and the study of the target land subsidence mechanism. Distributed optical fiber technology refers to the technical method of laying sensing optical cables in the object under test to realize the continuity test of multi-physical parameters in one-dimensional direction. Brillouin scattering can reflect the characteristics of optical fiber strain. It is calculated by formula that the construction process of this method is relatively simple, and multi-level continuous settlement information can be obtained. It is widely used, but its construction cost is relatively high, and it can be used for settlement center/level Central stratified monitoring.

分别表示与光纤类型有关的应变和温度的比例系数。In the formula, υ _B (ε, T) is the drift of the fiber Brillouin frequency when the ambient temperature is T and the strain is ε; υ _B (0, T ₀ ) is the fiber cloth when the temperature is T ₀ and the strain is 0 The amount of drift in the rilloin frequency;

are the scaling factors for strain and temperature, respectively, related to the fiber type.

FBGs reflect light of a specific wavelength that satisfies the following conditions:

λ _B =2n _eff Λ Equation (30)

In the formula, λ _B is the center wavelength of the reflected light; n _eff is the effective refractive index of the fiber core; Λ is the spatial period of the refractive index modulation of the fiber grating.

External stress and temperature changes will cause changes in refractive index and grating pitch, resulting in the shift of FBG wavelength λ _B , which satisfies the linear relationship:

where Δλ is the wavelength change of the FBG, ε is the axial strain of the fiber, ΔT is the temperature change, P _s is the photoelastic coefficient of the fiber, α is the thermal expansion coefficient of the fiber, and ζ is the thermo-optic coefficient of the fiber.

Groundwater overexploitation is the main influencing factor in the severely developed areas of land subsidence in my country. The decrease of groundwater level leads to dissipation of pore water pressure, and the increase of effective stress leads to subsidence of compressed soil layers. Therefore, the amplitude of groundwater level variation can indicate the subsidence of the aquifer system.

Layered monitoring of land subsidence is realized through groundwater-land subsidence coupled model or empirical model. Its formula is as follows:

where: Ω—seepage area; h—water level elevation of groundwater; K—permeability coefficient; w—source-sink term of aquifer; _{h0 —} initial water level; _Ss —water storage rate; ₂ —the second type of boundary; K _n —the permeability coefficient in the normal direction of the second type of boundary interface, n—the normal direction of the second type of boundary interface; Δb—the amount of deformation; b—the thickness of the soil layer; S _sk —the skeleton water storage (when the water level is lower than the previous minimum water level, the parameter is S _skv —the inelastic skeleton water storage rate, when the water level is higher than the previous minimum water level, the parameter is S _skv — the elastic skeleton water storage rate).

Where: S _sk and S _skv are elastic water storage rate and inelastic water storage rate, respectively:

where: C _c compression index; C _r rebound index; γ _w water bulk density; σ effective stress; e ₀ initial void ratio.

In this embodiment, based on the multi-objective land subsidence evaluation standard, a land subsidence early warning classification standard is established in each land subsidence unit.

Based on the comprehensive unit of land subsidence, the internal early warning standard of each unit is established. The early warning standard integrates the “National Land Subsidence Prevention and Control Plan”, the “Regional Land Subsidence Prevention and Control Plan”, the threshold of each building/linear engineering deformation caused by land subsidence in the unit, and the “regional land subsidence disaster risk assessment factor level”. Comprehensive comparison, according to the thresholds of receptors in different subsidence layers, the early warning levels are divided from small to large. The data to be monitored is obtained, and the layered settlement value in each unit is calculated by solving the model. Combined with the grading standards, determine the risk level of land subsidence and carry out early warning. Comprehensive application of cloud computing, big data, and WebGIS technologies to realize the display of early warning results.

For example, the ground subsidence monitoring and early warning device can be used as a computing device for the downward extension process. As described in this paper, the ground subsidence monitoring and early warning device can be used to implement the function of downward extension calculation of gravity or magnetic field data in a computing device. The land subsidence monitoring and early warning device may be implemented in a single node, or the functions of the land subsidence monitoring and early warning device may be implemented in multiple nodes in the network. Those skilled in the art should realize that the term land subsidence monitoring and early warning device includes equipment in a broad sense, of which the land subsidence monitoring and early warning device is only one example. The ground subsidence monitoring and early warning device is included for clarity, and is not intended to limit the application of the present invention to a specific embodiment of the ground subsidence monitoring and early warning device or a certain type of ground subsidence monitoring and early warning device embodiment. At least some of the features/methods described in the present invention may be implemented in a network device or component, such as a land subsidence monitoring and early warning device. For example, the features/methods of the present invention may be implemented in hardware, firmware, and/or software installed and running on hardware. The land subsidence monitoring and early warning device can be any device that processes, stores and/or forwards data frames through a network, for example, a server, a client, a data source, and the like. The ground subsidence monitoring and early warning device may include a transceiver (Tx/Rx) 210, which may be a transmitter, a receiver, or a combination thereof. The Tx/Rx 210 may be coupled to multiple ports 250 (eg, upstream and/or downstream interfaces) for sending and/or receiving frames from other nodes. Processor 230 may be coupled to Tx/Rx 210 to process the frame and/or determine which nodes to send the frame to. Processor 230 may include one or more multi-core processors and/or memory device 232, which may function as data storage, buffers, and the like. The processor 230 may be implemented as a general-purpose processor, or may be part of one or more application specific integrated circuits (ASICs) and/or digital signal processors (DSPs).

Based on the comprehensive unit of land subsidence, an early warning standard is established. The early warning standard integrates the “National Land Subsidence Prevention and Control Plan”, the “Regional Land Subsidence Prevention and Control Plan”, the threshold of each building/linear engineering deformation caused by land subsidence, and the “regional land subsidence disaster risk assessment factor level”, and comprehensively compares various thresholds. Yes, divide the warning levels from small to large. Regional early warning indicators

1) Groundwater level

The annual prevention and control indicators of land subsidence are decomposed into quarters, months and days according to the development law of the year, and the compression amount of key subsidence strata and the groundwater decline of adjacent aquifers that cause subsidence form a clear alternative model on the basis of continuous iteration. The subsidence value is decomposed in detail in the vertical space, and the groundwater level is used as the main early warning threshold and index, and the real-time monitoring and early warning is carried out with the results of the dynamic monitoring data of the groundwater level.

2) Surface deformation

The annual prevention and control indicators of land subsidence are decomposed into quarters, months and days according to the development law of the year, and the monitoring results are implemented by using GPS continuous stations in typical areas. Compared with the prevention and control indicators, real-time monitoring and early warning are carried out.

(3) Classification of early warning levels

By constructing a dual early-warning index system that combines the subsidence index represented by the dynamic cumulative subsidence value during the year and the groundwater level index, the early-warning and forecasting is realized. The early warning risk prevention and control levels are divided into three levels according to the size of the risk. Each level corresponds to the corresponding subsidence and groundwater level indicators. The first level is the highest, the display color is red, the second level is the second, the display color is orange, and the third level is the lowest. Display color is yellow.

The above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced. Without departing from the spirit and scope of the technical solutions of the present invention, all of them should be included in the scope of the claims of the present invention. The technology, shape, and structural part that are not described in detail in the present invention are all well-known technologies.

Claims (8)

1. a regional land subsidence early warning method based on multi-source heterogeneous data, is characterized in that, this method comprises data acquisition module, data transmission module, solution module, analysis module and early warning module, and described data acquisition module is used for ground Identification and quantification of main control factors of settlement, screening and evaluation of multi-source monitoring technology, and the calculation module includes monitoring data calculation and early warning index system construction. 2. A kind of regional land subsidence early warning method based on multi-source heterogeneous data according to claim 1, is characterized in that, identification and quantification of main control factors of land subsidence are obtained by the water level measuring instrument placed in the groundwater monitoring well; tall buildings The static loads of objects and dense building groups are selected using the method based on the remote sensing building land index, and the remote sensing images covering the study area are selected. The remote sensing image preprocessing process includes: geometric correction, radiometric calibration, atmospheric correction, image masking, and then the vegetation index SAVI extraction formula is as shown in Equation 1: SAVI=[(NIR-Red)(1+L)]/(NIR+Red+L) Formula (1) Among them, NIR is the pixel brightness value in the near-infrared band, Red is the pixel brightness value in the red band, and L is the soil adjustment factor, which ranges from 0 to 1. 3. a kind of regional land subsidence early warning method based on multi-source heterogeneous data according to claim 2, is characterized in that, the normalized water body index MNDWI of correction is extracted, as formula (2) MNDWI=(Green-MIR)/(Green+MIR) Formula (2) In the formula, Green represents the pixel brightness value in the green light band, MIR is the pixel brightness value in the mid-infrared band, Normalized building land index NDBI extraction, such as formula (3) NDBI=(MIR-NIR)/(MIR+NIR) Formula (3) In the formula, MIR is the gray value of the pixel in the mid-infrared band, NIR is the brightness value of the pixel in the near-infrared band, and NDBI is -1 to 1. 4. a kind of regional land subsidence early warning method based on multi-source heterogeneous data according to claim 1, it is characterized in that, described information collection module is based on the remote sensing construction land index IBI extraction of index, first superimposes three index bands Generate a new image, further analyze the spectral characteristics of each band of the generated new image, and find that the building land, vegetation and water body have the maximum values under the conditions of NDBI, SAVI and MNDWI, respectively, and use these three index bands to construct an index-based building land index IBI. , the formula is as formula (4): IBI=[NDBI-(SAVI+MNDWI)/2]/[NDBI+(SAVI+MNDWI)/2] Formula (4) Dynamic load of linear work such as rail transit: select the average static load of rail transit engineering, and the calculation formula is as follows (5) P _ES = 0.26PE (1+0.004V _{E )} ;

In the formula, P _E : static load of rail traffic; V _E vehicle running speed; P _ES : average static load of vehicles; n: number of vehicles; Mi: contribution rate; Pi average static load of vehicles PVS linear engineering average static load. 5. A kind of regional land subsidence early warning method based on multi-source heterogeneous data according to claim 1, is characterized in that, described analysis module comprises the method of SRCC, random forest, by the method obtained by claim 2-4 The data is calculated by the following formula, as shown in formula (6)

Among them: Pk represents: the proportion of the k-th type of samples in the current sample set D is Pk. 6. a kind of regional land subsidence early warning method based on multi-source heterogeneous data according to claim 1, is characterized in that, described multi-source monitoring technology screening and evaluation comprise InSAR, GPS, leveling, hierarchical standard, distribution The method of using InSAR, GNSS, and distributed optical fiber as the superior monitoring method for ground subsidence monitoring and early warning. InSAR calculation adopts the small baseline interferometry (SBAS-InSAR) technology, and the formula is as formula (7)

where ν( ) and β( ) are the average velocity and nonlinear components in the residual deformation of the high-resolution single-level differential interferogram, respectively, and ΔZ(x, r) is the high-resolution single-level differential interferogram The terrain component in , Δn(x, r) is the noise error, for the estimation of ν( ) and ΔZ(x, r), they must satisfy the maximum time coherence factor, which is expressed as formula (8)

where _δφmo is the analog phase, expressed as follows:

Subtract formulas (8) and (9) to obtain a new phase, including β( ) and Δm(x, r), and use the singular value decomposition method to remove the nonlinear deformation rate β( ), the overall deformation It can be expressed as formula (10) d( _tn ,x,r) _=dL ( _tn ,x,r)+( _{tn- t0} )v(x,r)+β( _tn ,x,r), n=0,1 , 2...N formula (10). 7. A kind of regional land subsidence early warning method based on multi-source heterogeneous data according to claim 1, is characterized in that, early warning index system construction and threshold determination are based on land subsidence main control factor identification and quantification results, to determine land subsidence early warning For the various evaluation factors and their weights required for the construction of the index system, cluster analysis and fuzzy mathematics methods are used to conduct coupling and superposition analysis of various single-factor evaluation index graphs to establish a land subsidence early warning evaluation index system. 8. A land subsidence monitoring and early warning device, characterized by comprising one or more processors; a storage device for storing one or more programs, when the one or more programs are processed by the one or more programs The one or more processors implement the method for early warning of regional land subsidence based on multi-source heterogeneous data according to any one of claims 1-7.