CN115630686A - Method for recovering land water reserve abnormity from satellite gravity data by machine learning - Google Patents
Method for recovering land water reserve abnormity from satellite gravity data by machine learning Download PDFInfo
- Publication number
- CN115630686A CN115630686A CN202211241455.3A CN202211241455A CN115630686A CN 115630686 A CN115630686 A CN 115630686A CN 202211241455 A CN202211241455 A CN 202211241455A CN 115630686 A CN115630686 A CN 115630686A
- Authority
- CN
- China
- Prior art keywords
- data
- land
- spherical harmonic
- water reserve
- water
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Geophysics And Detection Of Objects (AREA)
Abstract
The invention discloses a method for recovering land water reserve abnormality from satellite gravity data by machine learning, which comprises the following steps: s1: acquiring satellite data of a GRACE gravity satellite, land-water reserve data of a hydrological model and boundary and longitude and latitude information of a preset area, and setting boundary and longitude and latitude information of a background area according to a range in which gravity signals of the preset area are likely to leak; s2: calculating the smooth land water reserves abnormal of the GRACE gravity satellite inversion in the preset area and the background area; s3: acquiring lunar scale land-ground water reserve abnormal data of a global hydrological model, and carrying out leveling operation according to a research time period; s4: carrying out forward simulation on land and water reserve abnormal data of the globe, and acquiring smoothed land and water reserve abnormal data according to the boundary and longitude and latitude information of a background area; s5: constructing and training a neural network model; s6: and (3) inputting the land water reserves abnormal data after smoothing of the background area obtained in the step (S2) into a trained neural network model to recover the leakage signal.
Description
Technical Field
The invention relates to the field of gravity satellite technology and hydrology, in particular to the application field of gravity satellite inversion of land water reserve abnormity, and more particularly relates to a method for recovering land water reserve abnormity from satellite gravity data by using machine learning, namely a method for recovering a gravity satellite leakage signal to obtain land water reserve abnormity based on a convolutional neural network.
Background
The sum of the various forms of water reserves on land is called the land water reserve, including vegetation canopy water, soil water content, surface water reserve, snow water equivalent, and ground water reserve. The land Water Storage Anomaly (TWSA) refers to a difference value between the land Water Storage in a study time period and an average land Water Storage in a study time period, and can reflect increase and decrease of the land Water Storage. Land water reserves are abnormal and are important indicators for measuring the condition of regional water resources and are also important components of the water-heat balance of the global land ecosystem. The method can provide powerful support for realizing sustainable management of water resources by quantitatively and accurately estimating the abnormal space-time change of regional land water reserves.
Common land water reserves monitoring methods include hydrometeorology observation, land process modeling, and the like. The traditional hydrological meteorological observation method is easily restricted by conditions such as space-time distribution of space sites, complex landforms, capability of people and property and the like; land hydrological models such as Global Land Data Acquisition Systems (GLDAS) and Global hydrological models (hereinafter collectively referred to as hydrological models) can also simulate the change of Land water reserves near the surface of the earth, but due to the limitations of the hydrological models themselves and the absence of certain hydrological elements (such as groundwater), the area TWSA cannot be estimated accurately. Gravity field Recovery and Climate test satellite GRACE (Gravity Recovery and Climate Experiment) has been widely used in the inversion of TWSA since 2002 emission. However, the high-order coefficients of the Grace spherical harmonic product have large random errors and system errors, and need to be removed through low-pass filtering, thereby causing the problem of signal leakage. The traditional signal leakage correction method mainly adopts a scale factor method, a multiplication correction method, an addition correction method and an iteration correction method. However, the former two methods only rely on one kind of hydrological model when applied, and the accuracy of the hydrological model is relatively relied on, and if the hydrological model is used inaccurately, the calculation deviation is often very large, and the leaked signals cannot be recovered well. In addition, the multiplicative correction method assumes that the TWSA of a region is uniform, but the TWSA of an actual region tends to hardly conform to this assumption. Constrained iteration in the iterative correction method usually needs to obtain the source range of the signal, and unconstrained iteration has higher requirement on TWSA uniformity of an inversion region, so that spatial distribution of the signal is easy to be too smooth, and distortion is caused. In summary, there is a lack in the art of a method for recovering the GRACE signal leakage that is independent of a single hydrological model, has low prior signal dependency, and can ensure TWSA spatial details. The present invention aims to solve the above problems and provide a method for correcting a GRACE leakage signal based on a convolutional neural network.
Disclosure of Invention
In order to solve the above problems, the present invention provides a method for recovering Land water reserve abnormality from satellite gravity data by using machine learning, which is based on a convolutional neural network, and uses a Land water reserve abnormality GRACE leakage signal obtained by using five hydrological models, namely, WGHM (waterfap Global Hydrology Model), noah (Noah Land Model), CLSM (watershed Surface Model), mos (clinical Land Surface Model), and VIC (Variable permeability macro Model) to simulate Model training, and the neural network Model obtained by training is used for correcting the leakage signal in the process of processing the gravel sensitivity scale-2 spherical coefficient product. The method of the invention avoids the deficiency of prior knowledge during signal leakage error correction, does not depend on a single hydrological model, can consider the characteristics of spatial relationship by means of a convolutional neural network, can accurately invert the spatial details of land water reserve change, and expands the application space of the gravity satellite technology in the land water reserve change inversion aspect and the hydrological field.
To achieve the above object, the present invention provides a method for recovering terrestrial water reserve abnormality from satellite gravity data by machine learning, comprising:
step S1: acquiring satellite data of a GRACE gravity satellite, land-water reserve data of a hydrological model and boundary and longitude and latitude information of a preset area, and setting boundary and longitude and latitude information of a background area according to a range in which gravity signals of the preset area are likely to leak, wherein the background area is an area which is out of the range of the preset area and contains most of leaked signals;
step S2: according to the data obtained in the step S1 and longitude and latitude information of the preset area and the background area, calculating the abnormal smooth land water reserves inverted by GRACE gravity satellites in the preset area and the background area;
and step S3: acquiring month scale land water reserve abnormal data of a global hydrological model, and performing leveling operation according to a research time period;
and step S4: carrying out forward simulation on land water reserve abnormal data of the world, reducing errors in high-order spherical harmonic coefficients through low-pass filtering, and acquiring smoothed land water reserve abnormal data according to the boundary and longitude and latitude information of a background area;
step S5: constructing and training a neural network model;
step S6: and (3) inputting the smoothed land water storage quantity abnormal data obtained by the background region GRACE gravity satellite inversion in the step (S2) into a trained neural network model to recover the leakage signal.
In an embodiment of the invention, the GRACE gravity satellite data is a level-2 spherical harmonic coefficient product, and the land water reserve information represented by the GRACE gravity satellite data comprises surface water, soil water and underground water; the acquired hydrological models comprise five hydrological models of WGHM, noah, CLSM, mosaic and VIC, and land water reserve data of the hydrological models comprise the sum of snow water equivalent output results, soil water reserve output results and underground water reserve output results simulated by the hydrological models.
In an embodiment of the present invention, the step S2 of calculating the land-water reserve anomaly according to the satellite data of the GRACE gravity satellite is a process of converting level-2 spherical harmonic data with random errors and system errors in a high order into land-water reserve anomaly data with a smooth background region, and the land-water reserve anomaly data is converted into a land-water reserve anomaly value with a smooth background region according to longitude and latitude information of the background region through constructing an earth gravity field model, low-pass filtering, band-removing filtering and distance leveling operations.
In an embodiment of the present invention, the low-pass filtering is implemented by gaussian filtering with a radius of 300km, and the de-banding filtering is preferably implemented by a P3M10 manner, that is, spherical harmonic coefficients of the first 10 × 10 orders are kept unchanged, a 3-order polynomial is used to fit the spherical harmonic coefficients of more than or equal to 10 orders, odd-order and even-order polynomials are used to fit separately, and fitting values are subtracted from the original spherical harmonic coefficients, and the specific process includes:
cutting off a GRACE level-2 spherical harmonic coefficient product to a spherical harmonic coefficient of a certain order;
removing the 0 order term;
replacing the C20 item by adopting the result of the satellite laser ranging data calculation;
the Technical Note TN-13 data published by JPL was used to replace the 1 st order term of the level-2 spherical harmonic coefficient product.
In an embodiment of the present invention, step S2 further includes a conversion between a spherical harmonic coefficient and a land-water reserve abnormality, which is smoothed grid data obtained by converting a level-2 spherical harmonic coefficient of a GRACE gravity satellite into a background area, and specifically includes:
step S201: assuming that the earth mass weight distribution causing gravity anomaly is only concentrated on the earth surface layer, mainly comprising the changes of atmosphere, sea, ice cover and land water reserves, converting the density change corresponding to the mass weight distribution into the caused ground level surface height change by using radial integration, wherein the change comprises two parts of contribution of ground mass direct gravity attraction to the ground level surface change and load deformation caused by solid earth after the ground surface load change, and the formula is expressed as formula (1),
wherein θ represents the earth's remaining latitude (0 to 1)80 degrees, lambda represents the east meridian of the earth (-180 degrees to 180 degrees), l and m respectively represent the order and the frequency of the spherical harmonic expansion of the gravity field, and the dimensionless coefficient delta C lm And Δ S lm Representing the spherical harmonic coefficient of the change of the geohorizon,
is a standard associated Legendre function of order l m, ρ e Represents the average density of the earth (about 5517 kg/m) 3 )，k l Expressing the load lux number of order l, Δ σ (θ, λ) is the surface mass density variation, a is the earth mean radius (about 6378 km);
step S202: if the change of the earth surface mass density delta sigma (theta, A) is also subjected to spherical harmonic expansion, then
In the formula, ρ w Is the density of water (1000 kg/m) 3 ) In comparison with equation (1), ρ is known w The delta sigma/rho is often adopted for surface mass density conversion w Is expressed in terms of, thus yielding:
in the formula (I), the compound is shown in the specification,
and
spherical harmonic coefficients for surface density variations;
step S203: obtaining spherical harmonic coefficient of surface density variation according to formula (1) and formula (3)
And
spherical harmonic coefficient Delta C of large ground level surface change lm And Δ S lm The functional relationship between the two is as follows:
step S204: substituting formula (4) into formula (2) yields:
wherein, the formula (5) is a basic formula for calculating the change of the earth surface mass density by using the geodetic level spherical harmonic coefficient data published by the GRACE data center, namely a basic conversion formula between the spherical harmonic coefficient and the land water reserve abnormality;
step S205: when only the land water reserve change is considered, the land water reserve change is estimated by combining the formula (1) according to the acquired spherical harmonic coefficient of the ground level change provided by the GRACE satellite.
In an embodiment of the present invention, step S3 specifically includes:
step S301: according to the global longitude and latitude, the soil water reserves, the surface water reserves and other land water reserve change components of the five hydrological models are respectively extracted through the following formula, so that five land water reserve change data are obtained:
TWS＝SMS+GWS+SWS+OS
wherein TWS is land water reserve, SMS is soil water reserve, GWS is groundwater reserve, SWS is surface water reserve, OS is other land water reserve change component except SMS, GWS and SWS;
step S302: and (4) carrying out distance leveling processing on the obtained land water reserve data, namely deducting the average value of the land water reserve in the research time period from the value of each month to obtain land water reserve abnormal data of the background area simulated by the hydrological model.
In an embodiment of the present invention, the forward modeling in step S4 is to convert the land water reserve abnormal data simulated by the hydrological model into spherical harmonic coefficients, refer to equation (3), and perform a data processing procedure similar to the spherical harmonic coefficients of the gravity satellite on the spherical harmonic coefficients, where the specific processing procedure includes:
the low-pass filtering adopts Gaussian filtering with the radius of 300 km;
removing the 0 order term;
cutting off the GRACE level-2 spherical harmonic coefficient product to a spherical harmonic coefficient of a certain order;
and in the step S4, land water storage quantity abnormal data after smoothing is obtained according to the boundary and longitude and latitude information of the background area and is converted according to the formula (5).
In an embodiment of the present invention, step S5 specifically includes:
step S501: taking the land water reserve abnormal value after smoothing of the background area as input data of the neural network model, taking the land water reserve abnormal value of the preset area as target data of the neural network model, and establishing the neural network model;
step S502: the training of the model is done using Python language.
Compared with the prior art, the method for recovering the land water reserve abnormity from the satellite gravity data by using machine learning can avoid the deficiency of priori knowledge during signal leakage error correction to a greater extent, does not depend on a single hydrological model, better reduces the influence of leakage errors, and can accurately invert the space details of the land water reserve abnormity.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, it is obvious that the drawings in the following description are only some embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to the drawings without creative efforts.
FIG. 1 is a flow chart of a method according to an embodiment of the present invention;
FIG. 2 is a schematic diagram of a basic structure of a neural network model according to an embodiment of the present invention;
FIG. 3 is a comparison of the performance of the test set of the hydrological model in Qinghai-Tibet plateau in the convolutional neural network model in an embodiment of the present invention, wherein (a) represents the average value of the test set of the land water reserve anomaly spherical harmonic expansion simulated by the hydrological model; (b) The average value of the test set representing land water reserve abnormality simulated by the hydrological model; (c) representing an average of the convolutional neural network output results; (d) represents the difference between (b) and (c);
fig. 4 is a comparison between land water reserve anomalies of 2003-2016 year after GRACE inversion by convolutional neural network recovery and other methods in Qinghai-Tibet plateau in an embodiment of the present invention, wherein (a), (b), (c), and (d) respectively represent trend values of the land water reserve anomalies by GRACE inversion by convolutional neural network recovery, trend values of scale factor method, trend values of Mascon data issued by CSR official, trend values of Mascon data issued by JPL official, and (e) represents a time series of four land water reserve anomalies.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be obtained by a person skilled in the art without inventive effort based on the embodiments of the present invention, are within the scope of the present invention.
In a specific embodiment, the technical method of the present invention can be realized by the following method steps. It should be noted that, in this embodiment, the following steps are not strictly upper and lower step relationships in a logical order, but only to describe the specific implementation process and the main contents of the present solution in detail as much as possible, for example, steps S2 to S4, except that corresponding steps of data having a mathematical computation logical relationship between the upper and lower steps have a front-back relationship, data computation in other parallel relationships may be implemented by parallel steps, and are not necessarily implemented by a fixed upper-lower step relationship, and the following numerical order of S1 to S6 should not be understood as a limitation to the protection scope of the present solution.
Fig. 1 is a flowchart of a method according to an embodiment of the present invention, and as shown in fig. 1, the embodiment provides a method for recovering a terrestrial water reserve anomaly from satellite gravity data by using machine learning, which includes:
step S1: acquiring satellite data of a GRACE gravity satellite, land-water reserve data of a hydrological model and boundary and longitude and latitude information of a preset area, and setting boundary and longitude and latitude information of a background area according to a range in which gravity signals of the preset area are likely to leak;
in this embodiment, the GRACE gravity satellite data obtained in step S1 includes level-2 spherical harmonic coefficient products issued by the spatial research Center (CSR) of texas university, usa, and also includes level-2 spherical harmonic coefficient products issued by other domestic and foreign institutions such as Jet Power Laboratories (JPL), the germany bosentan Geocenter (GFZ), and the like, and land water reserve information represented by the products includes surface water, soil water, groundwater, and the like; the obtained hydrological models comprise five hydrological models, namely WGHM (WaterGAP global hydrology model), noah (Noah land model), CLSM (river basin surface model), mosaic (Mosaic land model) and VIC (variable permeability macroscopic model), and land water reserve data of the hydrological models comprise the sum of snow water equivalent output results, soil water reserve output results and underground water reserve output results simulated by the hydrological models. Wherein, the WGHM and CLSM hydrological models in the hydrological model used by the invention contain output results of underground water reserve change, and the other three models do not contain.
The background area is an area which is outside the range of the preset area and contains most of leakage signals, and the range of the background area can be set to be 10 degrees wider than the latitude and longitude of the preset area towards the periphery. For example, if the minimum longitude, the maximum longitude, the minimum latitude, and the maximum latitude of the preset region are 110 °, 120 °, 40 °, and 50 °, respectively, the minimum longitude, the maximum longitude, the minimum latitude, and the maximum latitude of the background region are 100 °, 130 °, 30 °, and 60 °, respectively.
Step S2: according to the data obtained in the step S1 and longitude and latitude information of the preset area and the background area, calculating the abnormal smooth land water reserves inverted by GRACE gravity satellites in the preset area and the background area;
in this embodiment, the step S2 of calculating the land-water reserve anomaly according to the satellite data of the GRACE gravity satellite is specifically a process of converting level-2 spherical harmonic data with a random error and a system error at a high order into land-water reserve anomaly data with a smooth background region, and the land-water reserve anomaly data can be converted into a land-water reserve anomaly value with a smooth background region according to longitude and latitude information of the background region through constructing an earth gravity field model, low-pass filtering, band-removing filtering and distance leveling operations.
In this embodiment, the low-pass filtering may preferably adopt gaussian filtering with a radius of 300km, and may also adopt other low-pass filtering for processing; the band elimination filtering preferably adopts a P3M10 mode, namely, the spherical harmonic coefficient of the first 10 multiplied by 10 orders is kept unchanged, a 3-order polynomial is used for fitting the spherical harmonic coefficient of more than or equal to 10 orders, the odd-order and the even-order are separately fitted, and the fitting value is deducted from the original spherical harmonic coefficient, and the specific process comprises the following steps:
cutting off a GRACE level-2 spherical harmonic coefficient product to a spherical harmonic coefficient of a certain order, such as 60 orders;
removing the 0 order term;
replacing the C20 items by adopting the result of the satellite laser ranging data calculation;
the level 1 term of the level-2 spherical harmonic coefficient product was replaced with Technical Note TN-13 data published by JPL.
It should be noted that when the filter and truncation and other processing are performed on the level-2 spherical harmonics of the GRACE gravity satellite, the real signal will be lost while the high-order error is attenuated, and the embodiment will recover the real signal in step S6. The gaussian filter used in this embodiment is substantially a weighted function of spatial averaging, and the signal is smoothed by performing weighted averaging in a space of a certain radius range. The de-banding filtering adopted by the embodiment can remove the correlation among the spherical harmonic coefficients, but also can cause distortion influence on a real signal; an important characteristic of the order of the spherical harmonic coefficient is to reflect the spatial resolution of data, and theoretically, the spherical harmonic function can be expanded to an infinite order to show more spatial details, but the higher the order is, more errors are brought.
In this embodiment, the step S2 further includes a conversion between a spherical harmonic coefficient and a terrestrial water reserve abnormality, where the conversion is smooth grid data obtained by converting a level-2 spherical harmonic coefficient of a GRACE gravity satellite into a background region, and specifically includes:
step S201: assuming that the earth mass weight distribution causing gravity anomaly is only concentrated on the earth surface layer (within 10-15 km), mainly comprising the changes of atmosphere, sea, ice cover, land water reserves and the like, converting the density change corresponding to the mass redistribution into the caused ground level height change by using radial integration, wherein the change comprises two parts of contribution of ground mass direct gravity attraction to the ground level change and load deformation caused by solid earth after the ground load change, and the formula is expressed as formula (1),
step S202: if the change of the earth surface mass density delta sigma (theta, lambda) is also subjected to spherical harmonic expansion, then
In the formula, ρ w Is the density of water (1000 kg/m) 3 ) In comparison with the formula (1), ρ is w Delta sigma/rho is often adopted due to surface mass density transformation w Is expressed, thus yielding:
in the formula (I), the compound is shown in the specification,
and
spherical harmonic coefficients for surface density variations;
step S203: obtaining spherical harmonic coefficient of surface density change according to formula (1) and formula (3)
And
spherical harmonic coefficient Delta C of ground level surface change lm And Δ S lm The functional relationship between the two is as follows:
step S204: substituting formula (4) into formula (2) yields:
wherein, the formula (5) is a basic formula for calculating the change of the earth surface mass density by using the geodetic level spherical harmonic coefficient data published by the GRACE data center, namely a basic conversion formula between the spherical harmonic coefficient and the land water reserve abnormality;
in the hydrological study, land water reserves are usually concerned, therefore, based on the 'thin layer' assumption (the assumption of step S201), the part of disturbance position signals (or ground level) caused by mass redistribution such as the atmosphere and the ocean can be simulated by using relevant model simulation dataChange) and subtracted from the original GRACE signal, the resulting residual perturbation bit signal (or ground level change) is mainly caused by land water reserve changes. On the premise of only considering the change of the land water reserves, by using the spherical harmonic coefficient of the change of the ground level provided by the GRACE satellite, the change of the land water reserves (or the equivalent water height change) can be estimated according to the formula (1) Δ σ/ρ w )。
Step S205: when only the land water reserve change is considered, the land water reserve change is estimated by combining the formula (1) according to the acquired spherical harmonic coefficient of the ground level change provided by the GRACE satellite.
And step S3: acquiring lunar scale land-ground water reserve abnormal data of a global hydrological model, and carrying out leveling operation according to a research time period; the hydrological model of the embodiment includes five hydrological models of WGHM, noah, CLSM, mosaic and VIC, and in addition, other embodiments may also include other hydrological models which are issued by other organizations at home and abroad and can acquire water storage data such as regional soil water storage, snow water equivalent and the like;
in this embodiment, step S3 specifically includes:
step S301: according to the global longitude and latitude, the soil water reserves, the surface water reserves and other land water reserve change components of the five hydrological models are respectively extracted through the following formula, so that five land water reserve change data are obtained:
TWS＝SMS+GWS+SWS+OS
wherein TWS is land water reserve, SMS is soil water reserve, GWS is groundwater reserve, SWS is surface water reserve, OS is other land water reserve change component except SMS, GWS and SWS;
step S302: and (4) carrying out distance leveling processing on the obtained land water reserve data, namely deducting the average value of the land water reserve in the research time period from the value of each month to obtain land water reserve abnormal data of the background area simulated by the hydrological model.
Among them, it should be noted that: since the resolutions of the five hydrological models are different, as a more preferable embodiment, all of the five hydrological models can be converted into the resolution of 1 ° × 1 ° in this embodiment. Wherein, WGHM and CLSM contain output results of groundwater reserves change in five hydrological models, and the other three models do not contain.
And step S4: carrying out forward simulation on land water reserve abnormal data of the world, reducing errors in high-order spherical harmonic coefficients through low-pass filtering, and acquiring smoothed land water reserve abnormal data according to the boundary and longitude and latitude information of a background area;
in this embodiment, the forward modeling in step S4 is to convert the land water reserve abnormal data simulated by the hydrological model into spherical harmonic coefficients, and may refer to formula (3), and perform a data processing procedure similar to the spherical harmonic coefficients of the gravity satellite on the spherical harmonic coefficients, where the specific processing procedure includes:
the low-pass filtering adopts Gaussian filtering with the radius of 300 km;
removing 0 order item;
truncating the GRACE level-2 spherical harmonic coefficient product to a spherical harmonic coefficient of a certain order, for example, to 60 orders;
and in the step S4, the land water storage quantity abnormal data after smoothing is obtained according to the boundary and longitude and latitude information of the background area and is converted according to the formula (5).
Calculating the land water reserves of the hydrological model into smoothed land water reserve abnormalities specifically comprises performing spherical harmonic expansion on the land water reserves of the hydrological model, and acquiring the smoothed land water reserve abnormalities by adopting low-pass filtering and pitch leveling operations which are the same as GRACE gravity satellite data, wherein the operations are also called forward modeling.
Step S5: constructing and training a neural network model;
fig. 2 is a schematic diagram of a basic structure of a neural network model adopted in an embodiment of the present invention, as shown in fig. 2, the neural network adopted in the embodiment is preferably a convolutional neural network model, in other embodiments, other application network models may also be adopted, a platform for constructing the neural network model is a tersorflow 2.2.0 and a Keras 2.4.3 packet of Python 3.8, a learning rate of the model is 0.01, a batch size is 32, an iteration number (epoch) of training is 300, a drop rate (dropout) is 0.1, an optimizer adopts a stochastic gradient descent optimizer (SGD), and a loss function is MSE.
In this embodiment, step S5 specifically includes:
step S501: taking the land water reserve abnormal value after smoothing of the background area as input data of the neural network model, taking the land water reserve abnormal value of the preset area as target data of the neural network model, and establishing the neural network model;
specifically, the land water reserve abnormal data of the five hydrological models in the preset area in the step S3 and the land water reserve abnormal data of the background area, which is smoothed by the forward modeling of the five hydrological models in the step S4, can be used as training of the models, the former is used as target data of the neural network model, and the latter is used as input data of the neural network model. The input data and the target data of the five hydrological models for training the convolutional neural network model can be respectively extracted 80% for model training, and the other 20% for model verification.
Step S502: the training of the model is done using Python language.
In order to confirm the recovery effect of the neural network model on the leakage signal, the evaluation result of the neural network model is performed on 20% of randomly extracted data, and is shown in fig. 3, fig. 3 is a comparison of the performance of the test set of the hydrological model in the Tibet plateau in the convolutional neural network model in one embodiment of the present invention, wherein (a) represents the average value of the test set of the land water reserve abnormal spherical harmonic expansion simulated by the hydrological model; (b) The average value of the test set representing the land water reserve abnormality simulated by the hydrological model; (c) representing an average of the convolutional neural network output results; (d) represents the difference between (b) and (c).
In this example, because the input data adopted in the training of the model is the land-water reservoir anomaly after the smoothing of the background region of the hydrological model, and the target data is the land-water reservoir anomaly value of the preset region of the hydrological model, the trained convolutional neural network model can be used for recovering the land-water reservoir anomaly after the smoothing of the background region into the land-water reservoir anomaly of the preset region.
Step S6: and (3) inputting the smoothed land water storage quantity abnormal data obtained by the background region GRACE gravity satellite inversion in the step (S2) into a trained neural network model to recover the leakage signal.
And inputting the data of the land water reserves with abnormal smooth background areas obtained by GRACE inversion into a trained convolutional neural network model to obtain the land water reserves abnormal data after recovery of the leakage signals. Fig. 4 is a comparison of land water reserve anomalies of 2003-2016 years of GRACE inversion recovered by convolutional neural network with those of other methods in Qinghai-Tibet plateau in an embodiment of the present invention, as shown in fig. 4, (a), (b), (c), and (d) respectively represent trend values of land water reserve anomalies of GRACE inversion recovered by convolutional neural network, trend values of scale factor method, trend values of Mascon data issued by CSR official, trend values of Mascon data issued by JPL official, and (e) represents a time series of four land water reserve anomalies.
The invention provides a method for recovering terrestrial water reserve abnormity from satellite gravity data by machine learning, which is a method for recovering leakage signals with terrestrial water reserve abnormity based on GRACE gravity satellite level-2 spherical harmonic data. Compared with the prior art, the method can avoid the deficiency of priori knowledge during signal leakage error correction, does not depend on a single hydrological model, better reduces the influence of leakage errors, and can accurately invert the space details of land water reserves.
Those of ordinary skill in the art will understand that: the figures are merely schematic representations of one embodiment, and the blocks or flow diagrams in the figures are not necessarily required to practice the present invention.
Finally, it should be noted that: the above examples are only intended to illustrate the technical solution of the present invention, and not to limit it; although the present invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some technical features may be equivalently replaced; and such modifications or substitutions do not depart from the spirit and scope of the corresponding technical solutions of the embodiments of the present invention.
Claims (8)
1. A method for recovering terrestrial water reserve anomalies from satellite gravity data using machine learning, comprising:
step S1: acquiring satellite data of a GRACE gravity satellite, land-water reserve data of a hydrological model and boundary and longitude and latitude information of a preset area, and setting boundary and longitude and latitude information of a background area according to a range in which gravity signals of the preset area are likely to leak, wherein the background area is an area which is out of the range of the preset area and contains most of leaked signals;
step S2: according to the data obtained in the step S1 and longitude and latitude information of the preset area and the background area, calculating the abnormal smooth land water reserves inverted by GRACE gravity satellites in the preset area and the background area;
and step S3: acquiring month scale land water reserve abnormal data of a global hydrological model, and performing leveling operation according to a research time period;
and step S4: forward modeling is carried out on land water reserve abnormal data of the whole world, errors in high-order spherical harmonic coefficients are reduced through low-pass filtering, and smoothed land water reserve abnormal data are obtained according to the boundary and longitude and latitude information of a background area;
step S5: constructing and training a neural network model;
step S6: and (3) inputting the smoothed land water storage quantity abnormal data obtained by the background region GRACE gravity satellite inversion in the step (S2) into a trained neural network model to recover the leakage signal.
2. The method for recovering terrestrial water reserves abnormality from satellite gravity data using machine learning as claimed in claim 1 wherein GRACE gravity satellite data is level-2 spherical harmonic coefficient product, which represents terrestrial water reserves information including surface water, soil water and ground water; the acquired hydrological models comprise five hydrological models of WGHM, noah, CLSM, mosaic and VIC, and land water reserve data of the hydrological models comprise the sum of snow water equivalent output results, soil water reserve output results and underground water reserve output results simulated by the hydrological models.
3. The method for recovering terrestrial water reservoir anomaly from satellite gravity data by using machine learning according to claim 2, wherein the step S2 of calculating the terrestrial water reservoir anomaly from the satellite data of the GRACE gravity satellite is a process of converting level-2 spherical harmonic data containing random errors and system errors at a high order into terrestrial water reservoir anomaly data with a smoothed background region by constructing an earth gravity field model, and converting the terrestrial water reservoir anomaly data into a smoothed late terrestrial water reservoir anomaly value according to latitude and longitude information of the background region through low pass filtering, band-removing filtering and distance leveling operations.
4. The method for recovering terrestrial water reserve abnormality from satellite gravity data by machine learning according to claim 3, wherein the low pass filtering is gaussian filtering with a radius of 300km, the de-banding filtering is preferably P3M10, that is, the first 10 x 10 order spherical harmonic coefficients are kept unchanged, a 3 rd order polynomial is used to fit the spherical harmonic coefficients with a value greater than or equal to 10 th order, the odd and even orders are fitted separately, and the fitted value is subtracted from the original spherical harmonic coefficients, and the method comprises:
cutting off a GRACE level-2 spherical harmonic coefficient product to a spherical harmonic coefficient of a certain order;
removing the 0 order term;
replacing the C20 item by adopting the result of the satellite laser ranging data calculation;
the Technical Note TN-13 data published by JPL was used to replace the 1 st order term of the level-2 spherical harmonic coefficient product.
5. The method for recovering terrestrial water reserves anomalies from satellite gravity data by machine learning as claimed in claim 4, wherein step S2 further comprises the conversion between spherical harmonics and terrestrial water reserves anomalies, which is the smoothed grid data of the level-2 spherical harmonics of the GRACE gravity satellite into background regions, specifically comprising:
step S201: assuming that the earth mass weight distribution causing gravity anomaly is only concentrated on the earth surface layer, mainly comprising the changes of atmosphere, sea, ice cover and land water reserves, converting the density change corresponding to the mass weight distribution into the caused ground level surface height change by using radial integration, wherein the change comprises two parts of contribution of ground mass direct gravity attraction to the ground level surface change and load deformation caused by solid earth after the ground surface load change, and the formula is expressed as formula (1),
in the formula, theta represents earth complementary latitude, lambda represents earth east longitude, l and m respectively represent the order and the number of times of gravity field spherical harmonic expansion, and a dimensionless coefficient delta C lm And Δ S lm Representing the spherical harmonic coefficient of the change of the geohorizon,
is a standard associated Legendre function of order l m, ρ e Representing the mean density of the earth, k l Expressing the load lux number of the order l, wherein delta sigma (theta, lambda) is the change of the mass density of the earth surface, and a is the average radius of the earth;
step S202: if the change of the earth surface mass density delta sigma (theta, lambda) is also subjected to spherical harmonic expansion, then
In the formula, ρ w Rho is the density of water, and is known from the formula (1) w The delta sigma/rho is often adopted for surface mass density conversion w Is expressed in terms of, thus yielding:
in the formula (I), the compound is shown in the specification,
and
spherical harmonic coefficients for surface density variations;
step S203: obtaining spherical harmonic coefficient of surface density variation according to formula (1) and formula (3)
And
spherical harmonic coefficient Delta C of large ground level surface change lm And Δ S lm The functional relationship between the two is as follows:
step S204: substituting formula (4) into formula (2) yields:
wherein, the formula (5) is a basic formula for calculating the change of the earth surface mass density by using the terrestrial level spherical harmonic coefficient data published by the GRACE data center, namely a basic conversion formula between the spherical harmonic coefficient and the land water reserve abnormality;
step S205: when only the land water reserve change is considered, the land water reserve change is estimated according to the acquired spherical harmonic coefficient of the large ground level change provided by the GRACE satellite in combination with the formula (1).
6. The method of claim 5, wherein the step S3 comprises:
step S301: according to the global longitude and latitude, the soil water reserves, the surface water reserves and other land water reserve change components of the five hydrological models are respectively extracted through the following formula, so that five land water reserve change data are obtained:
TWS＝SMS+GWS+SWS+OS
wherein TWS is land water reserve, SMS is soil water reserve, GWS is groundwater reserve, SWS is surface water reserve, OS is other land water reserve change component except SMS, GWS and SWS;
step S302: and (4) carrying out distance processing on the obtained land water reserves data, namely deducting the average land water reserves of the research time period from the value of each month to obtain land water reserve abnormal data of the background area simulated by the hydrological model.
7. The method for recovering terrestrial water reserve abnormality from satellite gravity data by machine learning according to claim 6, wherein the forward modeling in step S4 is to convert the terrestrial water reserve abnormality data simulated by the hydrological model into spherical harmonic coefficients, refer to equation (3), and perform a data processing procedure similar to the spherical harmonic coefficients of the gravity satellite on the spherical harmonic coefficients, and the specific processing procedure includes:
the low-pass filtering adopts Gaussian filtering with the radius of 300 km;
removing the 0 order term;
cutting off the GRACE level-2 spherical harmonic coefficient product to a spherical harmonic coefficient of a certain order;
in step S4, land and water storage quantity abnormal data after smoothing is obtained according to the boundary and longitude and latitude information of the background area and is converted according to the formula (5).
8. The method of claim 7, wherein the step S5 comprises:
step S501: taking the land water reserve abnormal value after smoothing of the background area as input data of the neural network model, taking the land water reserve abnormal value of the preset area as target data of the neural network model, and establishing the neural network model;
step S502: the training of the model is done using Python language.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211241455.3A CN115630686B (en) | 2022-10-11 | 2022-10-11 | Method for recovering land water reserve anomalies from satellite gravity data using machine learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211241455.3A CN115630686B (en) | 2022-10-11 | 2022-10-11 | Method for recovering land water reserve anomalies from satellite gravity data using machine learning |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN115630686A true CN115630686A (en) | 2023-01-20 |
| CN115630686B CN115630686B (en) | 2023-06-23 |
Family
ID=84905640
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211241455.3A Active CN115630686B (en) | 2022-10-11 | 2022-10-11 | Method for recovering land water reserve anomalies from satellite gravity data using machine learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115630686B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117152629A (en) * | 2023-08-23 | 2023-12-01 | 武汉大学 | Method and system for filling gravity vacancy data of drainage basin scale time-varying satellite |
Citations (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106529164A (en) * | 2016-11-03 | 2017-03-22 | 清华大学 | Method and system for acquiring ground water storage variation value by combining GRACE satellite |
| CN110175214A (en) * | 2019-02-01 | 2019-08-27 | 中国空间技术研究院 | A kind of method and system changed using Gravity Satellite data monitoring extreme climate |
| CN110852473A (en) * | 2019-09-24 | 2020-02-28 | 广州地理研究所 | Neural network algorithm-based land water reserve prediction method and equipment |
| CN111752934A (en) * | 2020-05-29 | 2020-10-09 | 首都师范大学 | High-resolution aquifer water storage coefficient inversion method and system based on gravity satellite |
| CN112200050A (en) * | 2020-09-30 | 2021-01-08 | 同济大学 | Lake water volume space-time change analysis method and device based on multi-source satellite data |
| CN112989557A (en) * | 2021-01-14 | 2021-06-18 | 中国空间技术研究院 | Method for improving water reserve change prediction reliability based on neural network selectable model |
| US20210341453A1 (en) * | 2020-04-30 | 2021-11-04 | Institute Of Geochemistry, Chinese Academy Of Sciences | Method and System for Estimating Groundwater Recharge Based on Pixel Scale |
| US20210341445A1 (en) * | 2020-04-30 | 2021-11-04 | Inst Geochemistry Cas | Method for monitoring change of vegetation water conservation |
| CN113868855A (en) * | 2021-09-24 | 2021-12-31 | 首都师范大学 | Groundwater reserve change satellite gravity forward modeling method integrating water level data |
| CN113962056A (en) * | 2021-04-25 | 2022-01-21 | 中国空间技术研究院 | Method for improving accuracy of water reserve change of GRACE land |
| CN114491967A (en) * | 2021-12-30 | 2022-05-13 | 中国科学院地理科学与资源研究所 | Land water reserve prediction method, device, equipment and storage medium |
| CN114676621A (en) * | 2022-01-25 | 2022-06-28 | 中国空间技术研究院 | Method for improving accuracy of land water reserve abnormity based on deep learning weight load |
| CN114858238A (en) * | 2022-04-29 | 2022-08-05 | 南方科技大学 | Underground water storage and transformation monitoring and early warning system |
-
2022
- 2022-10-11 CN CN202211241455.3A patent/CN115630686B/en active Active
Patent Citations (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106529164A (en) * | 2016-11-03 | 2017-03-22 | 清华大学 | Method and system for acquiring ground water storage variation value by combining GRACE satellite |
| CN110175214A (en) * | 2019-02-01 | 2019-08-27 | 中国空间技术研究院 | A kind of method and system changed using Gravity Satellite data monitoring extreme climate |
| CN110852473A (en) * | 2019-09-24 | 2020-02-28 | 广州地理研究所 | Neural network algorithm-based land water reserve prediction method and equipment |
| US20210341453A1 (en) * | 2020-04-30 | 2021-11-04 | Institute Of Geochemistry, Chinese Academy Of Sciences | Method and System for Estimating Groundwater Recharge Based on Pixel Scale |
| US20210341445A1 (en) * | 2020-04-30 | 2021-11-04 | Inst Geochemistry Cas | Method for monitoring change of vegetation water conservation |
| CN111752934A (en) * | 2020-05-29 | 2020-10-09 | 首都师范大学 | High-resolution aquifer water storage coefficient inversion method and system based on gravity satellite |
| US20210375034A1 (en) * | 2020-05-29 | 2021-12-02 | Capital Normal University | Method and system for inversion of high-resolution aquifer storage coefficient based on gravity satellite data |
| CN112200050A (en) * | 2020-09-30 | 2021-01-08 | 同济大学 | Lake water volume space-time change analysis method and device based on multi-source satellite data |
| CN112989557A (en) * | 2021-01-14 | 2021-06-18 | 中国空间技术研究院 | Method for improving water reserve change prediction reliability based on neural network selectable model |
| CN113962056A (en) * | 2021-04-25 | 2022-01-21 | 中国空间技术研究院 | Method for improving accuracy of water reserve change of GRACE land |
| CN113868855A (en) * | 2021-09-24 | 2021-12-31 | 首都师范大学 | Groundwater reserve change satellite gravity forward modeling method integrating water level data |
| CN114491967A (en) * | 2021-12-30 | 2022-05-13 | 中国科学院地理科学与资源研究所 | Land water reserve prediction method, device, equipment and storage medium |
| CN114676621A (en) * | 2022-01-25 | 2022-06-28 | 中国空间技术研究院 | Method for improving accuracy of land water reserve abnormity based on deep learning weight load |
| CN114858238A (en) * | 2022-04-29 | 2022-08-05 | 南方科技大学 | Underground water storage and transformation monitoring and early warning system |
Non-Patent Citations (7)
| Title |
|---|
| SHAOXING MO等: "Bayesian convolutional neural networks for predicting the terrestrial water storage anomalies during GRACE and GRACE-FO gap" * |
| ZHUOYA SHI等: "Improving the Reliability of the Prediction of Terrestrial Water Storage in Yunnan Using the Artificial Neural Network Selective Joint Prediction Model" * |
| 吴豪等: "利用GRACE重力测量反演地下水给水度——以海河流域为例" * |
| 周旭华;吴斌;彭碧波;许厚泽;: "全球水储量变化的GRACE卫星检测", 地球物理学报 * |
| 李婉秋;王伟;章传银;杨强;冯伟;刘阳;: "利用GRACE卫星重力数据监测关中地区地下水储量变化" * |
| 禤键豪等: "联合GRACE与GRACE-FO反演2002～2020年长江流域陆地水储量变化" * |
| 郑秋月;陈石;: "应用GRACE卫星重力数据计算陆地水变化的相关进展评述" * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117152629A (en) * | 2023-08-23 | 2023-12-01 | 武汉大学 | Method and system for filling gravity vacancy data of drainage basin scale time-varying satellite |
| CN117152629B (en) * | 2023-08-23 | 2024-03-22 | 武汉大学 | Method and system for filling gravity vacancy data of drainage basin scale time-varying satellite |
Also Published As
| Publication number | Publication date |
|---|---|
| CN115630686B (en) | 2023-06-23 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Towner et al. | Assessing the performance of global hydrological models for capturing peak river flows in the Amazon basin | |
| Liu et al. | Spatiotemporal distributions of surface ozone levels in China from 2005 to 2017: A machine learning approach | |
| Soltani et al. | Review of assimilating GRACE terrestrial water storage data into hydrological models: Advances, challenges and opportunities | |
| Humphrey et al. | Using satellite-based terrestrial water storage data: a review | |
| CN113868855B (en) | Groundwater reserve change satellite gravity forward modeling method fusing water level data | |
| Llovel et al. | Global land water storage change from GRACE over 2002–2009; Inference on sea level | |
| Jin et al. | Re-estimation of glacier mass loss in Greenland from GRACE with correction of land–ocean leakage effects | |
| CN111192282B (en) | Lake and reservoir time sequence water level reconstruction method for lakeside zone virtual station | |
| Marti et al. | Monitoring the ocean heat content change and the Earth energy imbalance from space altimetry and space gravimetry | |
| CN115630686A (en) | Method for recovering land water reserve abnormity from satellite gravity data by machine learning | |
| CN113962056A (en) | Method for improving accuracy of water reserve change of GRACE land | |
| Li et al. | Ellipsoidal correction in GRACE surface mass change estimation | |
| Bergmann et al. | Climate-driven interannual ice mass evolution in Greenland | |
| Tregoning et al. | ANU GRACE data analysis: Characteristics and benefits of using irregularly shaped mascons | |
| Revel et al. | Assimilation of transformed water surface elevation to improve river discharge estimation in a continental-scale river | |
| CN114046774B (en) | Ground deformation continuous monitoring method integrating CORS network and multi-source data | |
| CN113268869B (en) | Method and system for monitoring change of earth surface quality | |
| Quinn et al. | Estimating weights for the use of time‐dependent gravity recovery and climate experiment data in constraining ocean models | |
| Chao et al. | Multi-Lagrange multiplier method to improve the region-specific GRACE estimation of water storage change in eleven sub-basins of the Yangtze River | |
| Stanešić | Assimilation system at DHMZ: Development and first verification results | |
| CN117152629B (en) | Method and system for filling gravity vacancy data of drainage basin scale time-varying satellite | |
| Zhou et al. | HUST-Grace2024: a new GRACE-only gravity field time series based on more than 20 years of satellite geodesy data and a hybrid processing chain | |
| Berezowski et al. | CPLFD-GDPT5: high-resolution gridded daily precipitation and temperature dataset for two largest Polish river basins | |
| Martin et al. | Product user manual for Ostia near real time level 4 SST products over the global ocean SST-GLO-SST-L4-NRT-OBSERVATIONS-010-001 | |
| CN115906404B (en) | Collaborative forward modeling method for groundwater reserve change under observation well missing condition |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |