CN112200050A - Lake water volume space-time change analysis method and device based on multi-source satellite data - Google Patents

Lake water volume space-time change analysis method and device based on multi-source satellite data Download PDF

Info

Publication number
CN112200050A
CN112200050A CN202011061648.1A CN202011061648A CN112200050A CN 112200050 A CN112200050 A CN 112200050A CN 202011061648 A CN202011061648 A CN 202011061648A CN 112200050 A CN112200050 A CN 112200050A
Authority
CN
China
Prior art keywords
lake
data
water
satellite
image
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.)
Pending
Application number
CN202011061648.1A
Other languages
Chinese (zh)
Inventor
林怡
超能芳
李鑫
张婷慧
宇洁
叶勤
谢光顺
张文豪
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Tongji University
Original Assignee
Tongji University
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Tongji University filed Critical Tongji University
Priority to CN202011061648.1A priority Critical patent/CN112200050A/en
Publication of CN112200050A publication Critical patent/CN112200050A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V20/00—Scenes; Scene-specific elements
    • G06V20/10—Terrestrial scenes
    • G06V20/13—Satellite images
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/24—Classification techniques
    • G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
    • G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N20/00—Machine learning
    • G06N20/10—Machine learning using kernel methods, e.g. support vector machines [SVM]
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/004—Artificial life, i.e. computing arrangements simulating life
    • G06N3/006—Artificial life, i.e. computing arrangements simulating life based on simulated virtual individual or collective life forms, e.g. social simulations or particle swarm optimisation [PSO]
    • 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
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/045—Combinations of networks
    • 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
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/60—Analysis of geometric attributes
    • G06T7/62—Analysis of geometric attributes of area, perimeter, diameter or volume
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/40—Extraction of image or video features
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/10—Image acquisition modality
    • G06T2207/10032—Satellite or aerial image; Remote sensing
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/30—Subject of image; Context of image processing
    • G06T2207/30181—Earth observation

Abstract

The invention relates to a lake water volume space-time change analysis method and a device based on multisource satellite data, wherein the method specifically comprises the following steps: s1: acquiring a research area image of the lake through a satellite, preprocessing the research area image to obtain an image classification map, and acquiring the water level of the lake through the satellite; s2: acquiring time sequence data of the lake water volume according to the image classification map and the lake water level of the lake; inputting the image of the research area of the small lake into a KELM model to obtain an image classification diagram, and inputting the image of the research area of the large lake into a trained SVM model to obtain an image classification diagram; the training process is as follows: acquiring an experimental area image, dividing pixel points on the experimental area image into a water body and a non-water body to form a training sample, and training a KELM model and an SVM model by using the training sample, wherein parameter optimization is carried out on the KELM model by adopting a PSO algorithm. Compared with the prior art, the method has the advantages of good stability, good robustness, high efficiency and the like.

Description

Lake water volume space-time change analysis method and device based on multi-source satellite data
Technical Field
The invention relates to a lake water amount monitoring technology, in particular to a lake water amount space-time change analysis method and device based on multi-source satellite data.
Background
The lake water amount change is a main control factor for establishing a lake water amount balance relation and further analyzing water amount balance of different climate areas, and has important scientific significance for researching the influence of climate change and human activities on regional water circulation. The lake is a complex water circulation system, each hydrological process in the system has uncertainty, and great difficulty is brought to the improvement of the accuracy of the study on the water quantity change of the lake. The traditional water resource survey has various technical means mainly based on a hydrological observation station, the lake water capacity is calculated by combining lake bottom topographic data and field observation data, large-scale large-range accurate quantitative evaluation is difficult to perform, the data acquired by different types of space sensors has the advantages of rapidness, comprehensiveness, objectivity, visual information grasping and the like, and the method is almost the only and desirable method for the areas with severe environment, backward traffic and sparse ground monitoring station networks. In recent thirty years, with the development of aerospace technology, remote sensing and satellite geodetic measurement technology provides new methods for long-term observation of material migration and change, such as optical images, interference radar observation, satellite height measurement, gravity observation and the like, and the system accumulates satellite monitoring data and products of large-area rivers, lakes/reservoirs.
The detection of the lake water surface area is crucial to the research precision when the study of the lake water volume change is carried out, and the optical sensor is widely applied to the lake area detection and the related change analysis by virtue of the advantages of high time resolution, high spatial resolution, low cost and the like, and is also a main method for the study of the lake water area change at present. The key point for determining the area of the lake water area is the determination of the lake boundary, however, the lake contour is in an irregular state and has a large contrast with surrounding ground objects, so that the discrimination between a water body and a non-water body can be enhanced by using a remote sensing water body extraction method, and the lake area information can be conveniently and quickly extracted. The traditional methods comprise a single-waveband threshold value method, a multi-waveband method and the like, the methods are used for extracting water bodies, more logic judgments are used in the operation process, and the automation degree is low. With the development of artificial intelligence and machine learning theory and technology, researchers have proposed various artificial neural network methods to solve the classification problem, mainly with radial basis neural networks, multi-layer perceptron networks, self-organizing mapping networks, wavelet neural networks and other various neural network classification methods, and have proposed various optimization methods for these neural networks. Although neural networks have a strong learning ability, they are slow to learn and tend to be limited to local minima, making them less than satisfactory in many engineering studies.
Disclosure of Invention
The invention aims to overcome the defects in the prior art and provide a lake water amount space-time change analysis method and device based on multi-source satellite data, which have the advantages of good stability, good robustness and high efficiency.
The purpose of the invention can be realized by the following technical scheme:
a lake water volume space-time change analysis method based on multi-source satellite data specifically comprises the following steps:
s1: acquiring a research area image of the lake through a satellite, classifying water bodies and non-water bodies in the research area image to obtain an image classification diagram, and acquiring the water level of the lake through the satellite;
s2: acquiring time sequence data of the lake water volume by adopting a geometric analysis method or a physical analysis method according to the image classification diagram and the lake water level of the lake;
the specific process of step S1 is as follows:
respectively enabling the lakes to be small lakes and large lakes according to prior information, if the lakes are small lakes, inputting images of the research areas of the lakes into the trained KELM model to obtain image classification maps, and otherwise, inputting the images of the research areas of the lakes into the trained SVM model to obtain image classification maps;
the calculation formula of the output function f (x) of the KELM model is as follows:
Figure BDA0002712584930000021
Kmix(x,x')＝KGauss(xGauss,xGauss')+KPoly(xPoly,xPoly')
ΩELM＝HHT＝h(xi)·h(xj)＝K(xi,xj)
wherein, KmixAs a mixed kernel function, KGaussIs a Gaussian kernel function, xGaussIs the input value of the Gaussian kernel function, C is the regularization parameter, KPolyFor polynomial kernel, H is the output matrix of the hidden layer of the KELM model, xPolyIs the input value of the polynomial kernel function;
the training process of the KELM model and the SVM model is specifically as follows: acquiring an experimental area image, dividing pixel points on the experimental area image into a water body and a non-water body to form a training sample, and training a KELM model and an SVM model by using the training sample, wherein a PSO algorithm is adopted to KGaussAnd KPolyAnd optimizing the parameters to obtain an optimal solution of the parameters.
Further, the geometric analysis method specifically comprises:
obtaining the lake area according to the image classification diagram, namely the pixel sum of the water body in the image classification diagram, wherein the calculation formula is as follows:
Area＝points×resolution×resolution
wherein, Area is the lake Area, points is the sum of pixels classified as water in the image classification map, resolution is the spatial resolution of the image in the research Area;
obtaining a relative water level WLALL based on the lowest lake water level, selecting lake areas with similar obtaining dates, fitting a first mathematical model of the lake areas relative to the WLALL, carrying out differential calculation on the first mathematical model to obtain a second mathematical model of the water quantity WVALL above the lowest water level relative to the WLALL, and obtaining time series data of the lake water quantity according to the WLALL at different times and the corresponding WVALL.
Further, the physical analysis method is based on the GRACE satellite observation data, combines a least square method mathematical model and an optimal prior model, and specifically comprises the following steps:
dividing the lake into N sub-watersheds, wherein a least square mathematical model is as follows:
Figure BDA0002712584930000031
wherein, yt(x) For the quality of the surface of each basin, x is the geographical position coordinate, αtIs a uniform scale factor, alphatjIs a uniform scale factor of the jth sub-watershed at the time t, omega is a research area, epsilont(x) As fitting error, bj(x) For each sub-basin, said bj(x) The calculation formula of (2) is as follows:
Figure BDA0002712584930000032
wherein, inside base represents that the geographic position coordinate is located in the sub-flow domain, and else where represents that the geographic position coordinate is located outside the sub-flow domain.
Using a Gaussian smoothing function on yt(x) The calculation formula of (2) is smoothed to obtain the mass change z after Gaussian smoothingtThe calculation formula is as follows:
zt＝Aαt+γt
wherein A is bj(x) T is the fitting error after Gaussian smoothing;
obtaining the area of the lake according to the image classification diagram of the lake, obtaining the covariance sigma of the water quality change of each sub-basin and the optimal prior covariance W of the water quality change of the lake according to the area of each sub-basin and the lake level, and solving alpha at the time ttThe calculation formula is as follows:
αt＝(σ-1ATA+W-1)-1ATσ-1zt
according to alphatSolving the water quality of each sub-basin
Mtj(x)＝ρμjαtjbj(x)
Wherein x is a geographical position coordinate, αtIs a uniform scale factor, alphatjIs the jth alphatRho is the density of lake water, mujIs the area of the jth sub-basin;
and then, obtaining the water quality change of the lake according to the water quality change of each sub-basin, and obtaining the time sequence data of the lake water volume according to the water quality change.
Further, the lake water level obtaining process specifically comprises:
the method includes the steps that a plurality of groups of GDR data are obtained periodically according to Jason-1, Jason-2 and Jason-3 height measurement satellites, Jason-1, Jason-2 and Jason-3 have the same satellite orbit height, orbit inclination angle and revisit period, data continuity is guaranteed, one lake water level measurement value H is obtained according to each group of GDR data1Calculate all H in one cycle1Is taken as the lake level in the period, and H is1The calculation formula is as follows:
H1＝Ralt-R-ΔR
wherein，RaltThe height of the altimeter satellite relative to the reference ellipsoid is measured, R is the distance measurement value from the altimeter to the lake surface, and delta R is the error correction value.
Further, the calculation formula of Δ R is:
ΔR＝Wet+Dry+Iono+Sea+Set+Pol
wherein, Wet term correction value of troposphere is Wet, Dry term correction value of troposphere is Dry term correction value of troposphere, Iono is ionosphere correction value, Sea state deviation correction value is Sea state deviation correction value, Set is solid tide correction value, Pol is extreme tide correction value, and Wet, Dry, Iono, Sea, Set and Pol are respectively data items of rad _ Wet _ trop _ corr, model _ Dry _ trop _ corr, Iono _ corr _ alt _ ku, Sea _ state _ bias _ ku, solid _ earth _ tide and pole _ tide in GDR data.
Further, in order to fully utilize GDR data of valid data, H which does not satisfy any of the following judgment conditions is eliminated1：
The latitude of the height measuring satellite is between the latitudes covered by the lakes;
surface _ type in GDR data is equal to 1;
the qual _ alt _1hz _ range _ ku in the GDR data is equal to 0;
dry, Iono, Sea, Set and Pol are all in a Set range;
the difference between the GDR data in the same period and the average value of the period is within the error range of three times
Each group of GDR internal data has a plurality of sampling points, and H is taken1Points within error of three times the mean of the period are the significant points.
Further, due to the influence of instruments, satellite orbits and revising periods, some system deviation always exists between different satellite altimeters, the average difference of GDR data between Jason-1 and Jason-2 is added to the GDR data of Jason-1, and the average difference of GDR data between Jason-2 and Jason-3 is added to the GDR data of Jason-3, so that correction of the system error is completed.
And further, extracting the water surface information of the small lake through a Landsat remote sensing satellite, and performing radiation correction, wherein the radiation correction comprises radiometric calibration and atmospheric correction to obtain a research area image.
And further, extracting water surface information of the large lake through 500m earth surface reflectivity data synthesized by a Terra/MODIS satellite for 8 days, if the water surface information comprises two or more local images, firstly inlaying all the images into a complete image, and then carrying out re-projection and cutting on the complete image to obtain a research area image.
An analysis apparatus employing the analysis method comprises a memory storing a computer program and a processor calling the program instructions to perform any of the methods.
Compared with the prior art, the invention has the following beneficial effects:
(1) the invention firstly divides the lake into small-sized lake and large-sized lake, classifies the area image of the small-sized lake by adopting a KELM model, classifies the area image of the large-sized lake by adopting an SVM model, the KELM model has relatively low calculation speed and high accuracy, the SVM model has high calculation speed but relatively low accuracy, and balances the speed and the accuracy of the area image classification, wherein the KELM model is an ELM model based on a kernel function, does not need to know a hidden layer output matrix, can effectively avoid the use of random parameters, has good stability and robustness, has locality of a Gaussian kernel function, has better anti-interference capability on noise existing in data, takes a polynomial kernel function as a global kernel function, combines the kernel function of the invention with the Gaussian kernel function and the polynomial kernel function, has good fitting effect, and ensures high classification accuracy of the area image of the lake, the detection precision is high, the PSO algorithm is adopted to optimize the parameters of the kernel function, the method is suitable for the problems of multivariable, nonlinearity, discontinuity and infinitesimal classification of lake water bodies and non-water bodies, the operation is simple, and the algorithm search efficiency is high;
(2) according to the method, the height measurement data of a Jason series satellite is used for calculating the water level of the lake, firstly, the ellipsoidal height of the water surface of the lake is calculated according to the basic principle of satellite height measurement, a data editing rule is formulated according to the characteristics of inland water areas to eliminate abnormal sampling data, all qualified sampling data in the same period are averaged to obtain the average height of the period, water level changes in different time scales are obtained on the basis, and available long-time sequence water level data are obtained, so that the precision is high;
(3) according to the method, the lake area is obtained according to the image classification diagram, the first mathematical model of the lake area about the relative water level based on the lowest lake water level is fitted, differential calculation is carried out on the first mathematical model, the second mathematical model of the water quantity above the lowest water level about the relative water level is obtained, time series data of the lake water quantity are further obtained, and large-scale and large-range quantitative estimation of the water quantity by using a remote sensing means is realized;
(4) the method adopts an improved least square method, combines optimal prior covariance information and GRACE satellite observation data to construct a method equation, inverts the area quality change, determines the lake water volume change, can improve the accuracy of the GRACE data in inverting the small area land water storage volume change, and realizes accurate estimation of the lake water volume change from the physical quality perspective.
Drawings
FIG. 1 is a flow chart of a method of the present invention;
FIG. 2 is a view of the elevation of the lake surface of the Victoria lake before the systematic deviation is eliminated;
FIG. 3 is a graph of Victoria lake surface height for Jason-1 and Jason-2 overlap periods;
FIG. 4 is a graph of Victoria lake surface height for Jason-2 and Jason-3 overlap periods;
FIG. 5 is a view showing the height of the lake surface of Victoria lake after the system deviation is eliminated;
FIG. 6 is a scatter plot of WLALL and GRLM data sets;
FIG. 7 is a line graph of WLALL and GRLM datasets;
FIG. 8 is a line graph showing the change in the amount of lake water in Victoria;
FIG. 9 is a scatter diagram of the lake water volume obtained by geometric analysis and physical analysis;
fig. 10 is a water volume change wavelet variance plot of victoria lake.
Detailed Description
The invention is described in detail below with reference to the figures and specific embodiments. The present embodiment is implemented on the premise of the technical solution of the present invention, and a detailed implementation manner and a specific operation process are given, but the scope of the present invention is not limited to the following embodiments.
Example 1
A lake water volume space-time change analysis method based on multi-source satellite data is shown in figure 1 and specifically comprises the following steps:
s1: acquiring a research area image of the lake through a satellite, classifying water bodies and non-water bodies in the research area image to obtain an image classification diagram, and acquiring the water level of the lake through the satellite;
s2: obtaining time sequence data of the lake water volume by adopting a geometric analysis method according to the image classification diagram and the lake water level of the lake;
the specific process of step S1 is as follows:
respectively enabling the lakes to be small lakes and large lakes according to prior information, if the lakes are small lakes, inputting images of the research areas of the lakes into the trained KELM model to obtain image classification maps, and otherwise, inputting the images of the research areas of the lakes into the trained SVM model to obtain image classification maps;
the output function f (x) of the KELM model is calculated by the formula:
Figure BDA0002712584930000071
Kmix(x,x')＝KGauss(xGauss,xGauss')+KPoly(xPoly,xPoly')
ΩELM＝HHT＝h(xi)·h(xj)＝K(xi,xj)
wherein, KmixAs a mixed kernel function, KGaussIs a Gaussian kernel function, xGaussIs the input value of the Gaussian kernel function, C is the regularization parameter, KPolyFor polynomial kernel, H is the output matrix of the hidden layer of the KELM model, xPolyIs the input value of the polynomial kernel function;
the training process of the KELM model and the SVM model is specifically as follows: collected fruitThe method comprises the steps of examining an area image, dividing pixel points on the experiment area image into a water body and a non-water body to form training samples, and training a KELM model and an SVM model by using the training samples, wherein a PSO algorithm is adopted to carry out KGauss、KPolyAnd the regularization parameter C is used for optimizing the parameters to obtain the optimal solution of the parameters, and the searching efficiency is high.
The geometric analysis method specifically comprises the following steps:
the method comprises the steps of obtaining lake areas according to an image classification diagram, setting constant water quantity of the lowest water level of a lake to be zero, obtaining a relative water level WLALL based on the lowest lake water level, selecting lake areas with similar obtaining dates, fitting a first mathematical model of the lake areas relative to the WLALL, carrying out differential calculation on the first mathematical model, obtaining a second mathematical model of water quantity WVALL above the lowest water level relative to the WLALL, and obtaining time sequence data of the lake water quantity according to the WLALL at different times and the corresponding WVALL.
The traditional water level data mainly depends on the data of a hydrological station, the coverage range is small, and the accuracy is low. Based on the influence of environmental factors such as weather on the satellite measurement water level, the acquisition process of the lake water level specifically is as follows:
the method comprises the steps that a plurality of groups of GDR data are acquired periodically according to Jason-1, Jason-2 and Jason-3 height measurement satellites, due to the influence of instruments, satellite orbits and revisiting periods, some system deviation always exists between different satellite height measurement instruments, the average difference of the GDR data between Jason-1 and Jason-2 is added to the GDR data of Jason-1, the average difference of the GDR data between Jason-2 and Jason-3 is added to the GDR data of Jason-3, and correction of system errors is completed;
respectively obtaining a lake water level measured value H according to each group of GDR data1Calculate all H in one cycle1Is taken as the lake level in the period, H1The calculation formula is as follows:
H1＝Ralt-R-ΔR
wherein R isaltThe height of the altimeter satellite relative to the reference ellipsoid is measured, R is the distance measurement value from the altimeter to the lake surface, and delta R is the error correction value.
Δ R is calculated as:
ΔR＝Wet+Dry+Iono+Sea+Set+Pol
wherein, Wet term correction value of troposphere is Wet, Dry term correction value of troposphere is Dry term correction value of troposphere, Iono is ionosphere correction value, Sea state deviation correction value is Sea state deviation correction value, Set is solid tide correction value, Pol is extreme tide correction value, and Wet, Dry, Iono, Sea, Set and Pol are rad _ Wet _ trop _ corr, model _ Dry _ trop _ corr, Iono _ corr _ alt _ ku, Sea _ state _ bias _ ku, solid _ earth _ tide and pole _ tide data items in GDR data respectively.
The correction values are required to be within the effective range, as shown in table 1:
TABLE 1 threshold value table of correction values of respective items in error correction values
Figure BDA0002712584930000081
Removing H which does not meet any judgment condition1：
The latitude of the height measuring satellite is between the latitudes covered by the lakes;
surface _ type in the GDR data is equal to 1, which indicates that the ground type is inland lake or water area;
the qual _ alt _1Hz _ range _ Ku of a Ku waveband 1Hz altimeter data valid flag in the GDR data is equal to 0; dry, Iono, Sea, Set and Pol are all in a Set range;
the difference between the GDR data in the same cycle and the average value of the cycle is within a three-fold error range.
Because each group of GDR data has a plurality of sampling points, H is taken1Points within error of three times the mean of the period are the significant points.
And finally, comparing the calculated lake water level with the GRLM data set, and analyzing and verifying the calculated water level elevation result.
The method comprises the steps of extracting water surface information of the small lake through a Landsat remote sensing satellite, and performing radiation correction, wherein the radiation correction comprises radiometric calibration and atmospheric correction, and the atmospheric correction is performed on a Landsat remote sensing image by adopting a FLAASH atmospheric correction method of ENVI5.1 software to obtain a research area image. When receiving electromagnetic wave radiant energy of a ground object, the Landsat remote sensing image sensor generates gray level deviation, namely, radiation error of a remote sensing image, and needs to be corrected, so that various distortions attached to a radiation brightness value are eliminated, and intrinsic radiation characteristics of the ground object are recovered.
Extracting water surface information of a large lake through 500m ground surface reflectivity data MOD09A1 synthesized by Terra/MODIS satellites for 8 days, if the water surface information comprises two or more local images, inlaying all the images into a complete image by using MODIS reproduction Tool software, and then carrying out Reprojection and cutting on the complete image to obtain a research area image.
ELM is an extreme learning machine, ELM only needs to set the number of the hidden layer node neurons of the network in the learning process, the weight vector parameters between the input layer and the hidden layer and the bias vector parameters on the hidden layer do not need to be adjusted repeatedly through iteration, the ELM model can generate a unique optimal solution, and the ELM model has the advantages of less training parameters, high training speed, global generalization capability and the like, but in the ELM learning process, the model can often perfectly fit the training data, but in the solving process, an overfitting phenomenon is easy to generate based on an empirical risk minimization criterion, meanwhile, the anti-gross error capability is poor, the fitting effect on the test data is not ideal, the random algorithm used by the standard ELM cannot ensure the stability and the persistence according to the learning theory, therefore, the classification model needs to be constrained in the classification process, the appropriate regularization parameters are calculated according to a data set, the regularization ELM model with more stability and better generalization performance is constructed, the ELM model based on the kernel function can solve the result without knowing the hidden layer output matrix, namely the activation function and the hidden layer node, the kernel function can effectively avoid the use of random parameters, thereby increasing the stability and robustness of the model, and the kernel function has different effects on the difference of the distance between a data point and a central point,
the types of kernel functions are divided into two categories: a local kernel function and a global kernel function. The Gaussian kernel function is one of radial basis functions, has better anti-interference capability on noise existing in data, and has strong locality. A gaussian function is generally defined as a monotonic function of the euclidean distance from a point in space to the center, with the effect often being local, and weaker the further away from the center. The invention combines the advantages of the Gaussian kernel function and the polynomial kernel function, combines the two types of kernel functions to be used as a mixed kernel function model, and has high precision.
The PSO algorithm is a group intelligent optimization algorithm, has no continuous and differentiable requirements on a function to be optimized, has great advantages on solving problems of multivariable, nonlinearity, discontinuity and insensitivity, optimizes Gaussian kernel function parameters and polynomial kernel function parameters by adopting the PSO algorithm, and is simple to operate and high in algorithm searching efficiency.
In this embodiment, a lake of the lake is taken as a research area, a landmass around the lake of 2015 is divided into six types of land objects including a water body, a construction land, a forest green land, a farmland, a bare land and a factory by selecting a landmass of Landsat-8/OLI image of day 3 of month 8 as experimental data, the ground resolution is 30m, 6 spectral bands are totally obtained, the operation period is 16 days, PSO-KELM, standard ELM, maximum likelihood classification MLC and SVM are respectively used for classification experiments, PSO-KELM is a KELM model based on a PSO algorithm, and an image-taking sample set in the experiments is as follows: 302 water bodies, 305 construction land, 306 forest land, 309 farmlands, 304 bare land and 315 factories. 1841 sample points are selected altogether, 1200 sample points are randomly selected as training samples, the rest 641 sample points are used as test samples, classification results of four classification methods are evaluated by utilizing a Kappa coefficient, the Kappa coefficient is an index for measuring classification accuracy based on a confusion matrix, and the evaluation table is shown in table 2:
TABLE 2 Classification results evaluation table for lake image of lake of the lake
Figure BDA0002712584930000101
In the image experiment of the lake region in the lake region, the overall accuracy of the PSO-KELM and SVM method is higher than that of the MLC and standard ELM method, the overall accuracy of the PSO-KELM is 92.67 percent at most, in the aspect of computing time, the time consumption of the standard ELM is the least, the fact that the computing time of the classification of the ELM and the PSO-KELM is related to the size of the image is shown, when the number of input pixels of the image is small, the PSO-KELM computing speed is high, and the classification accuracy is high.
In this embodiment, a landrace-8/OLI image obtained in 5 months in 2017 is selected, an SVM, a traditional ELM and a PSO-KELM are used to classify the polder field experimental area of the ring nest lake drainage basin, the types of the land features in the experimental area are divided into five types, namely buildings, bare land, forest land, cultivated land and water, a point pixel is taken as a unit, a representative pixel point of the land feature is selected, the range extends over the whole experimental area, and the specific sample condition is as follows: 604 buildings, 472 forest lands, 587 cultivated lands, 582 bare lands and 443 water bodies, wherein the total number of samples is 2688, 1793 training samples and 895 testing samples;
parameter optimization using the PSO algorithm with Gaussian kernel parameter σ in KELM1340.3857, the polynomial kernel function parameter d is 6, and the classification results of SVM, conventional ELM and PSO-KELM are evaluated by using Kappa coefficient, as shown in Table 3:
table 3 ring nest lake image classification result evaluation table
Figure BDA0002712584930000102
As can be seen from Table 3, the accuracy of the PSO-KELM, SVM and standard ELM algorithms decreases in sequence.
In the embodiment, victoria lake is taken as a research area, MOD09a1 images from 2003 to 2017 are obtained, the ground resolution of the image data is 500m, 7 spectral bands are totally obtained, and the time resolution is 8 days. Firstly, image mosaicing, re-projecting and image cutting are carried out on an image, then 30 ground object sample sets of a water body and a non-water body are uniformly selected from the image, 15 ground object sample sets are randomly selected from the sample sets to serve as training sample sets, the rest ground object sample sets serve as verification sample sets, an SVM classification algorithm is used for classifying 104 scene MODIS images in total, a confusion matrix is used for carrying out precision evaluation on classification results of the 104 scene images, the overall precision is over 95%, the Kappa coefficient is over 0.93, and the SVM classification algorithm can better extract the water body information of Victoria lake.
The area of the victoria lake is extracted,
Area＝points×resolution×resolution
wherein, Area is the lake Area, points is the sum of pixels classified as water in the image classification map, resolution is the spatial resolution of the image in the research Area;
GDR data from Jason-1, Jason-2 and Jason-3 altimetric satellites are shown in Table 4:
TABLE 4 Jason satellite altimetry data sheet
Figure BDA0002712584930000111
As shown in figure 2, the satellite measurement period is 10 days, each satellite can averagely obtain 27 effective data points in each period, and the lake level H in the same period is measured for each satellite1And taking the average value to obtain water level line graphs of the Victoria lake in different years.
There are two overlapping periods in the 15 years. One overlap period is 21 cycles of Jason-1 and Jason-2 between 2008 and 1 2009 as in FIG. 3, and another overlap period includes 24 cycles of Jason-2 and Jason-3 between 2016 and 2 to 2016 and 9 as in FIG. 2. To eliminate the bias between the two different satellite elevation missions, the bias was set as the mean bias between the overlapping data sets, Jason-1 and Jason-2 mean biases were added to all Jason-1 results to eliminate the bias between the two, the bias between Jason-2 and Jason-3 was calibrated in the same way, Jason-2 and Jason-3 mean biases were added to all Jason-3 results, as in FIG. 5, to form the final Victoria lake height time series, with the maximum lake height observed being 1119.61m, occurring on day 11/5/2016, the minimum lake height being 1117.30 meters, occurring on day 21/10/2006, based on eliminating the systematic biases. The maximum change in lake surface height over the 15 years is 2.31 m. The lowest lake height between 15 years is subtracted from all the lake height results to obtain WLALL, and the WLALL obtained by calculation is compared with the GRLM data set for verification, as shown in FIGS. 6 and 7, and the correlation coefficient R20.9435, mean difference 0.8879m, standard deviation close to 0, indicatingThe two have high correlation, and the variation trends are completely consistent and consistent.
Setting the lowest water level constant water quantity of the lake in 2017 of 2003 + as zero, selecting Area areas and water level WLALL with similar dates for matching, establishing a first mathematical model about the Area areas and the WLALL of the lake, selecting 55 pairs of data of the Area areas and the water level of the lake for fitting, wherein the recording date interval of each pair of the Area areas and the water level of the lake is less than 1 day, and because the accuracy of a field measurement data verification model is lacked, randomly selecting 49 pairs of data for establishing a relational model, and using the remaining 6 pairs of data for evaluating the accuracy of the established model, respectively establishing linear, polynomial, power and exponential function model relations according to time sequence data of the Area of the Victoria lake and the elevation of the relative water level, and eliminating gross errors to obtain a first mathematical model, wherein the first mathematical model specifically comprises the following steps:
Area＝422.6e0.3805WLALL+65290
the fit relationship was verified using the 6 pairs of data retained randomly, and the results are shown in table 5:
TABLE 5 accuracy verification table of fitting function relation
Figure BDA0002712584930000121
The average absolute error of the two is 35.1418km2The average value of the relative errors is about 0.05%, which shows that the water area calculated by adopting a fitting relational expression is basically consistent with the result of remote sensing measurement, and the expression can effectively describe the relation between the area and WLALL.
And carrying out differential calculation on the first mathematical model to obtain a second mathematical model about WLALL about the water quantity WVALL above the lowest water level, which specifically comprises the following steps:
WVALL＝1110.6439e0.3805WLALL+6.529WLALL×104
as shown in fig. 8, a line graph of the change in water amount of victoria lake was obtained according to the second mathematical model.
And (3) calculating a month, season and year time sequence of relative water quantities of the Doria lake in 2003-2017, and carrying out correlation analysis on the change period of the water quantities of the lake.
As shown in fig. 10, wavelet analysis is performed on victoria lake by using Morlet wavelet based on a water volume change month time sequence of satellite height measurement and remote sensing images to obtain a wavelet variance graph of victoria lake, which can reflect the distribution of wave energy along with the cycle scale of the water volume time sequence and can be used for determining a main cycle existing in the water volume evolution process, wherein the wavelet variance graph has three peaks respectively corresponding to cycle scales of 18 months, 9 months and 6 months. Wherein, the maximum peak value corresponds to the cycle scale of 18 months, which shows that the cycle oscillation of about 18 months is the strongest, and is the first main cycle of water quantity change; the period scale of 9 months corresponds to a second peak value which is a second main period of water volume change, the time-frequency structure of the lake can be known through the wavelet variance graph, and the change period of the lake water volume can be calculated.
Example 2
In this embodiment, step S2 is to obtain time series data of the lake water volume by physical analysis method according to the image classification map and the lake level of the lake, and the rest is the same as in embodiment 1.
The physical analysis method is based on a GRACE satellite, combines a least square method mathematical model and an optimal prior model, and specifically comprises the following steps:
the Victoria lake is divided into 4 sub-watersheds, and a least square mathematical model is as follows:
Figure BDA0002712584930000131
wherein, yt(x) For the quality of the surface of each basin, x is the geographical position coordinate, αtIs a uniform scale factor, alphatjIs the jth alphatOmega is the area of investigation, epsilont(x) As fitting error, bj(x) For each sub-basin coefficient, the formula is calculated as:
Figure BDA0002712584930000132
wherein, inside base represents that the geographic position coordinate is located in the sub-flow domain, and else where represents that the geographic position coordinate is located outside the sub-flow domain.
Because the time-varying gravity field spherical harmonic coefficient high-order item determined by satellite observation data has errors, a least square method mathematical model is improved by utilizing a Gaussian smoothing function, and the mass change z after Gaussian smoothing is obtainedtThe calculation formula is as follows:
zt＝Aαt+γt
wherein A is bj(x) In matrix form of γtFitting error after Gaussian smoothing;
obtaining the area of the lake according to the image classification diagram of the lake, obtaining the covariance sigma of the water quality change of each sub-basin and the optimal prior covariance W of the water quality change of the lake according to the area of each sub-basin, the prior information of the lake water level and the like, and solving alpha according to the optimal prior modeltThe calculation formula is as follows:
αt＝(σ-1ATA+W-1)-1ATσ-1zt
according to alphatSolving the water quality of each sub-basin, wherein the calculation formula is as follows:
Mtj(x)＝ρμjαtjbj(x)
wherein M istj(x) Is the water quality of the jth sub-basin at the time t, x is the geographic position coordinate, alphatIs a uniform scale factor, alphatjIs the jth alphatRho is the density of lake water, mujIs the area of the jth sub-basin;
and then the water quality change of the lake is obtained according to the water quality change of each sub-basin.
The water volume change of the Victoria lake in 2002-2016 and 12 months is obtained by a physical analysis method based on a GRACE satellite, as shown in figure 8, the water volume change of the Victoria lake determined by the two methods has good consistency, the period and the trend are consistent, 81 pairs of WVALLs with the interval of less than two days in the daily period are selected to be compared with the water volume of the lake obtained based on the GRACE satellite, the analysis result is shown in figure 9, the change trends are consistent,coefficient of correlation R20.8448, victoria lake surface water changes are also shown to be the primary cause of land water change signals detected by GRACE.
Example 3
An analysis apparatus employing the analysis method in embodiment 1, comprising a memory storing a computer program and a processor calling the program instructions to be able to execute any of the methods.
Embodiments 1, 2, and 3 provide a method and an apparatus for analyzing spatial and temporal changes of lake water volume based on multi-source satellite data, which break through the bottleneck of lack of ground hydrological observation data and underwater lake basin measurement data, can quickly and periodically acquire changes of lake area and water level, provide a convenient means for mastering the dynamic change rule of lakes and dynamic monitoring and protection of water resources, and have important scientific significance and application value for timely mastering the aspects of regional water volume balance condition, sustainable utilization of water resources, and the like; reliable water volume change results can be effectively obtained by combining multi-source satellite observation data, and lake water volume change monitoring and change course reconstruction are realized under the condition of lacking hydrological data and lake bottom terrain data.
The foregoing detailed description of the preferred embodiments of the invention has been presented. It should be understood that numerous modifications and variations could be devised by those skilled in the art in light of the present teachings without departing from the inventive concepts. Therefore, the technical solutions available to those skilled in the art through logic analysis, reasoning and limited experiments based on the prior art according to the concept of the present invention should be within the scope of protection defined by the claims.

Claims (10)

1. A lake water volume space-time change analysis method based on multi-source satellite data is characterized by comprising the following steps:
s1: acquiring a research area image of the lake through a satellite, preprocessing the research area image to obtain an image classification map, and acquiring the water level of the lake through the satellite;
s2: acquiring time sequence data of the lake water volume according to the image classification map and the lake water level of the lake;
the specific process of step S1 is as follows:
respectively enabling the lakes to be small lakes and large lakes according to prior information, if the lakes are small lakes, inputting images of the research areas of the lakes into the trained KELM model to obtain image classification maps, and otherwise, inputting the images of the research areas of the lakes into the trained SVM model to obtain image classification maps;
the calculation formula of the output function f (x) of the KELM model is as follows:
Figure FDA0002712584920000011
Kmix(x,x')＝KGauss(xGauss,xGauss')+KPoly(xPoly,xPoly')
ΩELM＝HHT＝h(xi)·h(xj)＝K(xi,xj)
wherein, KmixAs a mixed kernel function, KGaussIs a Gaussian kernel function, xGaussIs the input value of the Gaussian kernel function, C is the regularization parameter, KPolyFor polynomial kernel, H is the output matrix of the hidden layer of the KELM model, xPolyIs the input value of the polynomial kernel function;
the training process of the KELM model and the SVM model is specifically as follows: acquiring an experimental area image, dividing pixel points on the experimental area image into a water body and a non-water body to form a training sample, and training a KELM model and an SVM model by using the training sample, wherein a PSO algorithm is adopted to KGaussAnd KPolyAnd optimizing the parameters to obtain an optimal solution of the parameters.
2. The method for analyzing the spatial-temporal change of the lake water volume based on the multi-source satellite data as claimed in claim 1, wherein the step S2 adopts a geometric analysis method, specifically:
obtaining the lake area according to the image classification diagram, obtaining a relative water level WLALL based on the lowest lake water level, fitting a first mathematical model of the lake area about WLALL, carrying out differential calculation on the first mathematical model, obtaining a second mathematical model of the water quantity WVALL above the lowest water level about WLALL, and obtaining time sequence data of the lake water quantity according to the WLALL at different times and the corresponding WVALL.
3. The method for analyzing the spatial-temporal change of the lake water volume based on the multi-source satellite data as claimed in claim 1, wherein the step S2 adopts a physical analysis method based on a GRACE satellite, and specifically comprises the following steps:
dividing the lake into N sub-basins, obtaining the area of the lake according to an image classification chart of the lake, obtaining the covariance sigma of the water quality change of each sub-basin and the optimal prior covariance W of the water quality change of the lake according to the area of each sub-basin and the water level of the lake, and solving a uniform scale factor alphatThe calculation formula is as follows:
αt＝(σ-1ATA+W-1)-1ATσ-1zt
wherein A is a basin coefficient matrix of the sub-basins, obtained by GRACE satellite according to alphatSolving the water quality of each sub-basin, wherein the calculation formula is as follows:
Mtj(x)＝ρμjαtjbj(x)
wherein M istj(x) Is the water quality of the jth sub-basin at the time t, x is the geographic position coordinate, alphatjIs a uniform scale factor of the jth sub-basin at the time t, and rho is the density of the lake water, mujIs the area of the jth sub-basin;
and then, obtaining the water quality change of the lake according to the water quality change of each sub-basin, and obtaining the time sequence data of the lake water volume according to the water quality change.
4. The method for analyzing the spatial-temporal change of the lake water volume based on the multi-source satellite data as claimed in claim 1, wherein the lake water level obtaining process specifically comprises:
elevation measurement according to Jason-1, Jason-2 and Jason-3Periodically acquiring a plurality of groups of GDR data, wherein the data comprises the height R of the altimeter satellite relative to the reference ellipsoidaltAnd the distance measurement value R from the altimeter to the lake surface, and respectively obtaining a lake water level measurement value H according to each group of GDR data1Calculate all H in one cycle1Is taken as the lake level in the period, and H is1The calculation formula is as follows:
H1＝Ralt-R-ΔR
where Δ R is an error correction value.
5. The method for analyzing the spatial-temporal change of the lake water volume based on the multisource satellite data as claimed in claim 4, wherein the calculation formula of Δ R is as follows:
ΔR＝Wet+Dry+Iono+Sea+Set+Pol
wherein, Wet term correction value of troposphere is Wet, Dry term correction value of troposphere is Dry term correction value of troposphere, Iono is ionosphere correction value, Sea state deviation correction value is Sea state deviation correction value, Set is solid tide correction value, Pol is extreme tide correction value, and Wet, Dry, Iono, Sea, Set and Pol are respectively data items of rad _ Wet _ trop _ corr, model _ Dry _ trop _ corr, Iono _ corr _ alt _ ku, Sea _ state _ bias _ ku, solid _ earth _ tide and pole _ tide in GDR data.
6. The method for analyzing spatial and temporal variation of lake water volume based on multi-source satellite data as claimed in claim 4, wherein H which does not meet any judgment condition is removed1：
The latitude of the height measuring satellite is between the latitudes covered by the lakes;
surface _ type in GDR data is equal to 1;
the qual _ alt _1hz _ range _ ku in the GDR data is equal to 0;
dry, Iono, Sea, Set and Pol are all in a Set range;
the difference between the GDR data in the same cycle and the average value of the cycle is within a three-fold error range.
7. The method of claim 4, wherein the average difference of GDR data between Jason-1 and Jason-2 is added to GDR data of Jason-1, and the average difference of GDR data between Jason-2 and Jason-3 is added to GDR data of Jason-3.
8. The method for analyzing the spatial-temporal variation of the lake water volume based on the multisource satellite data as claimed in claim 1, wherein the water surface information of the small lake is extracted through a Landsat remote sensing satellite, and radiation correction is performed, wherein the radiation correction comprises radiation calibration and atmospheric correction, and a research area image is obtained.
9. The method for analyzing the spatial-temporal variation of the lake water volume based on the multisource satellite data as claimed in claim 1, wherein the water surface information of the large lake is extracted through 500m earth surface reflectivity data synthesized by a Terra/MODIS satellite for 8 days, if the water surface information comprises two or more local images, all the images are firstly embedded into a complete image, and then the complete image is re-projected and cut to obtain the image of the research area.
10. An analysis apparatus employing the analysis method according to claim 1, comprising a memory storing a computer program and a processor invoking the program instructions to be able to perform the method according to any one of claims 1 to 9.
CN202011061648.1A 2020-09-30 2020-09-30 Lake water volume space-time change analysis method and device based on multi-source satellite data Pending CN112200050A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202011061648.1A CN112200050A (en) 2020-09-30 2020-09-30 Lake water volume space-time change analysis method and device based on multi-source satellite data

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202011061648.1A CN112200050A (en) 2020-09-30 2020-09-30 Lake water volume space-time change analysis method and device based on multi-source satellite data

Publications (1)

Publication Number Publication Date
CN112200050A true CN112200050A (en) 2021-01-08

Family

ID=74013542

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202011061648.1A Pending CN112200050A (en) 2020-09-30 2020-09-30 Lake water volume space-time change analysis method and device based on multi-source satellite data

Country Status (1)

Country Link
CN (1) CN112200050A (en)

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112884342A (en) * 2021-03-10 2021-06-01 陕西九州遥感信息技术有限公司 Water color satellite atmospheric layer top radiation product quality evaluation and cross calibration method
CN114279415A (en) * 2021-12-24 2022-04-05 水利部交通运输部国家能源局南京水利科学研究院 Method for calculating and calibrating river and lake water level based on satellite height measurement data
CN115630686A (en) * 2022-10-11 2023-01-20 首都师范大学 Method for recovering land water reserve abnormity from satellite gravity data by machine learning
CN117113211A (en) * 2023-01-16 2023-11-24 杭州市水务集团有限公司 Intelligent water use monitoring system and method for solitary old people

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103063202A (en) * 2012-12-30 2013-04-24 同济大学 Cyanobacteria biomass spatial-temporal change monitoring and visualization method based on remote sensing image
CN107767946A (en) * 2017-09-26 2018-03-06 浙江工业大学 Breast cancer diagnosis system based on PCA and PSO KELM models
CN108153146A (en) * 2017-12-12 2018-06-12 东北大学 A kind of polynary molten steel quality MFA control system and method for blast furnace
CN111192282A (en) * 2019-12-19 2020-05-22 中国科学院南京地理与湖泊研究所 Lake and reservoir time sequence water level reconstruction method for lakeside zone virtual station
CN111504424A (en) * 2020-06-17 2020-08-07 水利部交通运输部国家能源局南京水利科学研究院 Lake water storage variable quantity monitoring method based on remote sensing
CN111695440A (en) * 2020-05-21 2020-09-22 河海大学 GA-SVR lake level measurement and prediction method based on radar altimeter

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103063202A (en) * 2012-12-30 2013-04-24 同济大学 Cyanobacteria biomass spatial-temporal change monitoring and visualization method based on remote sensing image
CN107767946A (en) * 2017-09-26 2018-03-06 浙江工业大学 Breast cancer diagnosis system based on PCA and PSO KELM models
CN108153146A (en) * 2017-12-12 2018-06-12 东北大学 A kind of polynary molten steel quality MFA control system and method for blast furnace
CN111192282A (en) * 2019-12-19 2020-05-22 中国科学院南京地理与湖泊研究所 Lake and reservoir time sequence water level reconstruction method for lakeside zone virtual station
CN111695440A (en) * 2020-05-21 2020-09-22 河海大学 GA-SVR lake level measurement and prediction method based on radar altimeter
CN111504424A (en) * 2020-06-17 2020-08-07 水利部交通运输部国家能源局南京水利科学研究院 Lake water storage variable quantity monitoring method based on remote sensing

Non-Patent Citations (3)

* Cited by examiner, † Cited by third party
Title
YI LIN ET AL.: "Spatio-Temporal Analysis ofWetland Changes Using a Kernel Extreme Learning Machine Approach", 《REMOTE SENSING》 *
YI LIN ET AL.: "Water Volume Variations Estimation and Analysis Using Multisource Satellite Data: A Case Study of Lake Victoria", 《REMOTE SENSING》 *
超能芳: "联合多源数据研究青藏高原及长江流域浅层物质迁移机制", 《中国博士学位论文全文数据库 基础科学辑》 *

Cited By (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112884342A (en) * 2021-03-10 2021-06-01 陕西九州遥感信息技术有限公司 Water color satellite atmospheric layer top radiation product quality evaluation and cross calibration method
CN112884342B (en) * 2021-03-10 2024-03-12 陕西九州遥感信息技术有限公司 Quality evaluation and cross calibration method for water-color satellite atmospheric roof radiation product
CN114279415A (en) * 2021-12-24 2022-04-05 水利部交通运输部国家能源局南京水利科学研究院 Method for calculating and calibrating river and lake water level based on satellite height measurement data
CN115630686A (en) * 2022-10-11 2023-01-20 首都师范大学 Method for recovering land water reserve abnormity from satellite gravity data by machine learning
CN115630686B (en) * 2022-10-11 2023-06-23 首都师范大学 Method for recovering land water reserve anomalies from satellite gravity data using machine learning
CN117113211A (en) * 2023-01-16 2023-11-24 杭州市水务集团有限公司 Intelligent water use monitoring system and method for solitary old people
CN117113211B (en) * 2023-01-16 2024-03-12 杭州市水务集团有限公司 Intelligent water use monitoring system and method for solitary old people

Similar Documents

Publication Publication Date Title
Chen et al. A new downscaling-integration framework for high-resolution monthly precipitation estimates: Combining rain gauge observations, satellite-derived precipitation data and geographical ancillary data
CN112200050A (en) Lake water volume space-time change analysis method and device based on multi-source satellite data
Tian et al. Aboveground mangrove biomass estimation in Beibu Gulf using machine learning and UAV remote sensing
CN109142679B (en) Forest soil nutrient space prediction method based on artificial neural network kriging interpolation
CN104613944B (en) Distributed water depth prediction method based on GWR (geographically weighted regression) and BP (back propagation) neural network
CN106767687A (en) A kind of method of utilization remote sensing moisture measurement beach elevation
Yan et al. Shoreline change detection and forecast along the Yancheng coast using a digital shoreline analysis system
Wang et al. A new likelihood function for consistent phase series estimation in distributed scatterer interferometry
CN105115883B (en) The detection method and device of soil water erosion amount
Gao et al. Analysis of flood inundation in ungauged basins based on multi-source remote sensing data
Yağmur et al. Opportunities provided by remote sensing data for watershed management: example of Konya Closed Basin
Fan et al. A comparative study of four merging approaches for regional precipitation estimation
Mohebzadeh et al. Spatial downscaling of MODIS Chlorophyll-a with machine learning techniques over the west coast of the Yellow Sea in South Korea
CN114091756B (en) Township tsunami risk assessment method based on Thiessen polygon
Zhao et al. Stability evaluation of tidal flats based on time-series satellite images: A case study of the Jiangsu central coast, China
Bacalhau et al. Bathymetry of reservoirs using altimetric data associated to optical images
CN112241844A (en) Method and device for determining and updating environmental risk source background list of drinking water source area
Wang et al. Study on Remote Sensing of Water Depths Based on BP Artificial Neural Network.
Bakış et al. Analysis and comparison of spatial rainfall distribution applying different interpolation methods in Porsuk river basin, Turkey
Kazemzadeh et al. Remote sensing of temporal and spatial variations of suspended sediment concentration in Bahmanshir Estuary, Iran
Tournadre et al. High-resolution imaging of the ocean surface backscatter by inversion of altimeter waveforms
CN115060656A (en) Satellite remote sensing water depth inversion method based on sparse prior actual measurement points
Zhou et al. Bathymetry of the Gulf of Mexico predicted with multilayer perceptron from multi-source marine geodetic data
Li et al. Retrieval of Tidal Flat Elevation Based on Remotely Sensed Moisture Approach
Jiang et al. Surface global and diffuse solar radiation over China acquired from geostationary Multi-functional Transport Satellite data

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